annotate lib/regcomp.c @ 5968:541fed6ae301

* modules/regex (Files): Add lib/regex_internal.c, lib/regex_internal.h, lib/regexec.c, lib/regcomp.c, m4/codeset.m4. (Depends-on): Add extensions. (Makefile.am): Remove lib_SOURCES; now done by m4 code. * config/srclist.txt: Add regcomp.c, regex.c, regex.h, regex_internal.c, regexec.c. Add regex_internal.h too, but as a comment, since the libc version is currently broken in gnulib mode. * lib/regex.c, lib/regex.h: Sync from libc. * lib/regcomp.c, lib/regexec_internal.c, lib/regex_internal.h, lib/regexec.c: New files, synced from libc, except that regex_internal.h currently has a small porting fix. * m4/regex.m4: Adjust to new libc regex implementation. (gl_INCLUDED_REGEX): Add AC_LIBSOURCES for all the .c and .h parts of (the new) regex. Quote the m4 stuff better. Check for RE_ICASE bug of old gnulib. Check for REG_STARTEND of recent libc. Rename local variables from jm_* to gl_*. Quote operand of "test -f". Say "recent enough" version of libc, not "version 2". (gl_PREREQ_REGEX): Remove AC_FUNC_ALLOCA, since alloca is a prerequisite module. Remove AC_HEADER_STDC; no longer needed. Check for locale.h, isblank, mbrtowc, wcrtomb, wcscoll. Remove check for btowc, isascii. Require AM_LANGINFO_CODESET.
author Paul Eggert <eggert@cs.ucla.edu>
date Thu, 07 Jul 2005 08:08:39 +0000
parents
children aa260da0bbbe
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1 /* Extended regular expression matching and search library.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2 Copyright (C) 2002, 2003, 2004, 2005 Free Software Foundation, Inc.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3 This file is part of the GNU C Library.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
4 Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
5
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
6 This program is free software; you can redistribute it and/or modify
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
7 it under the terms of the GNU General Public License as published by
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
8 the Free Software Foundation; either version 2, or (at your option)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
9 any later version.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
10
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
11 This program is distributed in the hope that it will be useful,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
14 GNU General Public License for more details.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
15
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
16 You should have received a copy of the GNU General Public License along
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
17 with this program; if not, write to the Free Software Foundation,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
18 Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
19
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
20 static reg_errcode_t re_compile_internal (regex_t *preg, const char * pattern,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
21 int length, reg_syntax_t syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
22 static void re_compile_fastmap_iter (regex_t *bufp,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
23 const re_dfastate_t *init_state,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
24 char *fastmap);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
25 static reg_errcode_t init_dfa (re_dfa_t *dfa, int pat_len);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
26 static void init_word_char (re_dfa_t *dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
27 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
28 static void free_charset (re_charset_t *cset);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
29 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
30 static void free_workarea_compile (regex_t *preg);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
31 static reg_errcode_t create_initial_state (re_dfa_t *dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
32 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
33 static void optimize_utf8 (re_dfa_t *dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
34 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
35 static reg_errcode_t analyze (regex_t *preg);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
36 static reg_errcode_t create_initial_state (re_dfa_t *dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
37 static reg_errcode_t preorder (bin_tree_t *root,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
38 reg_errcode_t (fn (void *, bin_tree_t *)),
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
39 void *extra);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
40 static reg_errcode_t postorder (bin_tree_t *root,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
41 reg_errcode_t (fn (void *, bin_tree_t *)),
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
42 void *extra);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
43 static reg_errcode_t optimize_subexps (void *extra, bin_tree_t *node);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
44 static reg_errcode_t lower_subexps (void *extra, bin_tree_t *node);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
45 static bin_tree_t *lower_subexp (reg_errcode_t *err, regex_t *preg,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
46 bin_tree_t *node);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
47 static reg_errcode_t calc_first (void *extra, bin_tree_t *node);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
48 static reg_errcode_t calc_next (void *extra, bin_tree_t *node);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
49 static reg_errcode_t link_nfa_nodes (void *extra, bin_tree_t *node);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
50 static reg_errcode_t duplicate_node_closure (re_dfa_t *dfa, int top_org_node,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
51 int top_clone_node, int root_node,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
52 unsigned int constraint);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
53 static reg_errcode_t duplicate_node (int *new_idx, re_dfa_t *dfa, int org_idx,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
54 unsigned int constraint);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
55 static int search_duplicated_node (re_dfa_t *dfa, int org_node,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
56 unsigned int constraint);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
57 static reg_errcode_t calc_eclosure (re_dfa_t *dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
58 static reg_errcode_t calc_eclosure_iter (re_node_set *new_set, re_dfa_t *dfa,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
59 int node, int root);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
60 static reg_errcode_t calc_inveclosure (re_dfa_t *dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
61 static int fetch_number (re_string_t *input, re_token_t *token,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
62 reg_syntax_t syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
63 static void fetch_token (re_token_t *result, re_string_t *input,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
64 reg_syntax_t syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
65 static int peek_token (re_token_t *token, re_string_t *input,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
66 reg_syntax_t syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
67 static int peek_token_bracket (re_token_t *token, re_string_t *input,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
68 reg_syntax_t syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
69 static bin_tree_t *parse (re_string_t *regexp, regex_t *preg,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
70 reg_syntax_t syntax, reg_errcode_t *err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
71 static bin_tree_t *parse_reg_exp (re_string_t *regexp, regex_t *preg,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
72 re_token_t *token, reg_syntax_t syntax,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
73 int nest, reg_errcode_t *err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
74 static bin_tree_t *parse_branch (re_string_t *regexp, regex_t *preg,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
75 re_token_t *token, reg_syntax_t syntax,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
76 int nest, reg_errcode_t *err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
77 static bin_tree_t *parse_expression (re_string_t *regexp, regex_t *preg,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
78 re_token_t *token, reg_syntax_t syntax,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
79 int nest, reg_errcode_t *err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
80 static bin_tree_t *parse_sub_exp (re_string_t *regexp, regex_t *preg,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
81 re_token_t *token, reg_syntax_t syntax,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
82 int nest, reg_errcode_t *err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
83 static bin_tree_t *parse_dup_op (bin_tree_t *dup_elem, re_string_t *regexp,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
84 re_dfa_t *dfa, re_token_t *token,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
85 reg_syntax_t syntax, reg_errcode_t *err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
86 static bin_tree_t *parse_bracket_exp (re_string_t *regexp, re_dfa_t *dfa,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
87 re_token_t *token, reg_syntax_t syntax,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
88 reg_errcode_t *err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
89 static reg_errcode_t parse_bracket_element (bracket_elem_t *elem,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
90 re_string_t *regexp,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
91 re_token_t *token, int token_len,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
92 re_dfa_t *dfa,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
93 reg_syntax_t syntax,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
94 int accept_hyphen);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
95 static reg_errcode_t parse_bracket_symbol (bracket_elem_t *elem,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
96 re_string_t *regexp,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
97 re_token_t *token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
98 #ifndef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
99 # ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
100 static reg_errcode_t build_range_exp (re_bitset_ptr_t sbcset,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
101 re_charset_t *mbcset, int *range_alloc,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
102 bracket_elem_t *start_elem,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
103 bracket_elem_t *end_elem);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
104 static reg_errcode_t build_collating_symbol (re_bitset_ptr_t sbcset,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
105 re_charset_t *mbcset,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
106 int *coll_sym_alloc,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
107 const unsigned char *name);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
108 # else /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
109 static reg_errcode_t build_range_exp (re_bitset_ptr_t sbcset,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
110 bracket_elem_t *start_elem,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
111 bracket_elem_t *end_elem);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
112 static reg_errcode_t build_collating_symbol (re_bitset_ptr_t sbcset,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
113 const unsigned char *name);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
114 # endif /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
115 #endif /* not _LIBC */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
116 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
117 static reg_errcode_t build_equiv_class (re_bitset_ptr_t sbcset,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
118 re_charset_t *mbcset,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
119 int *equiv_class_alloc,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
120 const unsigned char *name);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
121 static reg_errcode_t build_charclass (unsigned RE_TRANSLATE_TYPE trans,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
122 re_bitset_ptr_t sbcset,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
123 re_charset_t *mbcset,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
124 int *char_class_alloc,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
125 const unsigned char *class_name,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
126 reg_syntax_t syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
127 #else /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
128 static reg_errcode_t build_equiv_class (re_bitset_ptr_t sbcset,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
129 const unsigned char *name);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
130 static reg_errcode_t build_charclass (unsigned RE_TRANSLATE_TYPE trans,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
131 re_bitset_ptr_t sbcset,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
132 const unsigned char *class_name,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
133 reg_syntax_t syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
134 #endif /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
135 static bin_tree_t *build_charclass_op (re_dfa_t *dfa,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
136 unsigned RE_TRANSLATE_TYPE trans,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
137 const unsigned char *class_name,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
138 const unsigned char *extra,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
139 int non_match, reg_errcode_t *err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
140 static bin_tree_t *create_tree (re_dfa_t *dfa,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
141 bin_tree_t *left, bin_tree_t *right,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
142 re_token_type_t type);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
143 static bin_tree_t *create_token_tree (re_dfa_t *dfa,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
144 bin_tree_t *left, bin_tree_t *right,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
145 const re_token_t *token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
146 static bin_tree_t *duplicate_tree (const bin_tree_t *src, re_dfa_t *dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
147 static void free_token (re_token_t *node);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
148 static reg_errcode_t free_tree (void *extra, bin_tree_t *node);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
149 static reg_errcode_t mark_opt_subexp (void *extra, bin_tree_t *node);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
150
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
151 /* This table gives an error message for each of the error codes listed
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
152 in regex.h. Obviously the order here has to be same as there.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
153 POSIX doesn't require that we do anything for REG_NOERROR,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
154 but why not be nice? */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
155
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
156 const char __re_error_msgid[] attribute_hidden =
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
157 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
158 #define REG_NOERROR_IDX 0
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
159 gettext_noop ("Success") /* REG_NOERROR */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
160 "\0"
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
161 #define REG_NOMATCH_IDX (REG_NOERROR_IDX + sizeof "Success")
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
162 gettext_noop ("No match") /* REG_NOMATCH */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
163 "\0"
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
164 #define REG_BADPAT_IDX (REG_NOMATCH_IDX + sizeof "No match")
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
165 gettext_noop ("Invalid regular expression") /* REG_BADPAT */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
166 "\0"
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
167 #define REG_ECOLLATE_IDX (REG_BADPAT_IDX + sizeof "Invalid regular expression")
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
168 gettext_noop ("Invalid collation character") /* REG_ECOLLATE */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
169 "\0"
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
170 #define REG_ECTYPE_IDX (REG_ECOLLATE_IDX + sizeof "Invalid collation character")
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
171 gettext_noop ("Invalid character class name") /* REG_ECTYPE */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
172 "\0"
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
173 #define REG_EESCAPE_IDX (REG_ECTYPE_IDX + sizeof "Invalid character class name")
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
174 gettext_noop ("Trailing backslash") /* REG_EESCAPE */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
175 "\0"
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
176 #define REG_ESUBREG_IDX (REG_EESCAPE_IDX + sizeof "Trailing backslash")
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
177 gettext_noop ("Invalid back reference") /* REG_ESUBREG */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
178 "\0"
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
179 #define REG_EBRACK_IDX (REG_ESUBREG_IDX + sizeof "Invalid back reference")
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
180 gettext_noop ("Unmatched [ or [^") /* REG_EBRACK */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
181 "\0"
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
182 #define REG_EPAREN_IDX (REG_EBRACK_IDX + sizeof "Unmatched [ or [^")
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
183 gettext_noop ("Unmatched ( or \\(") /* REG_EPAREN */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
184 "\0"
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
185 #define REG_EBRACE_IDX (REG_EPAREN_IDX + sizeof "Unmatched ( or \\(")
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
186 gettext_noop ("Unmatched \\{") /* REG_EBRACE */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
187 "\0"
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
188 #define REG_BADBR_IDX (REG_EBRACE_IDX + sizeof "Unmatched \\{")
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
189 gettext_noop ("Invalid content of \\{\\}") /* REG_BADBR */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
190 "\0"
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
191 #define REG_ERANGE_IDX (REG_BADBR_IDX + sizeof "Invalid content of \\{\\}")
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
192 gettext_noop ("Invalid range end") /* REG_ERANGE */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
193 "\0"
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
194 #define REG_ESPACE_IDX (REG_ERANGE_IDX + sizeof "Invalid range end")
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
195 gettext_noop ("Memory exhausted") /* REG_ESPACE */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
196 "\0"
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
197 #define REG_BADRPT_IDX (REG_ESPACE_IDX + sizeof "Memory exhausted")
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
198 gettext_noop ("Invalid preceding regular expression") /* REG_BADRPT */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
199 "\0"
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
200 #define REG_EEND_IDX (REG_BADRPT_IDX + sizeof "Invalid preceding regular expression")
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
201 gettext_noop ("Premature end of regular expression") /* REG_EEND */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
202 "\0"
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
203 #define REG_ESIZE_IDX (REG_EEND_IDX + sizeof "Premature end of regular expression")
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
204 gettext_noop ("Regular expression too big") /* REG_ESIZE */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
205 "\0"
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
206 #define REG_ERPAREN_IDX (REG_ESIZE_IDX + sizeof "Regular expression too big")
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
207 gettext_noop ("Unmatched ) or \\)") /* REG_ERPAREN */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
208 };
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
209
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
210 const size_t __re_error_msgid_idx[] attribute_hidden =
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
211 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
212 REG_NOERROR_IDX,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
213 REG_NOMATCH_IDX,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
214 REG_BADPAT_IDX,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
215 REG_ECOLLATE_IDX,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
216 REG_ECTYPE_IDX,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
217 REG_EESCAPE_IDX,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
218 REG_ESUBREG_IDX,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
219 REG_EBRACK_IDX,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
220 REG_EPAREN_IDX,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
221 REG_EBRACE_IDX,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
222 REG_BADBR_IDX,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
223 REG_ERANGE_IDX,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
224 REG_ESPACE_IDX,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
225 REG_BADRPT_IDX,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
226 REG_EEND_IDX,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
227 REG_ESIZE_IDX,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
228 REG_ERPAREN_IDX
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
229 };
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
230
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
231 /* Entry points for GNU code. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
232
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
233 /* re_compile_pattern is the GNU regular expression compiler: it
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
234 compiles PATTERN (of length LENGTH) and puts the result in BUFP.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
235 Returns 0 if the pattern was valid, otherwise an error string.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
236
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
237 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
238 are set in BUFP on entry. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
239
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
240 const char *
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
241 re_compile_pattern (pattern, length, bufp)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
242 const char *pattern;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
243 size_t length;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
244 struct re_pattern_buffer *bufp;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
245 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
246 reg_errcode_t ret;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
247
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
248 /* And GNU code determines whether or not to get register information
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
249 by passing null for the REGS argument to re_match, etc., not by
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
250 setting no_sub, unless RE_NO_SUB is set. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
251 bufp->no_sub = !!(re_syntax_options & RE_NO_SUB);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
252
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
253 /* Match anchors at newline. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
254 bufp->newline_anchor = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
255
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
256 ret = re_compile_internal (bufp, pattern, length, re_syntax_options);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
257
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
258 if (!ret)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
259 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
260 return gettext (__re_error_msgid + __re_error_msgid_idx[(int) ret]);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
261 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
262 #ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
263 weak_alias (__re_compile_pattern, re_compile_pattern)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
264 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
265
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
266 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
267 also be assigned to arbitrarily: each pattern buffer stores its own
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
268 syntax, so it can be changed between regex compilations. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
269 /* This has no initializer because initialized variables in Emacs
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
270 become read-only after dumping. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
271 reg_syntax_t re_syntax_options;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
272
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
273
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
274 /* Specify the precise syntax of regexps for compilation. This provides
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
275 for compatibility for various utilities which historically have
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
276 different, incompatible syntaxes.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
277
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
278 The argument SYNTAX is a bit mask comprised of the various bits
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
279 defined in regex.h. We return the old syntax. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
280
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
281 reg_syntax_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
282 re_set_syntax (syntax)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
283 reg_syntax_t syntax;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
284 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
285 reg_syntax_t ret = re_syntax_options;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
286
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
287 re_syntax_options = syntax;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
288 return ret;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
289 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
290 #ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
291 weak_alias (__re_set_syntax, re_set_syntax)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
292 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
293
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
294 int
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
295 re_compile_fastmap (bufp)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
296 struct re_pattern_buffer *bufp;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
297 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
298 re_dfa_t *dfa = (re_dfa_t *) bufp->buffer;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
299 char *fastmap = bufp->fastmap;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
300
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
301 memset (fastmap, '\0', sizeof (char) * SBC_MAX);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
302 re_compile_fastmap_iter (bufp, dfa->init_state, fastmap);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
303 if (dfa->init_state != dfa->init_state_word)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
304 re_compile_fastmap_iter (bufp, dfa->init_state_word, fastmap);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
305 if (dfa->init_state != dfa->init_state_nl)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
306 re_compile_fastmap_iter (bufp, dfa->init_state_nl, fastmap);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
307 if (dfa->init_state != dfa->init_state_begbuf)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
308 re_compile_fastmap_iter (bufp, dfa->init_state_begbuf, fastmap);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
309 bufp->fastmap_accurate = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
310 return 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
311 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
312 #ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
313 weak_alias (__re_compile_fastmap, re_compile_fastmap)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
314 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
315
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
316 static inline void
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
317 __attribute ((always_inline))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
318 re_set_fastmap (char *fastmap, int icase, int ch)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
319 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
320 fastmap[ch] = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
321 if (icase)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
322 fastmap[tolower (ch)] = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
323 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
324
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
325 /* Helper function for re_compile_fastmap.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
326 Compile fastmap for the initial_state INIT_STATE. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
327
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
328 static void
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
329 re_compile_fastmap_iter (bufp, init_state, fastmap)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
330 regex_t *bufp;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
331 const re_dfastate_t *init_state;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
332 char *fastmap;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
333 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
334 re_dfa_t *dfa = (re_dfa_t *) bufp->buffer;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
335 int node_cnt;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
336 int icase = (dfa->mb_cur_max == 1 && (bufp->syntax & RE_ICASE));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
337 for (node_cnt = 0; node_cnt < init_state->nodes.nelem; ++node_cnt)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
338 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
339 int node = init_state->nodes.elems[node_cnt];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
340 re_token_type_t type = dfa->nodes[node].type;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
341
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
342 if (type == CHARACTER)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
343 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
344 re_set_fastmap (fastmap, icase, dfa->nodes[node].opr.c);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
345 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
346 if ((bufp->syntax & RE_ICASE) && dfa->mb_cur_max > 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
347 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
348 unsigned char *buf = alloca (dfa->mb_cur_max), *p;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
349 wchar_t wc;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
350 mbstate_t state;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
351
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
352 p = buf;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
353 *p++ = dfa->nodes[node].opr.c;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
354 while (++node < dfa->nodes_len
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
355 && dfa->nodes[node].type == CHARACTER
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
356 && dfa->nodes[node].mb_partial)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
357 *p++ = dfa->nodes[node].opr.c;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
358 memset (&state, 0, sizeof (state));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
359 if (mbrtowc (&wc, (const char *) buf, p - buf,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
360 &state) == p - buf
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
361 && (__wcrtomb ((char *) buf, towlower (wc), &state)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
362 != (size_t) -1))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
363 re_set_fastmap (fastmap, 0, buf[0]);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
364 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
365 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
366 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
367 else if (type == SIMPLE_BRACKET)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
368 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
369 int i, j, ch;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
370 for (i = 0, ch = 0; i < BITSET_UINTS; ++i)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
371 for (j = 0; j < UINT_BITS; ++j, ++ch)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
372 if (dfa->nodes[node].opr.sbcset[i] & (1 << j))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
373 re_set_fastmap (fastmap, icase, ch);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
374 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
375 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
376 else if (type == COMPLEX_BRACKET)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
377 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
378 int i;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
379 re_charset_t *cset = dfa->nodes[node].opr.mbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
380 if (cset->non_match || cset->ncoll_syms || cset->nequiv_classes
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
381 || cset->nranges || cset->nchar_classes)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
382 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
383 # ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
384 if (_NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES) != 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
385 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
386 /* In this case we want to catch the bytes which are
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
387 the first byte of any collation elements.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
388 e.g. In da_DK, we want to catch 'a' since "aa"
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
389 is a valid collation element, and don't catch
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
390 'b' since 'b' is the only collation element
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
391 which starts from 'b'. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
392 int j, ch;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
393 const int32_t *table = (const int32_t *)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
394 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
395 for (i = 0, ch = 0; i < BITSET_UINTS; ++i)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
396 for (j = 0; j < UINT_BITS; ++j, ++ch)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
397 if (table[ch] < 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
398 re_set_fastmap (fastmap, icase, ch);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
399 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
400 # else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
401 if (dfa->mb_cur_max > 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
402 for (i = 0; i < SBC_MAX; ++i)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
403 if (__btowc (i) == WEOF)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
404 re_set_fastmap (fastmap, icase, i);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
405 # endif /* not _LIBC */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
406 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
407 for (i = 0; i < cset->nmbchars; ++i)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
408 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
409 char buf[256];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
410 mbstate_t state;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
411 memset (&state, '\0', sizeof (state));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
412 if (__wcrtomb (buf, cset->mbchars[i], &state) != (size_t) -1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
413 re_set_fastmap (fastmap, icase, *(unsigned char *) buf);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
414 if ((bufp->syntax & RE_ICASE) && dfa->mb_cur_max > 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
415 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
416 if (__wcrtomb (buf, towlower (cset->mbchars[i]), &state)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
417 != (size_t) -1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
418 re_set_fastmap (fastmap, 0, *(unsigned char *) buf);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
419 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
420 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
421 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
422 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
423 else if (type == OP_PERIOD
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
424 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
425 || type == OP_UTF8_PERIOD
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
426 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
427 || type == END_OF_RE)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
428 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
429 memset (fastmap, '\1', sizeof (char) * SBC_MAX);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
430 if (type == END_OF_RE)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
431 bufp->can_be_null = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
432 return;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
433 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
434 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
435 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
436
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
437 /* Entry point for POSIX code. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
438 /* regcomp takes a regular expression as a string and compiles it.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
439
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
440 PREG is a regex_t *. We do not expect any fields to be initialized,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
441 since POSIX says we shouldn't. Thus, we set
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
442
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
443 `buffer' to the compiled pattern;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
444 `used' to the length of the compiled pattern;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
445 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
446 REG_EXTENDED bit in CFLAGS is set; otherwise, to
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
447 RE_SYNTAX_POSIX_BASIC;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
448 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
449 `fastmap' to an allocated space for the fastmap;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
450 `fastmap_accurate' to zero;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
451 `re_nsub' to the number of subexpressions in PATTERN.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
452
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
453 PATTERN is the address of the pattern string.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
454
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
455 CFLAGS is a series of bits which affect compilation.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
456
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
457 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
458 use POSIX basic syntax.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
459
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
460 If REG_NEWLINE is set, then . and [^...] don't match newline.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
461 Also, regexec will try a match beginning after every newline.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
462
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
463 If REG_ICASE is set, then we considers upper- and lowercase
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
464 versions of letters to be equivalent when matching.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
465
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
466 If REG_NOSUB is set, then when PREG is passed to regexec, that
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
467 routine will report only success or failure, and nothing about the
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
468 registers.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
469
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
470 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
471 the return codes and their meanings.) */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
472
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
473 int
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
474 regcomp (preg, pattern, cflags)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
475 regex_t *__restrict preg;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
476 const char *__restrict pattern;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
477 int cflags;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
478 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
479 reg_errcode_t ret;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
480 reg_syntax_t syntax = ((cflags & REG_EXTENDED) ? RE_SYNTAX_POSIX_EXTENDED
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
481 : RE_SYNTAX_POSIX_BASIC);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
482
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
483 preg->buffer = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
484 preg->allocated = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
485 preg->used = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
486
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
487 /* Try to allocate space for the fastmap. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
488 preg->fastmap = re_malloc (char, SBC_MAX);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
489 if (BE (preg->fastmap == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
490 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
491
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
492 syntax |= (cflags & REG_ICASE) ? RE_ICASE : 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
493
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
494 /* If REG_NEWLINE is set, newlines are treated differently. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
495 if (cflags & REG_NEWLINE)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
496 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
497 syntax &= ~RE_DOT_NEWLINE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
498 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
499 /* It also changes the matching behavior. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
500 preg->newline_anchor = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
501 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
502 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
503 preg->newline_anchor = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
504 preg->no_sub = !!(cflags & REG_NOSUB);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
505 preg->translate = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
506
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
507 ret = re_compile_internal (preg, pattern, strlen (pattern), syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
508
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
509 /* POSIX doesn't distinguish between an unmatched open-group and an
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
510 unmatched close-group: both are REG_EPAREN. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
511 if (ret == REG_ERPAREN)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
512 ret = REG_EPAREN;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
513
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
514 /* We have already checked preg->fastmap != NULL. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
515 if (BE (ret == REG_NOERROR, 1))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
516 /* Compute the fastmap now, since regexec cannot modify the pattern
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
517 buffer. This function never fails in this implementation. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
518 (void) re_compile_fastmap (preg);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
519 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
520 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
521 /* Some error occurred while compiling the expression. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
522 re_free (preg->fastmap);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
523 preg->fastmap = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
524 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
525
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
526 return (int) ret;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
527 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
528 #ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
529 weak_alias (__regcomp, regcomp)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
530 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
531
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
532 /* Returns a message corresponding to an error code, ERRCODE, returned
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
533 from either regcomp or regexec. We don't use PREG here. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
534
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
535 size_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
536 regerror (errcode, preg, errbuf, errbuf_size)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
537 int errcode;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
538 const regex_t *preg;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
539 char *errbuf;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
540 size_t errbuf_size;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
541 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
542 const char *msg;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
543 size_t msg_size;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
544
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
545 if (BE (errcode < 0
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
546 || errcode >= (int) (sizeof (__re_error_msgid_idx)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
547 / sizeof (__re_error_msgid_idx[0])), 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
548 /* Only error codes returned by the rest of the code should be passed
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
549 to this routine. If we are given anything else, or if other regex
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
550 code generates an invalid error code, then the program has a bug.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
551 Dump core so we can fix it. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
552 abort ();
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
553
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
554 msg = gettext (__re_error_msgid + __re_error_msgid_idx[errcode]);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
555
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
556 msg_size = strlen (msg) + 1; /* Includes the null. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
557
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
558 if (BE (errbuf_size != 0, 1))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
559 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
560 if (BE (msg_size > errbuf_size, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
561 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
562 #if defined HAVE_MEMPCPY || defined _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
563 *((char *) __mempcpy (errbuf, msg, errbuf_size - 1)) = '\0';
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
564 #else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
565 memcpy (errbuf, msg, errbuf_size - 1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
566 errbuf[errbuf_size - 1] = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
567 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
568 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
569 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
570 memcpy (errbuf, msg, msg_size);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
571 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
572
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
573 return msg_size;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
574 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
575 #ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
576 weak_alias (__regerror, regerror)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
577 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
578
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
579
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
580 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
581 /* This static array is used for the map to single-byte characters when
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
582 UTF-8 is used. Otherwise we would allocate memory just to initialize
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
583 it the same all the time. UTF-8 is the preferred encoding so this is
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
584 a worthwhile optimization. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
585 static const bitset utf8_sb_map =
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
586 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
587 /* Set the first 128 bits. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
588 # if UINT_MAX == 0xffffffff
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
589 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
590 # else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
591 # error "Add case for new unsigned int size"
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
592 # endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
593 };
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
594 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
595
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
596
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
597 static void
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
598 free_dfa_content (re_dfa_t *dfa)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
599 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
600 int i, j;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
601
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
602 if (dfa->nodes)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
603 for (i = 0; i < dfa->nodes_len; ++i)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
604 free_token (dfa->nodes + i);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
605 re_free (dfa->nexts);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
606 for (i = 0; i < dfa->nodes_len; ++i)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
607 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
608 if (dfa->eclosures != NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
609 re_node_set_free (dfa->eclosures + i);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
610 if (dfa->inveclosures != NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
611 re_node_set_free (dfa->inveclosures + i);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
612 if (dfa->edests != NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
613 re_node_set_free (dfa->edests + i);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
614 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
615 re_free (dfa->edests);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
616 re_free (dfa->eclosures);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
617 re_free (dfa->inveclosures);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
618 re_free (dfa->nodes);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
619
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
620 if (dfa->state_table)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
621 for (i = 0; i <= dfa->state_hash_mask; ++i)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
622 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
623 struct re_state_table_entry *entry = dfa->state_table + i;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
624 for (j = 0; j < entry->num; ++j)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
625 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
626 re_dfastate_t *state = entry->array[j];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
627 free_state (state);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
628 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
629 re_free (entry->array);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
630 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
631 re_free (dfa->state_table);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
632 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
633 if (dfa->sb_char != utf8_sb_map)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
634 re_free (dfa->sb_char);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
635 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
636 re_free (dfa->subexp_map);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
637 #ifdef DEBUG
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
638 re_free (dfa->re_str);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
639 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
640
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
641 re_free (dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
642 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
643
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
644
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
645 /* Free dynamically allocated space used by PREG. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
646
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
647 void
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
648 regfree (preg)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
649 regex_t *preg;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
650 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
651 re_dfa_t *dfa = (re_dfa_t *) preg->buffer;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
652 if (BE (dfa != NULL, 1))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
653 free_dfa_content (dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
654 preg->buffer = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
655 preg->allocated = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
656
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
657 re_free (preg->fastmap);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
658 preg->fastmap = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
659
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
660 re_free (preg->translate);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
661 preg->translate = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
662 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
663 #ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
664 weak_alias (__regfree, regfree)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
665 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
666
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
667 /* Entry points compatible with 4.2 BSD regex library. We don't define
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
668 them unless specifically requested. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
669
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
670 #if defined _REGEX_RE_COMP || defined _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
671
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
672 /* BSD has one and only one pattern buffer. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
673 static struct re_pattern_buffer re_comp_buf;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
674
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
675 char *
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
676 # ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
677 /* Make these definitions weak in libc, so POSIX programs can redefine
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
678 these names if they don't use our functions, and still use
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
679 regcomp/regexec above without link errors. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
680 weak_function
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
681 # endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
682 re_comp (s)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
683 const char *s;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
684 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
685 reg_errcode_t ret;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
686 char *fastmap;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
687
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
688 if (!s)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
689 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
690 if (!re_comp_buf.buffer)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
691 return gettext ("No previous regular expression");
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
692 return 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
693 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
694
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
695 if (re_comp_buf.buffer)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
696 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
697 fastmap = re_comp_buf.fastmap;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
698 re_comp_buf.fastmap = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
699 __regfree (&re_comp_buf);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
700 memset (&re_comp_buf, '\0', sizeof (re_comp_buf));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
701 re_comp_buf.fastmap = fastmap;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
702 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
703
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
704 if (re_comp_buf.fastmap == NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
705 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
706 re_comp_buf.fastmap = (char *) malloc (SBC_MAX);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
707 if (re_comp_buf.fastmap == NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
708 return (char *) gettext (__re_error_msgid
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
709 + __re_error_msgid_idx[(int) REG_ESPACE]);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
710 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
711
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
712 /* Since `re_exec' always passes NULL for the `regs' argument, we
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
713 don't need to initialize the pattern buffer fields which affect it. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
714
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
715 /* Match anchors at newlines. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
716 re_comp_buf.newline_anchor = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
717
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
718 ret = re_compile_internal (&re_comp_buf, s, strlen (s), re_syntax_options);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
719
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
720 if (!ret)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
721 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
722
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
723 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
724 return (char *) gettext (__re_error_msgid + __re_error_msgid_idx[(int) ret]);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
725 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
726
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
727 #ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
728 libc_freeres_fn (free_mem)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
729 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
730 __regfree (&re_comp_buf);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
731 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
732 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
733
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
734 #endif /* _REGEX_RE_COMP */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
735
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
736 /* Internal entry point.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
737 Compile the regular expression PATTERN, whose length is LENGTH.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
738 SYNTAX indicate regular expression's syntax. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
739
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
740 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
741 re_compile_internal (preg, pattern, length, syntax)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
742 regex_t *preg;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
743 const char * pattern;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
744 int length;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
745 reg_syntax_t syntax;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
746 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
747 reg_errcode_t err = REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
748 re_dfa_t *dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
749 re_string_t regexp;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
750
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
751 /* Initialize the pattern buffer. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
752 preg->fastmap_accurate = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
753 preg->syntax = syntax;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
754 preg->not_bol = preg->not_eol = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
755 preg->used = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
756 preg->re_nsub = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
757 preg->can_be_null = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
758 preg->regs_allocated = REGS_UNALLOCATED;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
759
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
760 /* Initialize the dfa. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
761 dfa = (re_dfa_t *) preg->buffer;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
762 if (BE (preg->allocated < sizeof (re_dfa_t), 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
763 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
764 /* If zero allocated, but buffer is non-null, try to realloc
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
765 enough space. This loses if buffer's address is bogus, but
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
766 that is the user's responsibility. If ->buffer is NULL this
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
767 is a simple allocation. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
768 dfa = re_realloc (preg->buffer, re_dfa_t, 1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
769 if (dfa == NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
770 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
771 preg->allocated = sizeof (re_dfa_t);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
772 preg->buffer = (unsigned char *) dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
773 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
774 preg->used = sizeof (re_dfa_t);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
775
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
776 __libc_lock_init (dfa->lock);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
777
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
778 err = init_dfa (dfa, length);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
779 if (BE (err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
780 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
781 free_dfa_content (dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
782 preg->buffer = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
783 preg->allocated = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
784 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
785 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
786 #ifdef DEBUG
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
787 dfa->re_str = re_malloc (char, length + 1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
788 strncpy (dfa->re_str, pattern, length + 1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
789 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
790
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
791 err = re_string_construct (&regexp, pattern, length, preg->translate,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
792 syntax & RE_ICASE, dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
793 if (BE (err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
794 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
795 re_compile_internal_free_return:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
796 free_workarea_compile (preg);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
797 re_string_destruct (&regexp);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
798 free_dfa_content (dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
799 preg->buffer = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
800 preg->allocated = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
801 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
802 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
803
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
804 /* Parse the regular expression, and build a structure tree. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
805 preg->re_nsub = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
806 dfa->str_tree = parse (&regexp, preg, syntax, &err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
807 if (BE (dfa->str_tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
808 goto re_compile_internal_free_return;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
809
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
810 /* Analyze the tree and create the nfa. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
811 err = analyze (preg);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
812 if (BE (err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
813 goto re_compile_internal_free_return;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
814
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
815 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
816 /* If possible, do searching in single byte encoding to speed things up. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
817 if (dfa->is_utf8 && !(syntax & RE_ICASE) && preg->translate == NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
818 optimize_utf8 (dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
819 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
820
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
821 /* Then create the initial state of the dfa. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
822 err = create_initial_state (dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
823
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
824 /* Release work areas. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
825 free_workarea_compile (preg);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
826 re_string_destruct (&regexp);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
827
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
828 if (BE (err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
829 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
830 free_dfa_content (dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
831 preg->buffer = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
832 preg->allocated = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
833 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
834
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
835 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
836 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
837
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
838 /* Initialize DFA. We use the length of the regular expression PAT_LEN
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
839 as the initial length of some arrays. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
840
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
841 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
842 init_dfa (dfa, pat_len)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
843 re_dfa_t *dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
844 int pat_len;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
845 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
846 int table_size;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
847 #ifndef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
848 char *codeset_name;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
849 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
850
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
851 memset (dfa, '\0', sizeof (re_dfa_t));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
852
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
853 /* Force allocation of str_tree_storage the first time. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
854 dfa->str_tree_storage_idx = BIN_TREE_STORAGE_SIZE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
855
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
856 dfa->nodes_alloc = pat_len + 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
857 dfa->nodes = re_malloc (re_token_t, dfa->nodes_alloc);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
858
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
859 dfa->states_alloc = pat_len + 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
860
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
861 /* table_size = 2 ^ ceil(log pat_len) */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
862 for (table_size = 1; table_size > 0; table_size <<= 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
863 if (table_size > pat_len)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
864 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
865
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
866 dfa->state_table = calloc (sizeof (struct re_state_table_entry), table_size);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
867 dfa->state_hash_mask = table_size - 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
868
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
869 dfa->mb_cur_max = MB_CUR_MAX;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
870 #ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
871 if (dfa->mb_cur_max == 6
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
872 && strcmp (_NL_CURRENT (LC_CTYPE, _NL_CTYPE_CODESET_NAME), "UTF-8") == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
873 dfa->is_utf8 = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
874 dfa->map_notascii = (_NL_CURRENT_WORD (LC_CTYPE, _NL_CTYPE_MAP_TO_NONASCII)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
875 != 0);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
876 #else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
877 # ifdef HAVE_LANGINFO_CODESET
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
878 codeset_name = nl_langinfo (CODESET);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
879 # else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
880 codeset_name = getenv ("LC_ALL");
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
881 if (codeset_name == NULL || codeset_name[0] == '\0')
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
882 codeset_name = getenv ("LC_CTYPE");
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
883 if (codeset_name == NULL || codeset_name[0] == '\0')
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
884 codeset_name = getenv ("LANG");
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
885 if (codeset_name == NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
886 codeset_name = "";
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
887 else if (strchr (codeset_name, '.') != NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
888 codeset_name = strchr (codeset_name, '.') + 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
889 # endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
890
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
891 if (strcasecmp (codeset_name, "UTF-8") == 0
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
892 || strcasecmp (codeset_name, "UTF8") == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
893 dfa->is_utf8 = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
894
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
895 /* We check exhaustively in the loop below if this charset is a
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
896 superset of ASCII. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
897 dfa->map_notascii = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
898 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
899
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
900 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
901 if (dfa->mb_cur_max > 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
902 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
903 if (dfa->is_utf8)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
904 dfa->sb_char = (re_bitset_ptr_t) utf8_sb_map;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
905 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
906 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
907 int i, j, ch;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
908
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
909 dfa->sb_char = (re_bitset_ptr_t) calloc (sizeof (bitset), 1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
910 if (BE (dfa->sb_char == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
911 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
912
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
913 /* Clear all bits by, then set those corresponding to single
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
914 byte chars. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
915 bitset_empty (dfa->sb_char);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
916
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
917 for (i = 0, ch = 0; i < BITSET_UINTS; ++i)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
918 for (j = 0; j < UINT_BITS; ++j, ++ch)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
919 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
920 wchar_t wch = __btowc (ch);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
921 if (wch != WEOF)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
922 dfa->sb_char[i] |= 1 << j;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
923 # ifndef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
924 if (isascii (ch) && wch != (wchar_t) ch)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
925 dfa->map_notascii = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
926 # endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
927 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
928 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
929 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
930 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
931
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
932 if (BE (dfa->nodes == NULL || dfa->state_table == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
933 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
934 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
935 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
936
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
937 /* Initialize WORD_CHAR table, which indicate which character is
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
938 "word". In this case "word" means that it is the word construction
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
939 character used by some operators like "\<", "\>", etc. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
940
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
941 static void
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
942 init_word_char (dfa)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
943 re_dfa_t *dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
944 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
945 int i, j, ch;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
946 dfa->word_ops_used = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
947 for (i = 0, ch = 0; i < BITSET_UINTS; ++i)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
948 for (j = 0; j < UINT_BITS; ++j, ++ch)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
949 if (isalnum (ch) || ch == '_')
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
950 dfa->word_char[i] |= 1 << j;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
951 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
952
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
953 /* Free the work area which are only used while compiling. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
954
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
955 static void
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
956 free_workarea_compile (preg)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
957 regex_t *preg;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
958 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
959 re_dfa_t *dfa = (re_dfa_t *) preg->buffer;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
960 bin_tree_storage_t *storage, *next;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
961 for (storage = dfa->str_tree_storage; storage; storage = next)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
962 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
963 next = storage->next;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
964 re_free (storage);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
965 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
966 dfa->str_tree_storage = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
967 dfa->str_tree_storage_idx = BIN_TREE_STORAGE_SIZE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
968 dfa->str_tree = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
969 re_free (dfa->org_indices);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
970 dfa->org_indices = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
971 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
972
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
973 /* Create initial states for all contexts. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
974
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
975 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
976 create_initial_state (dfa)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
977 re_dfa_t *dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
978 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
979 int first, i;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
980 reg_errcode_t err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
981 re_node_set init_nodes;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
982
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
983 /* Initial states have the epsilon closure of the node which is
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
984 the first node of the regular expression. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
985 first = dfa->str_tree->first->node_idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
986 dfa->init_node = first;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
987 err = re_node_set_init_copy (&init_nodes, dfa->eclosures + first);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
988 if (BE (err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
989 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
990
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
991 /* The back-references which are in initial states can epsilon transit,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
992 since in this case all of the subexpressions can be null.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
993 Then we add epsilon closures of the nodes which are the next nodes of
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
994 the back-references. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
995 if (dfa->nbackref > 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
996 for (i = 0; i < init_nodes.nelem; ++i)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
997 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
998 int node_idx = init_nodes.elems[i];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
999 re_token_type_t type = dfa->nodes[node_idx].type;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1000
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1001 int clexp_idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1002 if (type != OP_BACK_REF)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1003 continue;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1004 for (clexp_idx = 0; clexp_idx < init_nodes.nelem; ++clexp_idx)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1005 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1006 re_token_t *clexp_node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1007 clexp_node = dfa->nodes + init_nodes.elems[clexp_idx];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1008 if (clexp_node->type == OP_CLOSE_SUBEXP
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1009 && clexp_node->opr.idx == dfa->nodes[node_idx].opr.idx)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1010 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1011 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1012 if (clexp_idx == init_nodes.nelem)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1013 continue;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1014
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1015 if (type == OP_BACK_REF)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1016 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1017 int dest_idx = dfa->edests[node_idx].elems[0];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1018 if (!re_node_set_contains (&init_nodes, dest_idx))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1019 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1020 re_node_set_merge (&init_nodes, dfa->eclosures + dest_idx);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1021 i = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1022 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1023 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1024 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1025
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1026 /* It must be the first time to invoke acquire_state. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1027 dfa->init_state = re_acquire_state_context (&err, dfa, &init_nodes, 0);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1028 /* We don't check ERR here, since the initial state must not be NULL. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1029 if (BE (dfa->init_state == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1030 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1031 if (dfa->init_state->has_constraint)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1032 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1033 dfa->init_state_word = re_acquire_state_context (&err, dfa, &init_nodes,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1034 CONTEXT_WORD);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1035 dfa->init_state_nl = re_acquire_state_context (&err, dfa, &init_nodes,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1036 CONTEXT_NEWLINE);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1037 dfa->init_state_begbuf = re_acquire_state_context (&err, dfa,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1038 &init_nodes,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1039 CONTEXT_NEWLINE
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1040 | CONTEXT_BEGBUF);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1041 if (BE (dfa->init_state_word == NULL || dfa->init_state_nl == NULL
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1042 || dfa->init_state_begbuf == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1043 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1044 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1045 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1046 dfa->init_state_word = dfa->init_state_nl
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1047 = dfa->init_state_begbuf = dfa->init_state;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1048
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1049 re_node_set_free (&init_nodes);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1050 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1051 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1052
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1053 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1054 /* If it is possible to do searching in single byte encoding instead of UTF-8
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1055 to speed things up, set dfa->mb_cur_max to 1, clear is_utf8 and change
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1056 DFA nodes where needed. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1057
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1058 static void
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1059 optimize_utf8 (dfa)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1060 re_dfa_t *dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1061 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1062 int node, i, mb_chars = 0, has_period = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1063
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1064 for (node = 0; node < dfa->nodes_len; ++node)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1065 switch (dfa->nodes[node].type)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1066 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1067 case CHARACTER:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1068 if (dfa->nodes[node].opr.c >= 0x80)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1069 mb_chars = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1070 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1071 case ANCHOR:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1072 switch (dfa->nodes[node].opr.idx)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1073 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1074 case LINE_FIRST:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1075 case LINE_LAST:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1076 case BUF_FIRST:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1077 case BUF_LAST:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1078 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1079 default:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1080 /* Word anchors etc. cannot be handled. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1081 return;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1082 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1083 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1084 case OP_PERIOD:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1085 has_period = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1086 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1087 case OP_BACK_REF:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1088 case OP_ALT:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1089 case END_OF_RE:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1090 case OP_DUP_ASTERISK:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1091 case OP_OPEN_SUBEXP:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1092 case OP_CLOSE_SUBEXP:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1093 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1094 case COMPLEX_BRACKET:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1095 return;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1096 case SIMPLE_BRACKET:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1097 /* Just double check. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1098 for (i = 0x80 / UINT_BITS; i < BITSET_UINTS; ++i)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1099 if (dfa->nodes[node].opr.sbcset[i])
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1100 return;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1101 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1102 default:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1103 abort ();
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1104 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1105
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1106 if (mb_chars || has_period)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1107 for (node = 0; node < dfa->nodes_len; ++node)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1108 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1109 if (dfa->nodes[node].type == CHARACTER
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1110 && dfa->nodes[node].opr.c >= 0x80)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1111 dfa->nodes[node].mb_partial = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1112 else if (dfa->nodes[node].type == OP_PERIOD)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1113 dfa->nodes[node].type = OP_UTF8_PERIOD;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1114 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1115
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1116 /* The search can be in single byte locale. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1117 dfa->mb_cur_max = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1118 dfa->is_utf8 = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1119 dfa->has_mb_node = dfa->nbackref > 0 || has_period;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1120 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1121 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1122
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1123 /* Analyze the structure tree, and calculate "first", "next", "edest",
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1124 "eclosure", and "inveclosure". */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1125
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1126 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1127 analyze (preg)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1128 regex_t *preg;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1129 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1130 re_dfa_t *dfa = (re_dfa_t *) preg->buffer;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1131 reg_errcode_t ret;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1132
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1133 /* Allocate arrays. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1134 dfa->nexts = re_malloc (int, dfa->nodes_alloc);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1135 dfa->org_indices = re_malloc (int, dfa->nodes_alloc);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1136 dfa->edests = re_malloc (re_node_set, dfa->nodes_alloc);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1137 dfa->eclosures = re_malloc (re_node_set, dfa->nodes_alloc);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1138 if (BE (dfa->nexts == NULL || dfa->org_indices == NULL || dfa->edests == NULL
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1139 || dfa->eclosures == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1140 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1141
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1142 dfa->subexp_map = re_malloc (int, preg->re_nsub);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1143 if (dfa->subexp_map != NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1144 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1145 int i;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1146 for (i = 0; i < preg->re_nsub; i++)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1147 dfa->subexp_map[i] = i;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1148 preorder (dfa->str_tree, optimize_subexps, dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1149 for (i = 0; i < preg->re_nsub; i++)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1150 if (dfa->subexp_map[i] != i)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1151 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1152 if (i == preg->re_nsub)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1153 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1154 free (dfa->subexp_map);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1155 dfa->subexp_map = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1156 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1157 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1158
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1159 ret = postorder (dfa->str_tree, lower_subexps, preg);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1160 if (BE (ret != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1161 return ret;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1162 ret = postorder (dfa->str_tree, calc_first, dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1163 if (BE (ret != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1164 return ret;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1165 preorder (dfa->str_tree, calc_next, dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1166 ret = preorder (dfa->str_tree, link_nfa_nodes, dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1167 if (BE (ret != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1168 return ret;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1169 ret = calc_eclosure (dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1170 if (BE (ret != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1171 return ret;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1172
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1173 /* We only need this during the prune_impossible_nodes pass in regexec.c;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1174 skip it if p_i_n will not run, as calc_inveclosure can be quadratic. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1175 if ((!preg->no_sub && preg->re_nsub > 0 && dfa->has_plural_match)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1176 || dfa->nbackref)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1177 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1178 dfa->inveclosures = re_malloc (re_node_set, dfa->nodes_len);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1179 if (BE (dfa->inveclosures == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1180 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1181 ret = calc_inveclosure (dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1182 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1183
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1184 return ret;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1185 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1186
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1187 /* Our parse trees are very unbalanced, so we cannot use a stack to
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1188 implement parse tree visits. Instead, we use parent pointers and
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1189 some hairy code in these two functions. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1190 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1191 postorder (root, fn, extra)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1192 bin_tree_t *root;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1193 reg_errcode_t (fn (void *, bin_tree_t *));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1194 void *extra;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1195 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1196 bin_tree_t *node, *prev;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1197
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1198 for (node = root; ; )
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1199 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1200 /* Descend down the tree, preferably to the left (or to the right
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1201 if that's the only child). */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1202 while (node->left || node->right)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1203 if (node->left)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1204 node = node->left;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1205 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1206 node = node->right;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1207
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1208 do
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1209 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1210 reg_errcode_t err = fn (extra, node);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1211 if (BE (err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1212 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1213 if (node->parent == NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1214 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1215 prev = node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1216 node = node->parent;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1217 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1218 /* Go up while we have a node that is reached from the right. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1219 while (node->right == prev || node->right == NULL);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1220 node = node->right;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1221 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1222 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1223
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1224 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1225 preorder (root, fn, extra)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1226 bin_tree_t *root;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1227 reg_errcode_t (fn (void *, bin_tree_t *));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1228 void *extra;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1229 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1230 bin_tree_t *node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1231
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1232 for (node = root; ; )
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1233 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1234 reg_errcode_t err = fn (extra, node);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1235 if (BE (err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1236 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1237
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1238 /* Go to the left node, or up and to the right. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1239 if (node->left)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1240 node = node->left;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1241 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1242 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1243 bin_tree_t *prev = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1244 while (node->right == prev || node->right == NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1245 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1246 prev = node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1247 node = node->parent;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1248 if (!node)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1249 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1250 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1251 node = node->right;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1252 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1253 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1254 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1255
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1256 /* Optimization pass: if a SUBEXP is entirely contained, strip it and tell
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1257 re_search_internal to map the inner one's opr.idx to this one's. Adjust
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1258 backreferences as well. Requires a preorder visit. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1259 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1260 optimize_subexps (extra, node)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1261 void *extra;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1262 bin_tree_t *node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1263 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1264 re_dfa_t *dfa = (re_dfa_t *) extra;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1265
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1266 if (node->token.type == OP_BACK_REF && dfa->subexp_map)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1267 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1268 int idx = node->token.opr.idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1269 node->token.opr.idx = dfa->subexp_map[idx];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1270 dfa->used_bkref_map |= 1 << node->token.opr.idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1271 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1272
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1273 else if (node->token.type == SUBEXP
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1274 && node->left && node->left->token.type == SUBEXP)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1275 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1276 int other_idx = node->left->token.opr.idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1277
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1278 node->left = node->left->left;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1279 if (node->left)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1280 node->left->parent = node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1281
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1282 dfa->subexp_map[other_idx] = dfa->subexp_map[node->token.opr.idx];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1283 if (other_idx < 8 * sizeof (dfa->used_bkref_map))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1284 dfa->used_bkref_map &= ~(1 << other_idx);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1285 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1286
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1287 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1288 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1289
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1290 /* Lowering pass: Turn each SUBEXP node into the appropriate concatenation
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1291 of OP_OPEN_SUBEXP, the body of the SUBEXP (if any) and OP_CLOSE_SUBEXP. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1292 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1293 lower_subexps (extra, node)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1294 void *extra;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1295 bin_tree_t *node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1296 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1297 regex_t *preg = (regex_t *) extra;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1298 reg_errcode_t err = REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1299
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1300 if (node->left && node->left->token.type == SUBEXP)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1301 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1302 node->left = lower_subexp (&err, preg, node->left);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1303 if (node->left)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1304 node->left->parent = node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1305 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1306 if (node->right && node->right->token.type == SUBEXP)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1307 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1308 node->right = lower_subexp (&err, preg, node->right);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1309 if (node->right)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1310 node->right->parent = node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1311 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1312
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1313 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1314 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1315
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1316 static bin_tree_t *
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1317 lower_subexp (err, preg, node)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1318 reg_errcode_t *err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1319 regex_t *preg;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1320 bin_tree_t *node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1321 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1322 re_dfa_t *dfa = (re_dfa_t *) preg->buffer;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1323 bin_tree_t *body = node->left;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1324 bin_tree_t *op, *cls, *tree1, *tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1325
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1326 if (preg->no_sub
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1327 /* We do not optimize empty subexpressions, because otherwise we may
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1328 have bad CONCAT nodes with NULL children. This is obviously not
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1329 very common, so we do not lose much. An example that triggers
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1330 this case is the sed "script" /\(\)/x. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1331 && node->left != NULL
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1332 && (node->token.opr.idx >= 8 * sizeof (dfa->used_bkref_map)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1333 || !(dfa->used_bkref_map & (1 << node->token.opr.idx))))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1334 return node->left;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1335
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1336 /* Convert the SUBEXP node to the concatenation of an
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1337 OP_OPEN_SUBEXP, the contents, and an OP_CLOSE_SUBEXP. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1338 op = create_tree (dfa, NULL, NULL, OP_OPEN_SUBEXP);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1339 cls = create_tree (dfa, NULL, NULL, OP_CLOSE_SUBEXP);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1340 tree1 = body ? create_tree (dfa, body, cls, CONCAT) : cls;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1341 tree = create_tree (dfa, op, tree1, CONCAT);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1342 if (BE (tree == NULL || tree1 == NULL || op == NULL || cls == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1343 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1344 *err = REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1345 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1346 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1347
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1348 op->token.opr.idx = cls->token.opr.idx = node->token.opr.idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1349 op->token.opt_subexp = cls->token.opt_subexp = node->token.opt_subexp;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1350 return tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1351 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1352
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1353 /* Pass 1 in building the NFA: compute FIRST and create unlinked automaton
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1354 nodes. Requires a postorder visit. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1355 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1356 calc_first (extra, node)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1357 void *extra;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1358 bin_tree_t *node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1359 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1360 re_dfa_t *dfa = (re_dfa_t *) extra;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1361 if (node->token.type == CONCAT)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1362 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1363 node->first = node->left->first;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1364 node->node_idx = node->left->node_idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1365 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1366 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1367 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1368 node->first = node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1369 node->node_idx = re_dfa_add_node (dfa, node->token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1370 if (BE (node->node_idx == -1, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1371 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1372 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1373 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1374 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1375
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1376 /* Pass 2: compute NEXT on the tree. Preorder visit. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1377 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1378 calc_next (extra, node)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1379 void *extra;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1380 bin_tree_t *node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1381 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1382 switch (node->token.type)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1383 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1384 case OP_DUP_ASTERISK:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1385 node->left->next = node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1386 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1387 case CONCAT:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1388 node->left->next = node->right->first;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1389 node->right->next = node->next;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1390 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1391 default:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1392 if (node->left)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1393 node->left->next = node->next;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1394 if (node->right)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1395 node->right->next = node->next;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1396 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1397 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1398 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1399 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1400
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1401 /* Pass 3: link all DFA nodes to their NEXT node (any order will do). */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1402 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1403 link_nfa_nodes (extra, node)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1404 void *extra;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1405 bin_tree_t *node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1406 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1407 re_dfa_t *dfa = (re_dfa_t *) extra;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1408 int idx = node->node_idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1409 reg_errcode_t err = REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1410
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1411 switch (node->token.type)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1412 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1413 case CONCAT:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1414 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1415
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1416 case END_OF_RE:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1417 assert (node->next == NULL);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1418 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1419
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1420 case OP_DUP_ASTERISK:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1421 case OP_ALT:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1422 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1423 int left, right;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1424 dfa->has_plural_match = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1425 if (node->left != NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1426 left = node->left->first->node_idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1427 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1428 left = node->next->node_idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1429 if (node->right != NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1430 right = node->right->first->node_idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1431 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1432 right = node->next->node_idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1433 assert (left > -1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1434 assert (right > -1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1435 err = re_node_set_init_2 (dfa->edests + idx, left, right);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1436 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1437 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1438
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1439 case ANCHOR:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1440 case OP_OPEN_SUBEXP:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1441 case OP_CLOSE_SUBEXP:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1442 err = re_node_set_init_1 (dfa->edests + idx, node->next->node_idx);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1443 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1444
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1445 case OP_BACK_REF:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1446 dfa->nexts[idx] = node->next->node_idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1447 if (node->token.type == OP_BACK_REF)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1448 re_node_set_init_1 (dfa->edests + idx, dfa->nexts[idx]);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1449 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1450
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1451 default:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1452 assert (!IS_EPSILON_NODE (node->token.type));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1453 dfa->nexts[idx] = node->next->node_idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1454 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1455 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1456
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1457 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1458 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1459
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1460 /* Duplicate the epsilon closure of the node ROOT_NODE.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1461 Note that duplicated nodes have constraint INIT_CONSTRAINT in addition
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1462 to their own constraint. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1463
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1464 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1465 duplicate_node_closure (dfa, top_org_node, top_clone_node, root_node,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1466 init_constraint)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1467 re_dfa_t *dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1468 int top_org_node, top_clone_node, root_node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1469 unsigned int init_constraint;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1470 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1471 reg_errcode_t err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1472 int org_node, clone_node, ret;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1473 unsigned int constraint = init_constraint;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1474 for (org_node = top_org_node, clone_node = top_clone_node;;)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1475 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1476 int org_dest, clone_dest;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1477 if (dfa->nodes[org_node].type == OP_BACK_REF)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1478 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1479 /* If the back reference epsilon-transit, its destination must
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1480 also have the constraint. Then duplicate the epsilon closure
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1481 of the destination of the back reference, and store it in
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1482 edests of the back reference. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1483 org_dest = dfa->nexts[org_node];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1484 re_node_set_empty (dfa->edests + clone_node);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1485 err = duplicate_node (&clone_dest, dfa, org_dest, constraint);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1486 if (BE (err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1487 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1488 dfa->nexts[clone_node] = dfa->nexts[org_node];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1489 ret = re_node_set_insert (dfa->edests + clone_node, clone_dest);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1490 if (BE (ret < 0, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1491 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1492 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1493 else if (dfa->edests[org_node].nelem == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1494 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1495 /* In case of the node can't epsilon-transit, don't duplicate the
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1496 destination and store the original destination as the
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1497 destination of the node. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1498 dfa->nexts[clone_node] = dfa->nexts[org_node];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1499 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1500 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1501 else if (dfa->edests[org_node].nelem == 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1502 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1503 /* In case of the node can epsilon-transit, and it has only one
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1504 destination. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1505 org_dest = dfa->edests[org_node].elems[0];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1506 re_node_set_empty (dfa->edests + clone_node);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1507 if (dfa->nodes[org_node].type == ANCHOR)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1508 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1509 /* In case of the node has another constraint, append it. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1510 if (org_node == root_node && clone_node != org_node)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1511 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1512 /* ...but if the node is root_node itself, it means the
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1513 epsilon closure have a loop, then tie it to the
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1514 destination of the root_node. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1515 ret = re_node_set_insert (dfa->edests + clone_node,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1516 org_dest);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1517 if (BE (ret < 0, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1518 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1519 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1520 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1521 constraint |= dfa->nodes[org_node].opr.ctx_type;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1522 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1523 err = duplicate_node (&clone_dest, dfa, org_dest, constraint);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1524 if (BE (err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1525 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1526 ret = re_node_set_insert (dfa->edests + clone_node, clone_dest);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1527 if (BE (ret < 0, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1528 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1529 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1530 else /* dfa->edests[org_node].nelem == 2 */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1531 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1532 /* In case of the node can epsilon-transit, and it has two
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1533 destinations. In the bin_tree_t and DFA, that's '|' and '*'. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1534 org_dest = dfa->edests[org_node].elems[0];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1535 re_node_set_empty (dfa->edests + clone_node);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1536 /* Search for a duplicated node which satisfies the constraint. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1537 clone_dest = search_duplicated_node (dfa, org_dest, constraint);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1538 if (clone_dest == -1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1539 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1540 /* There are no such a duplicated node, create a new one. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1541 err = duplicate_node (&clone_dest, dfa, org_dest, constraint);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1542 if (BE (err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1543 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1544 ret = re_node_set_insert (dfa->edests + clone_node, clone_dest);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1545 if (BE (ret < 0, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1546 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1547 err = duplicate_node_closure (dfa, org_dest, clone_dest,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1548 root_node, constraint);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1549 if (BE (err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1550 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1551 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1552 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1553 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1554 /* There are a duplicated node which satisfy the constraint,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1555 use it to avoid infinite loop. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1556 ret = re_node_set_insert (dfa->edests + clone_node, clone_dest);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1557 if (BE (ret < 0, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1558 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1559 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1560
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1561 org_dest = dfa->edests[org_node].elems[1];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1562 err = duplicate_node (&clone_dest, dfa, org_dest, constraint);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1563 if (BE (err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1564 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1565 ret = re_node_set_insert (dfa->edests + clone_node, clone_dest);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1566 if (BE (ret < 0, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1567 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1568 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1569 org_node = org_dest;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1570 clone_node = clone_dest;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1571 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1572 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1573 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1574
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1575 /* Search for a node which is duplicated from the node ORG_NODE, and
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1576 satisfies the constraint CONSTRAINT. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1577
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1578 static int
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1579 search_duplicated_node (dfa, org_node, constraint)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1580 re_dfa_t *dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1581 int org_node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1582 unsigned int constraint;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1583 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1584 int idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1585 for (idx = dfa->nodes_len - 1; dfa->nodes[idx].duplicated && idx > 0; --idx)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1586 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1587 if (org_node == dfa->org_indices[idx]
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1588 && constraint == dfa->nodes[idx].constraint)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1589 return idx; /* Found. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1590 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1591 return -1; /* Not found. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1592 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1593
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1594 /* Duplicate the node whose index is ORG_IDX and set the constraint CONSTRAINT.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1595 The new index will be stored in NEW_IDX and return REG_NOERROR if succeeded,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1596 otherwise return the error code. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1597
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1598 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1599 duplicate_node (new_idx, dfa, org_idx, constraint)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1600 re_dfa_t *dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1601 int *new_idx, org_idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1602 unsigned int constraint;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1603 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1604 int dup_idx = re_dfa_add_node (dfa, dfa->nodes[org_idx]);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1605 if (BE (dup_idx == -1, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1606 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1607 dfa->nodes[dup_idx].constraint = constraint;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1608 if (dfa->nodes[org_idx].type == ANCHOR)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1609 dfa->nodes[dup_idx].constraint |= dfa->nodes[org_idx].opr.ctx_type;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1610 dfa->nodes[dup_idx].duplicated = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1611
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1612 /* Store the index of the original node. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1613 dfa->org_indices[dup_idx] = org_idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1614 *new_idx = dup_idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1615 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1616 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1617
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1618 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1619 calc_inveclosure (dfa)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1620 re_dfa_t *dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1621 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1622 int src, idx, ret;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1623 for (idx = 0; idx < dfa->nodes_len; ++idx)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1624 re_node_set_init_empty (dfa->inveclosures + idx);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1625
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1626 for (src = 0; src < dfa->nodes_len; ++src)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1627 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1628 int *elems = dfa->eclosures[src].elems;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1629 for (idx = 0; idx < dfa->eclosures[src].nelem; ++idx)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1630 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1631 ret = re_node_set_insert_last (dfa->inveclosures + elems[idx], src);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1632 if (BE (ret == -1, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1633 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1634 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1635 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1636
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1637 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1638 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1639
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1640 /* Calculate "eclosure" for all the node in DFA. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1641
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1642 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1643 calc_eclosure (dfa)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1644 re_dfa_t *dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1645 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1646 int node_idx, incomplete;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1647 #ifdef DEBUG
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1648 assert (dfa->nodes_len > 0);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1649 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1650 incomplete = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1651 /* For each nodes, calculate epsilon closure. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1652 for (node_idx = 0; ; ++node_idx)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1653 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1654 reg_errcode_t err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1655 re_node_set eclosure_elem;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1656 if (node_idx == dfa->nodes_len)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1657 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1658 if (!incomplete)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1659 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1660 incomplete = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1661 node_idx = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1662 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1663
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1664 #ifdef DEBUG
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1665 assert (dfa->eclosures[node_idx].nelem != -1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1666 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1667
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1668 /* If we have already calculated, skip it. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1669 if (dfa->eclosures[node_idx].nelem != 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1670 continue;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1671 /* Calculate epsilon closure of `node_idx'. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1672 err = calc_eclosure_iter (&eclosure_elem, dfa, node_idx, 1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1673 if (BE (err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1674 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1675
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1676 if (dfa->eclosures[node_idx].nelem == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1677 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1678 incomplete = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1679 re_node_set_free (&eclosure_elem);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1680 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1681 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1682 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1683 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1684
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1685 /* Calculate epsilon closure of NODE. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1686
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1687 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1688 calc_eclosure_iter (new_set, dfa, node, root)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1689 re_node_set *new_set;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1690 re_dfa_t *dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1691 int node, root;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1692 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1693 reg_errcode_t err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1694 unsigned int constraint;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1695 int i, incomplete;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1696 re_node_set eclosure;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1697 incomplete = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1698 err = re_node_set_alloc (&eclosure, dfa->edests[node].nelem + 1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1699 if (BE (err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1700 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1701
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1702 /* This indicates that we are calculating this node now.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1703 We reference this value to avoid infinite loop. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1704 dfa->eclosures[node].nelem = -1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1705
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1706 constraint = ((dfa->nodes[node].type == ANCHOR)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1707 ? dfa->nodes[node].opr.ctx_type : 0);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1708 /* If the current node has constraints, duplicate all nodes.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1709 Since they must inherit the constraints. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1710 if (constraint
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1711 && dfa->edests[node].nelem
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1712 && !dfa->nodes[dfa->edests[node].elems[0]].duplicated)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1713 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1714 int org_node, cur_node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1715 org_node = cur_node = node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1716 err = duplicate_node_closure (dfa, node, node, node, constraint);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1717 if (BE (err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1718 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1719 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1720
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1721 /* Expand each epsilon destination nodes. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1722 if (IS_EPSILON_NODE(dfa->nodes[node].type))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1723 for (i = 0; i < dfa->edests[node].nelem; ++i)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1724 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1725 re_node_set eclosure_elem;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1726 int edest = dfa->edests[node].elems[i];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1727 /* If calculating the epsilon closure of `edest' is in progress,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1728 return intermediate result. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1729 if (dfa->eclosures[edest].nelem == -1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1730 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1731 incomplete = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1732 continue;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1733 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1734 /* If we haven't calculated the epsilon closure of `edest' yet,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1735 calculate now. Otherwise use calculated epsilon closure. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1736 if (dfa->eclosures[edest].nelem == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1737 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1738 err = calc_eclosure_iter (&eclosure_elem, dfa, edest, 0);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1739 if (BE (err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1740 return err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1741 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1742 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1743 eclosure_elem = dfa->eclosures[edest];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1744 /* Merge the epsilon closure of `edest'. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1745 re_node_set_merge (&eclosure, &eclosure_elem);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1746 /* If the epsilon closure of `edest' is incomplete,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1747 the epsilon closure of this node is also incomplete. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1748 if (dfa->eclosures[edest].nelem == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1749 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1750 incomplete = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1751 re_node_set_free (&eclosure_elem);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1752 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1753 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1754
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1755 /* Epsilon closures include itself. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1756 re_node_set_insert (&eclosure, node);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1757 if (incomplete && !root)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1758 dfa->eclosures[node].nelem = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1759 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1760 dfa->eclosures[node] = eclosure;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1761 *new_set = eclosure;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1762 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1763 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1764
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1765 /* Functions for token which are used in the parser. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1766
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1767 /* Fetch a token from INPUT.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1768 We must not use this function inside bracket expressions. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1769
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1770 static void
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1771 fetch_token (result, input, syntax)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1772 re_token_t *result;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1773 re_string_t *input;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1774 reg_syntax_t syntax;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1775 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1776 re_string_skip_bytes (input, peek_token (result, input, syntax));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1777 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1778
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1779 /* Peek a token from INPUT, and return the length of the token.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1780 We must not use this function inside bracket expressions. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1781
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1782 static int
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1783 peek_token (token, input, syntax)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1784 re_token_t *token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1785 re_string_t *input;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1786 reg_syntax_t syntax;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1787 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1788 unsigned char c;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1789
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1790 if (re_string_eoi (input))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1791 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1792 token->type = END_OF_RE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1793 return 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1794 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1795
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1796 c = re_string_peek_byte (input, 0);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1797 token->opr.c = c;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1798
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1799 token->word_char = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1800 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1801 token->mb_partial = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1802 if (input->mb_cur_max > 1 &&
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1803 !re_string_first_byte (input, re_string_cur_idx (input)))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1804 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1805 token->type = CHARACTER;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1806 token->mb_partial = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1807 return 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1808 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1809 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1810 if (c == '\\')
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1811 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1812 unsigned char c2;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1813 if (re_string_cur_idx (input) + 1 >= re_string_length (input))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1814 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1815 token->type = BACK_SLASH;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1816 return 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1817 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1818
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1819 c2 = re_string_peek_byte_case (input, 1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1820 token->opr.c = c2;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1821 token->type = CHARACTER;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1822 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1823 if (input->mb_cur_max > 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1824 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1825 wint_t wc = re_string_wchar_at (input,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1826 re_string_cur_idx (input) + 1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1827 token->word_char = IS_WIDE_WORD_CHAR (wc) != 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1828 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1829 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1830 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1831 token->word_char = IS_WORD_CHAR (c2) != 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1832
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1833 switch (c2)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1834 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1835 case '|':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1836 if (!(syntax & RE_LIMITED_OPS) && !(syntax & RE_NO_BK_VBAR))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1837 token->type = OP_ALT;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1838 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1839 case '1': case '2': case '3': case '4': case '5':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1840 case '6': case '7': case '8': case '9':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1841 if (!(syntax & RE_NO_BK_REFS))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1842 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1843 token->type = OP_BACK_REF;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1844 token->opr.idx = c2 - '1';
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1845 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1846 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1847 case '<':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1848 if (!(syntax & RE_NO_GNU_OPS))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1849 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1850 token->type = ANCHOR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1851 token->opr.ctx_type = WORD_FIRST;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1852 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1853 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1854 case '>':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1855 if (!(syntax & RE_NO_GNU_OPS))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1856 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1857 token->type = ANCHOR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1858 token->opr.ctx_type = WORD_LAST;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1859 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1860 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1861 case 'b':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1862 if (!(syntax & RE_NO_GNU_OPS))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1863 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1864 token->type = ANCHOR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1865 token->opr.ctx_type = WORD_DELIM;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1866 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1867 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1868 case 'B':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1869 if (!(syntax & RE_NO_GNU_OPS))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1870 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1871 token->type = ANCHOR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1872 token->opr.ctx_type = NOT_WORD_DELIM;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1873 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1874 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1875 case 'w':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1876 if (!(syntax & RE_NO_GNU_OPS))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1877 token->type = OP_WORD;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1878 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1879 case 'W':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1880 if (!(syntax & RE_NO_GNU_OPS))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1881 token->type = OP_NOTWORD;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1882 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1883 case 's':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1884 if (!(syntax & RE_NO_GNU_OPS))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1885 token->type = OP_SPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1886 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1887 case 'S':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1888 if (!(syntax & RE_NO_GNU_OPS))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1889 token->type = OP_NOTSPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1890 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1891 case '`':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1892 if (!(syntax & RE_NO_GNU_OPS))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1893 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1894 token->type = ANCHOR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1895 token->opr.ctx_type = BUF_FIRST;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1896 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1897 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1898 case '\'':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1899 if (!(syntax & RE_NO_GNU_OPS))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1900 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1901 token->type = ANCHOR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1902 token->opr.ctx_type = BUF_LAST;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1903 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1904 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1905 case '(':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1906 if (!(syntax & RE_NO_BK_PARENS))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1907 token->type = OP_OPEN_SUBEXP;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1908 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1909 case ')':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1910 if (!(syntax & RE_NO_BK_PARENS))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1911 token->type = OP_CLOSE_SUBEXP;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1912 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1913 case '+':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1914 if (!(syntax & RE_LIMITED_OPS) && (syntax & RE_BK_PLUS_QM))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1915 token->type = OP_DUP_PLUS;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1916 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1917 case '?':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1918 if (!(syntax & RE_LIMITED_OPS) && (syntax & RE_BK_PLUS_QM))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1919 token->type = OP_DUP_QUESTION;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1920 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1921 case '{':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1922 if ((syntax & RE_INTERVALS) && (!(syntax & RE_NO_BK_BRACES)))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1923 token->type = OP_OPEN_DUP_NUM;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1924 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1925 case '}':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1926 if ((syntax & RE_INTERVALS) && (!(syntax & RE_NO_BK_BRACES)))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1927 token->type = OP_CLOSE_DUP_NUM;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1928 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1929 default:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1930 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1931 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1932 return 2;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1933 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1934
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1935 token->type = CHARACTER;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1936 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1937 if (input->mb_cur_max > 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1938 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1939 wint_t wc = re_string_wchar_at (input, re_string_cur_idx (input));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1940 token->word_char = IS_WIDE_WORD_CHAR (wc) != 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1941 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1942 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1943 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1944 token->word_char = IS_WORD_CHAR (token->opr.c);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1945
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1946 switch (c)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1947 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1948 case '\n':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1949 if (syntax & RE_NEWLINE_ALT)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1950 token->type = OP_ALT;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1951 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1952 case '|':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1953 if (!(syntax & RE_LIMITED_OPS) && (syntax & RE_NO_BK_VBAR))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1954 token->type = OP_ALT;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1955 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1956 case '*':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1957 token->type = OP_DUP_ASTERISK;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1958 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1959 case '+':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1960 if (!(syntax & RE_LIMITED_OPS) && !(syntax & RE_BK_PLUS_QM))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1961 token->type = OP_DUP_PLUS;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1962 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1963 case '?':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1964 if (!(syntax & RE_LIMITED_OPS) && !(syntax & RE_BK_PLUS_QM))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1965 token->type = OP_DUP_QUESTION;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1966 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1967 case '{':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1968 if ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1969 token->type = OP_OPEN_DUP_NUM;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1970 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1971 case '}':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1972 if ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1973 token->type = OP_CLOSE_DUP_NUM;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1974 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1975 case '(':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1976 if (syntax & RE_NO_BK_PARENS)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1977 token->type = OP_OPEN_SUBEXP;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1978 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1979 case ')':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1980 if (syntax & RE_NO_BK_PARENS)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1981 token->type = OP_CLOSE_SUBEXP;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1982 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1983 case '[':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1984 token->type = OP_OPEN_BRACKET;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1985 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1986 case '.':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1987 token->type = OP_PERIOD;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1988 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1989 case '^':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1990 if (!(syntax & (RE_CONTEXT_INDEP_ANCHORS | RE_CARET_ANCHORS_HERE)) &&
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1991 re_string_cur_idx (input) != 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1992 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1993 char prev = re_string_peek_byte (input, -1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1994 if (!(syntax & RE_NEWLINE_ALT) || prev != '\n')
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1995 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1996 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1997 token->type = ANCHOR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1998 token->opr.ctx_type = LINE_FIRST;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1999 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2000 case '$':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2001 if (!(syntax & RE_CONTEXT_INDEP_ANCHORS) &&
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2002 re_string_cur_idx (input) + 1 != re_string_length (input))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2003 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2004 re_token_t next;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2005 re_string_skip_bytes (input, 1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2006 peek_token (&next, input, syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2007 re_string_skip_bytes (input, -1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2008 if (next.type != OP_ALT && next.type != OP_CLOSE_SUBEXP)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2009 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2010 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2011 token->type = ANCHOR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2012 token->opr.ctx_type = LINE_LAST;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2013 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2014 default:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2015 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2016 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2017 return 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2018 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2019
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2020 /* Peek a token from INPUT, and return the length of the token.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2021 We must not use this function out of bracket expressions. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2022
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2023 static int
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2024 peek_token_bracket (token, input, syntax)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2025 re_token_t *token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2026 re_string_t *input;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2027 reg_syntax_t syntax;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2028 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2029 unsigned char c;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2030 if (re_string_eoi (input))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2031 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2032 token->type = END_OF_RE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2033 return 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2034 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2035 c = re_string_peek_byte (input, 0);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2036 token->opr.c = c;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2037
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2038 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2039 if (input->mb_cur_max > 1 &&
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2040 !re_string_first_byte (input, re_string_cur_idx (input)))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2041 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2042 token->type = CHARACTER;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2043 return 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2044 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2045 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2046
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2047 if (c == '\\' && (syntax & RE_BACKSLASH_ESCAPE_IN_LISTS)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2048 && re_string_cur_idx (input) + 1 < re_string_length (input))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2049 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2050 /* In this case, '\' escape a character. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2051 unsigned char c2;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2052 re_string_skip_bytes (input, 1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2053 c2 = re_string_peek_byte (input, 0);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2054 token->opr.c = c2;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2055 token->type = CHARACTER;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2056 return 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2057 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2058 if (c == '[') /* '[' is a special char in a bracket exps. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2059 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2060 unsigned char c2;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2061 int token_len;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2062 if (re_string_cur_idx (input) + 1 < re_string_length (input))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2063 c2 = re_string_peek_byte (input, 1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2064 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2065 c2 = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2066 token->opr.c = c2;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2067 token_len = 2;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2068 switch (c2)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2069 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2070 case '.':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2071 token->type = OP_OPEN_COLL_ELEM;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2072 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2073 case '=':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2074 token->type = OP_OPEN_EQUIV_CLASS;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2075 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2076 case ':':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2077 if (syntax & RE_CHAR_CLASSES)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2078 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2079 token->type = OP_OPEN_CHAR_CLASS;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2080 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2081 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2082 /* else fall through. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2083 default:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2084 token->type = CHARACTER;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2085 token->opr.c = c;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2086 token_len = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2087 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2088 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2089 return token_len;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2090 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2091 switch (c)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2092 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2093 case '-':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2094 token->type = OP_CHARSET_RANGE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2095 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2096 case ']':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2097 token->type = OP_CLOSE_BRACKET;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2098 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2099 case '^':
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2100 token->type = OP_NON_MATCH_LIST;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2101 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2102 default:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2103 token->type = CHARACTER;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2104 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2105 return 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2106 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2107
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2108 /* Functions for parser. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2109
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2110 /* Entry point of the parser.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2111 Parse the regular expression REGEXP and return the structure tree.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2112 If an error is occured, ERR is set by error code, and return NULL.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2113 This function build the following tree, from regular expression <reg_exp>:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2114 CAT
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2115 / \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2116 / \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2117 <reg_exp> EOR
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2118
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2119 CAT means concatenation.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2120 EOR means end of regular expression. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2121
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2122 static bin_tree_t *
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2123 parse (regexp, preg, syntax, err)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2124 re_string_t *regexp;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2125 regex_t *preg;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2126 reg_syntax_t syntax;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2127 reg_errcode_t *err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2128 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2129 re_dfa_t *dfa = (re_dfa_t *) preg->buffer;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2130 bin_tree_t *tree, *eor, *root;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2131 re_token_t current_token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2132 dfa->syntax = syntax;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2133 fetch_token (&current_token, regexp, syntax | RE_CARET_ANCHORS_HERE);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2134 tree = parse_reg_exp (regexp, preg, &current_token, syntax, 0, err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2135 if (BE (*err != REG_NOERROR && tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2136 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2137 eor = create_tree (dfa, NULL, NULL, END_OF_RE);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2138 if (tree != NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2139 root = create_tree (dfa, tree, eor, CONCAT);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2140 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2141 root = eor;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2142 if (BE (eor == NULL || root == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2143 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2144 *err = REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2145 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2146 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2147 return root;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2148 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2149
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2150 /* This function build the following tree, from regular expression
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2151 <branch1>|<branch2>:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2152 ALT
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2153 / \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2154 / \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2155 <branch1> <branch2>
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2156
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2157 ALT means alternative, which represents the operator `|'. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2158
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2159 static bin_tree_t *
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2160 parse_reg_exp (regexp, preg, token, syntax, nest, err)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2161 re_string_t *regexp;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2162 regex_t *preg;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2163 re_token_t *token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2164 reg_syntax_t syntax;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2165 int nest;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2166 reg_errcode_t *err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2167 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2168 re_dfa_t *dfa = (re_dfa_t *) preg->buffer;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2169 bin_tree_t *tree, *branch = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2170 tree = parse_branch (regexp, preg, token, syntax, nest, err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2171 if (BE (*err != REG_NOERROR && tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2172 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2173
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2174 while (token->type == OP_ALT)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2175 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2176 fetch_token (token, regexp, syntax | RE_CARET_ANCHORS_HERE);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2177 if (token->type != OP_ALT && token->type != END_OF_RE
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2178 && (nest == 0 || token->type != OP_CLOSE_SUBEXP))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2179 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2180 branch = parse_branch (regexp, preg, token, syntax, nest, err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2181 if (BE (*err != REG_NOERROR && branch == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2182 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2183 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2184 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2185 branch = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2186 tree = create_tree (dfa, tree, branch, OP_ALT);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2187 if (BE (tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2188 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2189 *err = REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2190 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2191 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2192 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2193 return tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2194 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2195
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2196 /* This function build the following tree, from regular expression
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2197 <exp1><exp2>:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2198 CAT
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2199 / \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2200 / \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2201 <exp1> <exp2>
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2202
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2203 CAT means concatenation. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2204
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2205 static bin_tree_t *
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2206 parse_branch (regexp, preg, token, syntax, nest, err)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2207 re_string_t *regexp;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2208 regex_t *preg;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2209 re_token_t *token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2210 reg_syntax_t syntax;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2211 int nest;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2212 reg_errcode_t *err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2213 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2214 bin_tree_t *tree, *exp;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2215 re_dfa_t *dfa = (re_dfa_t *) preg->buffer;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2216 tree = parse_expression (regexp, preg, token, syntax, nest, err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2217 if (BE (*err != REG_NOERROR && tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2218 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2219
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2220 while (token->type != OP_ALT && token->type != END_OF_RE
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2221 && (nest == 0 || token->type != OP_CLOSE_SUBEXP))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2222 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2223 exp = parse_expression (regexp, preg, token, syntax, nest, err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2224 if (BE (*err != REG_NOERROR && exp == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2225 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2226 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2227 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2228 if (tree != NULL && exp != NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2229 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2230 tree = create_tree (dfa, tree, exp, CONCAT);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2231 if (tree == NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2232 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2233 *err = REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2234 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2235 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2236 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2237 else if (tree == NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2238 tree = exp;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2239 /* Otherwise exp == NULL, we don't need to create new tree. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2240 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2241 return tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2242 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2243
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2244 /* This function build the following tree, from regular expression a*:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2245 *
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2246 |
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2247 a
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2248 */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2249
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2250 static bin_tree_t *
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2251 parse_expression (regexp, preg, token, syntax, nest, err)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2252 re_string_t *regexp;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2253 regex_t *preg;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2254 re_token_t *token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2255 reg_syntax_t syntax;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2256 int nest;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2257 reg_errcode_t *err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2258 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2259 re_dfa_t *dfa = (re_dfa_t *) preg->buffer;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2260 bin_tree_t *tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2261 switch (token->type)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2262 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2263 case CHARACTER:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2264 tree = create_token_tree (dfa, NULL, NULL, token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2265 if (BE (tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2266 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2267 *err = REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2268 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2269 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2270 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2271 if (dfa->mb_cur_max > 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2272 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2273 while (!re_string_eoi (regexp)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2274 && !re_string_first_byte (regexp, re_string_cur_idx (regexp)))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2275 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2276 bin_tree_t *mbc_remain;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2277 fetch_token (token, regexp, syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2278 mbc_remain = create_token_tree (dfa, NULL, NULL, token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2279 tree = create_tree (dfa, tree, mbc_remain, CONCAT);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2280 if (BE (mbc_remain == NULL || tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2281 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2282 *err = REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2283 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2284 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2285 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2286 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2287 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2288 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2289 case OP_OPEN_SUBEXP:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2290 tree = parse_sub_exp (regexp, preg, token, syntax, nest + 1, err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2291 if (BE (*err != REG_NOERROR && tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2292 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2293 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2294 case OP_OPEN_BRACKET:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2295 tree = parse_bracket_exp (regexp, dfa, token, syntax, err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2296 if (BE (*err != REG_NOERROR && tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2297 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2298 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2299 case OP_BACK_REF:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2300 if (!BE (dfa->completed_bkref_map & (1 << token->opr.idx), 1))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2301 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2302 *err = REG_ESUBREG;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2303 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2304 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2305 dfa->used_bkref_map |= 1 << token->opr.idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2306 tree = create_token_tree (dfa, NULL, NULL, token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2307 if (BE (tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2308 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2309 *err = REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2310 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2311 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2312 ++dfa->nbackref;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2313 dfa->has_mb_node = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2314 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2315 case OP_OPEN_DUP_NUM:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2316 if (syntax & RE_CONTEXT_INVALID_DUP)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2317 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2318 *err = REG_BADRPT;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2319 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2320 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2321 /* FALLTHROUGH */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2322 case OP_DUP_ASTERISK:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2323 case OP_DUP_PLUS:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2324 case OP_DUP_QUESTION:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2325 if (syntax & RE_CONTEXT_INVALID_OPS)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2326 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2327 *err = REG_BADRPT;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2328 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2329 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2330 else if (syntax & RE_CONTEXT_INDEP_OPS)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2331 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2332 fetch_token (token, regexp, syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2333 return parse_expression (regexp, preg, token, syntax, nest, err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2334 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2335 /* else fall through */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2336 case OP_CLOSE_SUBEXP:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2337 if ((token->type == OP_CLOSE_SUBEXP) &&
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2338 !(syntax & RE_UNMATCHED_RIGHT_PAREN_ORD))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2339 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2340 *err = REG_ERPAREN;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2341 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2342 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2343 /* else fall through */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2344 case OP_CLOSE_DUP_NUM:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2345 /* We treat it as a normal character. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2346
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2347 /* Then we can these characters as normal characters. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2348 token->type = CHARACTER;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2349 /* mb_partial and word_char bits should be initialized already
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2350 by peek_token. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2351 tree = create_token_tree (dfa, NULL, NULL, token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2352 if (BE (tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2353 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2354 *err = REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2355 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2356 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2357 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2358 case ANCHOR:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2359 if ((token->opr.ctx_type
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2360 & (WORD_DELIM | NOT_WORD_DELIM | WORD_FIRST | WORD_LAST))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2361 && dfa->word_ops_used == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2362 init_word_char (dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2363 if (token->opr.ctx_type == WORD_DELIM
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2364 || token->opr.ctx_type == NOT_WORD_DELIM)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2365 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2366 bin_tree_t *tree_first, *tree_last;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2367 if (token->opr.ctx_type == WORD_DELIM)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2368 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2369 token->opr.ctx_type = WORD_FIRST;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2370 tree_first = create_token_tree (dfa, NULL, NULL, token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2371 token->opr.ctx_type = WORD_LAST;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2372 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2373 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2374 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2375 token->opr.ctx_type = INSIDE_WORD;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2376 tree_first = create_token_tree (dfa, NULL, NULL, token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2377 token->opr.ctx_type = INSIDE_NOTWORD;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2378 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2379 tree_last = create_token_tree (dfa, NULL, NULL, token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2380 tree = create_tree (dfa, tree_first, tree_last, OP_ALT);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2381 if (BE (tree_first == NULL || tree_last == NULL || tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2382 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2383 *err = REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2384 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2385 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2386 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2387 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2388 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2389 tree = create_token_tree (dfa, NULL, NULL, token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2390 if (BE (tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2391 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2392 *err = REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2393 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2394 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2395 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2396 /* We must return here, since ANCHORs can't be followed
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2397 by repetition operators.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2398 eg. RE"^*" is invalid or "<ANCHOR(^)><CHAR(*)>",
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2399 it must not be "<ANCHOR(^)><REPEAT(*)>". */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2400 fetch_token (token, regexp, syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2401 return tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2402 case OP_PERIOD:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2403 tree = create_token_tree (dfa, NULL, NULL, token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2404 if (BE (tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2405 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2406 *err = REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2407 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2408 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2409 if (dfa->mb_cur_max > 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2410 dfa->has_mb_node = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2411 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2412 case OP_WORD:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2413 case OP_NOTWORD:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2414 tree = build_charclass_op (dfa, regexp->trans,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2415 (const unsigned char *) "alnum",
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2416 (const unsigned char *) "_",
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2417 token->type == OP_NOTWORD, err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2418 if (BE (*err != REG_NOERROR && tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2419 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2420 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2421 case OP_SPACE:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2422 case OP_NOTSPACE:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2423 tree = build_charclass_op (dfa, regexp->trans,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2424 (const unsigned char *) "space",
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2425 (const unsigned char *) "",
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2426 token->type == OP_NOTSPACE, err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2427 if (BE (*err != REG_NOERROR && tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2428 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2429 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2430 case OP_ALT:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2431 case END_OF_RE:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2432 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2433 case BACK_SLASH:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2434 *err = REG_EESCAPE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2435 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2436 default:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2437 /* Must not happen? */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2438 #ifdef DEBUG
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2439 assert (0);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2440 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2441 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2442 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2443 fetch_token (token, regexp, syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2444
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2445 while (token->type == OP_DUP_ASTERISK || token->type == OP_DUP_PLUS
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2446 || token->type == OP_DUP_QUESTION || token->type == OP_OPEN_DUP_NUM)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2447 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2448 tree = parse_dup_op (tree, regexp, dfa, token, syntax, err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2449 if (BE (*err != REG_NOERROR && tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2450 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2451 /* In BRE consecutive duplications are not allowed. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2452 if ((syntax & RE_CONTEXT_INVALID_DUP)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2453 && (token->type == OP_DUP_ASTERISK
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2454 || token->type == OP_OPEN_DUP_NUM))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2455 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2456 *err = REG_BADRPT;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2457 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2458 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2459 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2460
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2461 return tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2462 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2463
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2464 /* This function build the following tree, from regular expression
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2465 (<reg_exp>):
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2466 SUBEXP
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2467 |
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2468 <reg_exp>
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2469 */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2470
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2471 static bin_tree_t *
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2472 parse_sub_exp (regexp, preg, token, syntax, nest, err)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2473 re_string_t *regexp;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2474 regex_t *preg;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2475 re_token_t *token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2476 reg_syntax_t syntax;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2477 int nest;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2478 reg_errcode_t *err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2479 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2480 re_dfa_t *dfa = (re_dfa_t *) preg->buffer;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2481 bin_tree_t *tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2482 size_t cur_nsub;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2483 cur_nsub = preg->re_nsub++;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2484
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2485 fetch_token (token, regexp, syntax | RE_CARET_ANCHORS_HERE);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2486
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2487 /* The subexpression may be a null string. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2488 if (token->type == OP_CLOSE_SUBEXP)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2489 tree = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2490 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2491 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2492 tree = parse_reg_exp (regexp, preg, token, syntax, nest, err);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2493 if (BE (*err == REG_NOERROR && token->type != OP_CLOSE_SUBEXP, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2494 *err = REG_EPAREN;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2495 if (BE (*err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2496 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2497 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2498 dfa->completed_bkref_map |= 1 << cur_nsub;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2499
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2500 tree = create_tree (dfa, tree, NULL, SUBEXP);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2501 if (BE (tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2502 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2503 *err = REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2504 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2505 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2506 tree->token.opr.idx = cur_nsub;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2507 return tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2508 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2509
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2510 /* This function parse repetition operators like "*", "+", "{1,3}" etc. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2511
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2512 static bin_tree_t *
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2513 parse_dup_op (elem, regexp, dfa, token, syntax, err)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2514 bin_tree_t *elem;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2515 re_string_t *regexp;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2516 re_dfa_t *dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2517 re_token_t *token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2518 reg_syntax_t syntax;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2519 reg_errcode_t *err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2520 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2521 bin_tree_t *tree = NULL, *old_tree = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2522 int i, start, end, start_idx = re_string_cur_idx (regexp);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2523 re_token_t start_token = *token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2524
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2525 if (token->type == OP_OPEN_DUP_NUM)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2526 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2527 end = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2528 start = fetch_number (regexp, token, syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2529 if (start == -1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2530 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2531 if (token->type == CHARACTER && token->opr.c == ',')
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2532 start = 0; /* We treat "{,m}" as "{0,m}". */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2533 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2534 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2535 *err = REG_BADBR; /* <re>{} is invalid. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2536 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2537 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2538 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2539 if (BE (start != -2, 1))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2540 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2541 /* We treat "{n}" as "{n,n}". */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2542 end = ((token->type == OP_CLOSE_DUP_NUM) ? start
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2543 : ((token->type == CHARACTER && token->opr.c == ',')
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2544 ? fetch_number (regexp, token, syntax) : -2));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2545 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2546 if (BE (start == -2 || end == -2, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2547 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2548 /* Invalid sequence. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2549 if (BE (!(syntax & RE_INVALID_INTERVAL_ORD), 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2550 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2551 if (token->type == END_OF_RE)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2552 *err = REG_EBRACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2553 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2554 *err = REG_BADBR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2555
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2556 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2557 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2558
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2559 /* If the syntax bit is set, rollback. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2560 re_string_set_index (regexp, start_idx);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2561 *token = start_token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2562 token->type = CHARACTER;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2563 /* mb_partial and word_char bits should be already initialized by
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2564 peek_token. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2565 return elem;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2566 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2567
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2568 if (BE (end != -1 && start > end, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2569 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2570 /* First number greater than second. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2571 *err = REG_BADBR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2572 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2573 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2574 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2575 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2576 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2577 start = (token->type == OP_DUP_PLUS) ? 1 : 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2578 end = (token->type == OP_DUP_QUESTION) ? 1 : -1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2579 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2580
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2581 fetch_token (token, regexp, syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2582
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2583 if (BE (elem == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2584 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2585 if (BE (start == 0 && end == 0, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2586 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2587 postorder (elem, free_tree, NULL);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2588 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2589 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2590
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2591 /* Extract "<re>{n,m}" to "<re><re>...<re><re>{0,<m-n>}". */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2592 if (BE (start > 0, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2593 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2594 tree = elem;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2595 for (i = 2; i <= start; ++i)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2596 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2597 elem = duplicate_tree (elem, dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2598 tree = create_tree (dfa, tree, elem, CONCAT);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2599 if (BE (elem == NULL || tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2600 goto parse_dup_op_espace;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2601 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2602
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2603 if (start == end)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2604 return tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2605
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2606 /* Duplicate ELEM before it is marked optional. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2607 elem = duplicate_tree (elem, dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2608 old_tree = tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2609 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2610 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2611 old_tree = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2612
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2613 if (elem->token.type == SUBEXP)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2614 postorder (elem, mark_opt_subexp, (void *) (long) elem->token.opr.idx);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2615
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2616 tree = create_tree (dfa, elem, NULL, (end == -1 ? OP_DUP_ASTERISK : OP_ALT));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2617 if (BE (tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2618 goto parse_dup_op_espace;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2619
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2620 /* This loop is actually executed only when end != -1,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2621 to rewrite <re>{0,n} as (<re>(<re>...<re>?)?)?... We have
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2622 already created the start+1-th copy. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2623 for (i = start + 2; i <= end; ++i)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2624 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2625 elem = duplicate_tree (elem, dfa);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2626 tree = create_tree (dfa, tree, elem, CONCAT);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2627 if (BE (elem == NULL || tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2628 goto parse_dup_op_espace;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2629
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2630 tree = create_tree (dfa, tree, NULL, OP_ALT);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2631 if (BE (tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2632 goto parse_dup_op_espace;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2633 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2634
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2635 if (old_tree)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2636 tree = create_tree (dfa, old_tree, tree, CONCAT);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2637
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2638 return tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2639
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2640 parse_dup_op_espace:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2641 *err = REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2642 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2643 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2644
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2645 /* Size of the names for collating symbol/equivalence_class/character_class.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2646 I'm not sure, but maybe enough. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2647 #define BRACKET_NAME_BUF_SIZE 32
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2648
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2649 #ifndef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2650 /* Local function for parse_bracket_exp only used in case of NOT _LIBC.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2651 Build the range expression which starts from START_ELEM, and ends
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2652 at END_ELEM. The result are written to MBCSET and SBCSET.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2653 RANGE_ALLOC is the allocated size of mbcset->range_starts, and
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2654 mbcset->range_ends, is a pointer argument sinse we may
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2655 update it. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2656
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2657 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2658 # ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2659 build_range_exp (sbcset, mbcset, range_alloc, start_elem, end_elem)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2660 re_charset_t *mbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2661 int *range_alloc;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2662 # else /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2663 build_range_exp (sbcset, start_elem, end_elem)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2664 # endif /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2665 re_bitset_ptr_t sbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2666 bracket_elem_t *start_elem, *end_elem;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2667 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2668 unsigned int start_ch, end_ch;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2669 /* Equivalence Classes and Character Classes can't be a range start/end. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2670 if (BE (start_elem->type == EQUIV_CLASS || start_elem->type == CHAR_CLASS
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2671 || end_elem->type == EQUIV_CLASS || end_elem->type == CHAR_CLASS,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2672 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2673 return REG_ERANGE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2674
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2675 /* We can handle no multi character collating elements without libc
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2676 support. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2677 if (BE ((start_elem->type == COLL_SYM
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2678 && strlen ((char *) start_elem->opr.name) > 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2679 || (end_elem->type == COLL_SYM
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2680 && strlen ((char *) end_elem->opr.name) > 1), 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2681 return REG_ECOLLATE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2682
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2683 # ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2684 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2685 wchar_t wc, start_wc, end_wc;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2686 wchar_t cmp_buf[6] = {L'\0', L'\0', L'\0', L'\0', L'\0', L'\0'};
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2687
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2688 start_ch = ((start_elem->type == SB_CHAR) ? start_elem->opr.ch
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2689 : ((start_elem->type == COLL_SYM) ? start_elem->opr.name[0]
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2690 : 0));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2691 end_ch = ((end_elem->type == SB_CHAR) ? end_elem->opr.ch
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2692 : ((end_elem->type == COLL_SYM) ? end_elem->opr.name[0]
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2693 : 0));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2694 start_wc = ((start_elem->type == SB_CHAR || start_elem->type == COLL_SYM)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2695 ? __btowc (start_ch) : start_elem->opr.wch);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2696 end_wc = ((end_elem->type == SB_CHAR || end_elem->type == COLL_SYM)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2697 ? __btowc (end_ch) : end_elem->opr.wch);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2698 if (start_wc == WEOF || end_wc == WEOF)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2699 return REG_ECOLLATE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2700 cmp_buf[0] = start_wc;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2701 cmp_buf[4] = end_wc;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2702 if (wcscoll (cmp_buf, cmp_buf + 4) > 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2703 return REG_ERANGE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2704
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2705 /* Got valid collation sequence values, add them as a new entry.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2706 However, for !_LIBC we have no collation elements: if the
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2707 character set is single byte, the single byte character set
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2708 that we build below suffices. parse_bracket_exp passes
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2709 no MBCSET if dfa->mb_cur_max == 1. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2710 if (mbcset)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2711 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2712 /* Check the space of the arrays. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2713 if (BE (*range_alloc == mbcset->nranges, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2714 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2715 /* There is not enough space, need realloc. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2716 wchar_t *new_array_start, *new_array_end;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2717 int new_nranges;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2718
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2719 /* +1 in case of mbcset->nranges is 0. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2720 new_nranges = 2 * mbcset->nranges + 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2721 /* Use realloc since mbcset->range_starts and mbcset->range_ends
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2722 are NULL if *range_alloc == 0. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2723 new_array_start = re_realloc (mbcset->range_starts, wchar_t,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2724 new_nranges);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2725 new_array_end = re_realloc (mbcset->range_ends, wchar_t,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2726 new_nranges);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2727
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2728 if (BE (new_array_start == NULL || new_array_end == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2729 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2730
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2731 mbcset->range_starts = new_array_start;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2732 mbcset->range_ends = new_array_end;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2733 *range_alloc = new_nranges;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2734 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2735
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2736 mbcset->range_starts[mbcset->nranges] = start_wc;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2737 mbcset->range_ends[mbcset->nranges++] = end_wc;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2738 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2739
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2740 /* Build the table for single byte characters. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2741 for (wc = 0; wc < SBC_MAX; ++wc)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2742 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2743 cmp_buf[2] = wc;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2744 if (wcscoll (cmp_buf, cmp_buf + 2) <= 0
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2745 && wcscoll (cmp_buf + 2, cmp_buf + 4) <= 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2746 bitset_set (sbcset, wc);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2747 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2748 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2749 # else /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2750 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2751 unsigned int ch;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2752 start_ch = ((start_elem->type == SB_CHAR ) ? start_elem->opr.ch
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2753 : ((start_elem->type == COLL_SYM) ? start_elem->opr.name[0]
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2754 : 0));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2755 end_ch = ((end_elem->type == SB_CHAR ) ? end_elem->opr.ch
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2756 : ((end_elem->type == COLL_SYM) ? end_elem->opr.name[0]
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2757 : 0));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2758 if (start_ch > end_ch)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2759 return REG_ERANGE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2760 /* Build the table for single byte characters. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2761 for (ch = 0; ch < SBC_MAX; ++ch)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2762 if (start_ch <= ch && ch <= end_ch)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2763 bitset_set (sbcset, ch);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2764 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2765 # endif /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2766 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2767 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2768 #endif /* not _LIBC */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2769
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2770 #ifndef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2771 /* Helper function for parse_bracket_exp only used in case of NOT _LIBC..
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2772 Build the collating element which is represented by NAME.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2773 The result are written to MBCSET and SBCSET.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2774 COLL_SYM_ALLOC is the allocated size of mbcset->coll_sym, is a
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2775 pointer argument since we may update it. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2776
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2777 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2778 # ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2779 build_collating_symbol (sbcset, mbcset, coll_sym_alloc, name)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2780 re_charset_t *mbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2781 int *coll_sym_alloc;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2782 # else /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2783 build_collating_symbol (sbcset, name)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2784 # endif /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2785 re_bitset_ptr_t sbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2786 const unsigned char *name;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2787 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2788 size_t name_len = strlen ((const char *) name);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2789 if (BE (name_len != 1, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2790 return REG_ECOLLATE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2791 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2792 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2793 bitset_set (sbcset, name[0]);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2794 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2795 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2796 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2797 #endif /* not _LIBC */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2798
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2799 /* This function parse bracket expression like "[abc]", "[a-c]",
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2800 "[[.a-a.]]" etc. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2801
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2802 static bin_tree_t *
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2803 parse_bracket_exp (regexp, dfa, token, syntax, err)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2804 re_string_t *regexp;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2805 re_dfa_t *dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2806 re_token_t *token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2807 reg_syntax_t syntax;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2808 reg_errcode_t *err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2809 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2810 #ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2811 const unsigned char *collseqmb;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2812 const char *collseqwc;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2813 uint32_t nrules;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2814 int32_t table_size;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2815 const int32_t *symb_table;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2816 const unsigned char *extra;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2817
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2818 /* Local function for parse_bracket_exp used in _LIBC environement.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2819 Seek the collating symbol entry correspondings to NAME.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2820 Return the index of the symbol in the SYMB_TABLE. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2821
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2822 auto inline int32_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2823 __attribute ((always_inline))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2824 seek_collating_symbol_entry (name, name_len)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2825 const unsigned char *name;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2826 size_t name_len;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2827 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2828 int32_t hash = elem_hash ((const char *) name, name_len);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2829 int32_t elem = hash % table_size;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2830 int32_t second = hash % (table_size - 2);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2831 while (symb_table[2 * elem] != 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2832 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2833 /* First compare the hashing value. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2834 if (symb_table[2 * elem] == hash
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2835 /* Compare the length of the name. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2836 && name_len == extra[symb_table[2 * elem + 1]]
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2837 /* Compare the name. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2838 && memcmp (name, &extra[symb_table[2 * elem + 1] + 1],
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2839 name_len) == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2840 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2841 /* Yep, this is the entry. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2842 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2843 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2844
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2845 /* Next entry. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2846 elem += second;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2847 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2848 return elem;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2849 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2850
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2851 /* Local function for parse_bracket_exp used in _LIBC environement.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2852 Look up the collation sequence value of BR_ELEM.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2853 Return the value if succeeded, UINT_MAX otherwise. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2854
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2855 auto inline unsigned int
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2856 __attribute ((always_inline))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2857 lookup_collation_sequence_value (br_elem)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2858 bracket_elem_t *br_elem;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2859 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2860 if (br_elem->type == SB_CHAR)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2861 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2862 /*
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2863 if (MB_CUR_MAX == 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2864 */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2865 if (nrules == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2866 return collseqmb[br_elem->opr.ch];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2867 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2868 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2869 wint_t wc = __btowc (br_elem->opr.ch);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2870 return __collseq_table_lookup (collseqwc, wc);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2871 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2872 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2873 else if (br_elem->type == MB_CHAR)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2874 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2875 return __collseq_table_lookup (collseqwc, br_elem->opr.wch);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2876 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2877 else if (br_elem->type == COLL_SYM)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2878 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2879 size_t sym_name_len = strlen ((char *) br_elem->opr.name);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2880 if (nrules != 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2881 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2882 int32_t elem, idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2883 elem = seek_collating_symbol_entry (br_elem->opr.name,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2884 sym_name_len);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2885 if (symb_table[2 * elem] != 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2886 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2887 /* We found the entry. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2888 idx = symb_table[2 * elem + 1];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2889 /* Skip the name of collating element name. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2890 idx += 1 + extra[idx];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2891 /* Skip the byte sequence of the collating element. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2892 idx += 1 + extra[idx];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2893 /* Adjust for the alignment. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2894 idx = (idx + 3) & ~3;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2895 /* Skip the multibyte collation sequence value. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2896 idx += sizeof (unsigned int);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2897 /* Skip the wide char sequence of the collating element. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2898 idx += sizeof (unsigned int) *
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2899 (1 + *(unsigned int *) (extra + idx));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2900 /* Return the collation sequence value. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2901 return *(unsigned int *) (extra + idx);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2902 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2903 else if (symb_table[2 * elem] == 0 && sym_name_len == 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2904 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2905 /* No valid character. Match it as a single byte
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2906 character. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2907 return collseqmb[br_elem->opr.name[0]];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2908 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2909 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2910 else if (sym_name_len == 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2911 return collseqmb[br_elem->opr.name[0]];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2912 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2913 return UINT_MAX;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2914 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2915
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2916 /* Local function for parse_bracket_exp used in _LIBC environement.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2917 Build the range expression which starts from START_ELEM, and ends
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2918 at END_ELEM. The result are written to MBCSET and SBCSET.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2919 RANGE_ALLOC is the allocated size of mbcset->range_starts, and
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2920 mbcset->range_ends, is a pointer argument sinse we may
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2921 update it. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2922
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2923 auto inline reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2924 __attribute ((always_inline))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2925 build_range_exp (sbcset, mbcset, range_alloc, start_elem, end_elem)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2926 re_charset_t *mbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2927 int *range_alloc;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2928 re_bitset_ptr_t sbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2929 bracket_elem_t *start_elem, *end_elem;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2930 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2931 unsigned int ch;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2932 uint32_t start_collseq;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2933 uint32_t end_collseq;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2934
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2935 /* Equivalence Classes and Character Classes can't be a range
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2936 start/end. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2937 if (BE (start_elem->type == EQUIV_CLASS || start_elem->type == CHAR_CLASS
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2938 || end_elem->type == EQUIV_CLASS || end_elem->type == CHAR_CLASS,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2939 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2940 return REG_ERANGE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2941
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2942 start_collseq = lookup_collation_sequence_value (start_elem);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2943 end_collseq = lookup_collation_sequence_value (end_elem);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2944 /* Check start/end collation sequence values. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2945 if (BE (start_collseq == UINT_MAX || end_collseq == UINT_MAX, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2946 return REG_ECOLLATE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2947 if (BE ((syntax & RE_NO_EMPTY_RANGES) && start_collseq > end_collseq, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2948 return REG_ERANGE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2949
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2950 /* Got valid collation sequence values, add them as a new entry.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2951 However, if we have no collation elements, and the character set
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2952 is single byte, the single byte character set that we
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2953 build below suffices. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2954 if (nrules > 0 || dfa->mb_cur_max > 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2955 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2956 /* Check the space of the arrays. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2957 if (BE (*range_alloc == mbcset->nranges, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2958 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2959 /* There is not enough space, need realloc. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2960 uint32_t *new_array_start;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2961 uint32_t *new_array_end;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2962 int new_nranges;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2963
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2964 /* +1 in case of mbcset->nranges is 0. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2965 new_nranges = 2 * mbcset->nranges + 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2966 new_array_start = re_realloc (mbcset->range_starts, uint32_t,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2967 new_nranges);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2968 new_array_end = re_realloc (mbcset->range_ends, uint32_t,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2969 new_nranges);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2970
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2971 if (BE (new_array_start == NULL || new_array_end == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2972 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2973
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2974 mbcset->range_starts = new_array_start;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2975 mbcset->range_ends = new_array_end;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2976 *range_alloc = new_nranges;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2977 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2978
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2979 mbcset->range_starts[mbcset->nranges] = start_collseq;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2980 mbcset->range_ends[mbcset->nranges++] = end_collseq;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2981 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2982
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2983 /* Build the table for single byte characters. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2984 for (ch = 0; ch < SBC_MAX; ch++)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2985 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2986 uint32_t ch_collseq;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2987 /*
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2988 if (MB_CUR_MAX == 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2989 */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2990 if (nrules == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2991 ch_collseq = collseqmb[ch];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2992 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2993 ch_collseq = __collseq_table_lookup (collseqwc, __btowc (ch));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2994 if (start_collseq <= ch_collseq && ch_collseq <= end_collseq)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2995 bitset_set (sbcset, ch);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2996 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2997 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2998 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
2999
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3000 /* Local function for parse_bracket_exp used in _LIBC environement.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3001 Build the collating element which is represented by NAME.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3002 The result are written to MBCSET and SBCSET.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3003 COLL_SYM_ALLOC is the allocated size of mbcset->coll_sym, is a
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3004 pointer argument sinse we may update it. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3005
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3006 auto inline reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3007 __attribute ((always_inline))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3008 build_collating_symbol (sbcset, mbcset, coll_sym_alloc, name)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3009 re_charset_t *mbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3010 int *coll_sym_alloc;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3011 re_bitset_ptr_t sbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3012 const unsigned char *name;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3013 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3014 int32_t elem, idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3015 size_t name_len = strlen ((const char *) name);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3016 if (nrules != 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3017 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3018 elem = seek_collating_symbol_entry (name, name_len);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3019 if (symb_table[2 * elem] != 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3020 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3021 /* We found the entry. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3022 idx = symb_table[2 * elem + 1];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3023 /* Skip the name of collating element name. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3024 idx += 1 + extra[idx];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3025 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3026 else if (symb_table[2 * elem] == 0 && name_len == 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3027 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3028 /* No valid character, treat it as a normal
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3029 character. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3030 bitset_set (sbcset, name[0]);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3031 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3032 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3033 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3034 return REG_ECOLLATE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3035
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3036 /* Got valid collation sequence, add it as a new entry. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3037 /* Check the space of the arrays. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3038 if (BE (*coll_sym_alloc == mbcset->ncoll_syms, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3039 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3040 /* Not enough, realloc it. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3041 /* +1 in case of mbcset->ncoll_syms is 0. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3042 int new_coll_sym_alloc = 2 * mbcset->ncoll_syms + 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3043 /* Use realloc since mbcset->coll_syms is NULL
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3044 if *alloc == 0. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3045 int32_t *new_coll_syms = re_realloc (mbcset->coll_syms, int32_t,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3046 new_coll_sym_alloc);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3047 if (BE (new_coll_syms == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3048 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3049 mbcset->coll_syms = new_coll_syms;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3050 *coll_sym_alloc = new_coll_sym_alloc;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3051 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3052 mbcset->coll_syms[mbcset->ncoll_syms++] = idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3053 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3054 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3055 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3056 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3057 if (BE (name_len != 1, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3058 return REG_ECOLLATE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3059 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3060 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3061 bitset_set (sbcset, name[0]);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3062 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3063 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3064 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3065 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3066 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3067
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3068 re_token_t br_token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3069 re_bitset_ptr_t sbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3070 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3071 re_charset_t *mbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3072 int coll_sym_alloc = 0, range_alloc = 0, mbchar_alloc = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3073 int equiv_class_alloc = 0, char_class_alloc = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3074 #endif /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3075 int non_match = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3076 bin_tree_t *work_tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3077 int token_len;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3078 int first_round = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3079 #ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3080 collseqmb = (const unsigned char *)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3081 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_COLLSEQMB);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3082 nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3083 if (nrules)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3084 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3085 /*
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3086 if (MB_CUR_MAX > 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3087 */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3088 collseqwc = _NL_CURRENT (LC_COLLATE, _NL_COLLATE_COLLSEQWC);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3089 table_size = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_SYMB_HASH_SIZEMB);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3090 symb_table = (const int32_t *) _NL_CURRENT (LC_COLLATE,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3091 _NL_COLLATE_SYMB_TABLEMB);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3092 extra = (const unsigned char *) _NL_CURRENT (LC_COLLATE,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3093 _NL_COLLATE_SYMB_EXTRAMB);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3094 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3095 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3096 sbcset = (re_bitset_ptr_t) calloc (sizeof (unsigned int), BITSET_UINTS);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3097 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3098 mbcset = (re_charset_t *) calloc (sizeof (re_charset_t), 1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3099 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3100 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3101 if (BE (sbcset == NULL || mbcset == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3102 #else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3103 if (BE (sbcset == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3104 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3105 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3106 *err = REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3107 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3108 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3109
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3110 token_len = peek_token_bracket (token, regexp, syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3111 if (BE (token->type == END_OF_RE, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3112 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3113 *err = REG_BADPAT;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3114 goto parse_bracket_exp_free_return;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3115 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3116 if (token->type == OP_NON_MATCH_LIST)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3117 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3118 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3119 mbcset->non_match = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3120 #endif /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3121 non_match = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3122 if (syntax & RE_HAT_LISTS_NOT_NEWLINE)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3123 bitset_set (sbcset, '\0');
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3124 re_string_skip_bytes (regexp, token_len); /* Skip a token. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3125 token_len = peek_token_bracket (token, regexp, syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3126 if (BE (token->type == END_OF_RE, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3127 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3128 *err = REG_BADPAT;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3129 goto parse_bracket_exp_free_return;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3130 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3131 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3132
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3133 /* We treat the first ']' as a normal character. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3134 if (token->type == OP_CLOSE_BRACKET)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3135 token->type = CHARACTER;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3136
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3137 while (1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3138 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3139 bracket_elem_t start_elem, end_elem;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3140 unsigned char start_name_buf[BRACKET_NAME_BUF_SIZE];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3141 unsigned char end_name_buf[BRACKET_NAME_BUF_SIZE];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3142 reg_errcode_t ret;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3143 int token_len2 = 0, is_range_exp = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3144 re_token_t token2;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3145
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3146 start_elem.opr.name = start_name_buf;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3147 ret = parse_bracket_element (&start_elem, regexp, token, token_len, dfa,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3148 syntax, first_round);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3149 if (BE (ret != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3150 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3151 *err = ret;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3152 goto parse_bracket_exp_free_return;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3153 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3154 first_round = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3155
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3156 /* Get information about the next token. We need it in any case. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3157 token_len = peek_token_bracket (token, regexp, syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3158
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3159 /* Do not check for ranges if we know they are not allowed. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3160 if (start_elem.type != CHAR_CLASS && start_elem.type != EQUIV_CLASS)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3161 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3162 if (BE (token->type == END_OF_RE, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3163 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3164 *err = REG_EBRACK;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3165 goto parse_bracket_exp_free_return;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3166 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3167 if (token->type == OP_CHARSET_RANGE)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3168 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3169 re_string_skip_bytes (regexp, token_len); /* Skip '-'. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3170 token_len2 = peek_token_bracket (&token2, regexp, syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3171 if (BE (token2.type == END_OF_RE, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3172 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3173 *err = REG_EBRACK;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3174 goto parse_bracket_exp_free_return;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3175 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3176 if (token2.type == OP_CLOSE_BRACKET)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3177 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3178 /* We treat the last '-' as a normal character. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3179 re_string_skip_bytes (regexp, -token_len);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3180 token->type = CHARACTER;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3181 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3182 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3183 is_range_exp = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3184 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3185 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3186
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3187 if (is_range_exp == 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3188 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3189 end_elem.opr.name = end_name_buf;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3190 ret = parse_bracket_element (&end_elem, regexp, &token2, token_len2,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3191 dfa, syntax, 1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3192 if (BE (ret != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3193 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3194 *err = ret;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3195 goto parse_bracket_exp_free_return;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3196 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3197
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3198 token_len = peek_token_bracket (token, regexp, syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3199
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3200 #ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3201 *err = build_range_exp (sbcset, mbcset, &range_alloc,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3202 &start_elem, &end_elem);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3203 #else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3204 # ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3205 *err = build_range_exp (sbcset,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3206 dfa->mb_cur_max > 1 ? mbcset : NULL,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3207 &range_alloc, &start_elem, &end_elem);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3208 # else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3209 *err = build_range_exp (sbcset, &start_elem, &end_elem);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3210 # endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3211 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3212 if (BE (*err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3213 goto parse_bracket_exp_free_return;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3214 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3215 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3216 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3217 switch (start_elem.type)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3218 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3219 case SB_CHAR:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3220 bitset_set (sbcset, start_elem.opr.ch);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3221 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3222 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3223 case MB_CHAR:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3224 /* Check whether the array has enough space. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3225 if (BE (mbchar_alloc == mbcset->nmbchars, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3226 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3227 wchar_t *new_mbchars;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3228 /* Not enough, realloc it. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3229 /* +1 in case of mbcset->nmbchars is 0. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3230 mbchar_alloc = 2 * mbcset->nmbchars + 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3231 /* Use realloc since array is NULL if *alloc == 0. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3232 new_mbchars = re_realloc (mbcset->mbchars, wchar_t,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3233 mbchar_alloc);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3234 if (BE (new_mbchars == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3235 goto parse_bracket_exp_espace;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3236 mbcset->mbchars = new_mbchars;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3237 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3238 mbcset->mbchars[mbcset->nmbchars++] = start_elem.opr.wch;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3239 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3240 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3241 case EQUIV_CLASS:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3242 *err = build_equiv_class (sbcset,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3243 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3244 mbcset, &equiv_class_alloc,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3245 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3246 start_elem.opr.name);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3247 if (BE (*err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3248 goto parse_bracket_exp_free_return;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3249 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3250 case COLL_SYM:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3251 *err = build_collating_symbol (sbcset,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3252 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3253 mbcset, &coll_sym_alloc,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3254 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3255 start_elem.opr.name);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3256 if (BE (*err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3257 goto parse_bracket_exp_free_return;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3258 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3259 case CHAR_CLASS:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3260 *err = build_charclass (regexp->trans, sbcset,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3261 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3262 mbcset, &char_class_alloc,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3263 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3264 start_elem.opr.name, syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3265 if (BE (*err != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3266 goto parse_bracket_exp_free_return;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3267 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3268 default:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3269 assert (0);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3270 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3271 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3272 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3273 if (BE (token->type == END_OF_RE, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3274 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3275 *err = REG_EBRACK;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3276 goto parse_bracket_exp_free_return;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3277 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3278 if (token->type == OP_CLOSE_BRACKET)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3279 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3280 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3281
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3282 re_string_skip_bytes (regexp, token_len); /* Skip a token. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3283
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3284 /* If it is non-matching list. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3285 if (non_match)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3286 bitset_not (sbcset);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3287
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3288 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3289 /* Ensure only single byte characters are set. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3290 if (dfa->mb_cur_max > 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3291 bitset_mask (sbcset, dfa->sb_char);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3292
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3293 if (mbcset->nmbchars || mbcset->ncoll_syms || mbcset->nequiv_classes
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3294 || mbcset->nranges || (dfa->mb_cur_max > 1 && (mbcset->nchar_classes
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3295 || mbcset->non_match)))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3296 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3297 bin_tree_t *mbc_tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3298 int sbc_idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3299 /* Build a tree for complex bracket. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3300 dfa->has_mb_node = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3301 br_token.type = COMPLEX_BRACKET;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3302 br_token.opr.mbcset = mbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3303 mbc_tree = create_token_tree (dfa, NULL, NULL, &br_token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3304 if (BE (mbc_tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3305 goto parse_bracket_exp_espace;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3306 for (sbc_idx = 0; sbc_idx < BITSET_UINTS; ++sbc_idx)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3307 if (sbcset[sbc_idx])
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3308 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3309 /* If there are no bits set in sbcset, there is no point
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3310 of having both SIMPLE_BRACKET and COMPLEX_BRACKET. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3311 if (sbc_idx < BITSET_UINTS)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3312 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3313 /* Build a tree for simple bracket. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3314 br_token.type = SIMPLE_BRACKET;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3315 br_token.opr.sbcset = sbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3316 work_tree = create_token_tree (dfa, NULL, NULL, &br_token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3317 if (BE (work_tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3318 goto parse_bracket_exp_espace;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3319
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3320 /* Then join them by ALT node. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3321 work_tree = create_tree (dfa, work_tree, mbc_tree, OP_ALT);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3322 if (BE (work_tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3323 goto parse_bracket_exp_espace;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3324 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3325 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3326 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3327 re_free (sbcset);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3328 work_tree = mbc_tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3329 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3330 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3331 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3332 #endif /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3333 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3334 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3335 free_charset (mbcset);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3336 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3337 /* Build a tree for simple bracket. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3338 br_token.type = SIMPLE_BRACKET;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3339 br_token.opr.sbcset = sbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3340 work_tree = create_token_tree (dfa, NULL, NULL, &br_token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3341 if (BE (work_tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3342 goto parse_bracket_exp_espace;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3343 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3344 return work_tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3345
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3346 parse_bracket_exp_espace:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3347 *err = REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3348 parse_bracket_exp_free_return:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3349 re_free (sbcset);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3350 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3351 free_charset (mbcset);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3352 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3353 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3354 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3355
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3356 /* Parse an element in the bracket expression. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3357
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3358 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3359 parse_bracket_element (elem, regexp, token, token_len, dfa, syntax,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3360 accept_hyphen)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3361 bracket_elem_t *elem;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3362 re_string_t *regexp;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3363 re_token_t *token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3364 int token_len;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3365 re_dfa_t *dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3366 reg_syntax_t syntax;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3367 int accept_hyphen;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3368 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3369 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3370 int cur_char_size;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3371 cur_char_size = re_string_char_size_at (regexp, re_string_cur_idx (regexp));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3372 if (cur_char_size > 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3373 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3374 elem->type = MB_CHAR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3375 elem->opr.wch = re_string_wchar_at (regexp, re_string_cur_idx (regexp));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3376 re_string_skip_bytes (regexp, cur_char_size);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3377 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3378 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3379 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3380 re_string_skip_bytes (regexp, token_len); /* Skip a token. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3381 if (token->type == OP_OPEN_COLL_ELEM || token->type == OP_OPEN_CHAR_CLASS
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3382 || token->type == OP_OPEN_EQUIV_CLASS)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3383 return parse_bracket_symbol (elem, regexp, token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3384 if (BE (token->type == OP_CHARSET_RANGE, 0) && !accept_hyphen)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3385 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3386 /* A '-' must only appear as anything but a range indicator before
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3387 the closing bracket. Everything else is an error. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3388 re_token_t token2;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3389 (void) peek_token_bracket (&token2, regexp, syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3390 if (token2.type != OP_CLOSE_BRACKET)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3391 /* The actual error value is not standardized since this whole
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3392 case is undefined. But ERANGE makes good sense. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3393 return REG_ERANGE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3394 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3395 elem->type = SB_CHAR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3396 elem->opr.ch = token->opr.c;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3397 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3398 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3399
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3400 /* Parse a bracket symbol in the bracket expression. Bracket symbols are
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3401 such as [:<character_class>:], [.<collating_element>.], and
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3402 [=<equivalent_class>=]. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3403
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3404 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3405 parse_bracket_symbol (elem, regexp, token)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3406 bracket_elem_t *elem;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3407 re_string_t *regexp;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3408 re_token_t *token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3409 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3410 unsigned char ch, delim = token->opr.c;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3411 int i = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3412 if (re_string_eoi(regexp))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3413 return REG_EBRACK;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3414 for (;; ++i)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3415 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3416 if (i >= BRACKET_NAME_BUF_SIZE)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3417 return REG_EBRACK;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3418 if (token->type == OP_OPEN_CHAR_CLASS)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3419 ch = re_string_fetch_byte_case (regexp);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3420 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3421 ch = re_string_fetch_byte (regexp);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3422 if (re_string_eoi(regexp))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3423 return REG_EBRACK;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3424 if (ch == delim && re_string_peek_byte (regexp, 0) == ']')
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3425 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3426 elem->opr.name[i] = ch;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3427 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3428 re_string_skip_bytes (regexp, 1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3429 elem->opr.name[i] = '\0';
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3430 switch (token->type)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3431 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3432 case OP_OPEN_COLL_ELEM:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3433 elem->type = COLL_SYM;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3434 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3435 case OP_OPEN_EQUIV_CLASS:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3436 elem->type = EQUIV_CLASS;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3437 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3438 case OP_OPEN_CHAR_CLASS:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3439 elem->type = CHAR_CLASS;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3440 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3441 default:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3442 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3443 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3444 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3445 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3446
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3447 /* Helper function for parse_bracket_exp.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3448 Build the equivalence class which is represented by NAME.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3449 The result are written to MBCSET and SBCSET.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3450 EQUIV_CLASS_ALLOC is the allocated size of mbcset->equiv_classes,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3451 is a pointer argument sinse we may update it. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3452
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3453 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3454 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3455 build_equiv_class (sbcset, mbcset, equiv_class_alloc, name)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3456 re_charset_t *mbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3457 int *equiv_class_alloc;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3458 #else /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3459 build_equiv_class (sbcset, name)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3460 #endif /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3461 re_bitset_ptr_t sbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3462 const unsigned char *name;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3463 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3464 #if defined _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3465 uint32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3466 if (nrules != 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3467 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3468 const int32_t *table, *indirect;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3469 const unsigned char *weights, *extra, *cp;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3470 unsigned char char_buf[2];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3471 int32_t idx1, idx2;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3472 unsigned int ch;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3473 size_t len;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3474 /* This #include defines a local function! */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3475 # include <locale/weight.h>
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3476 /* Calculate the index for equivalence class. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3477 cp = name;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3478 table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3479 weights = (const unsigned char *) _NL_CURRENT (LC_COLLATE,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3480 _NL_COLLATE_WEIGHTMB);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3481 extra = (const unsigned char *) _NL_CURRENT (LC_COLLATE,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3482 _NL_COLLATE_EXTRAMB);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3483 indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3484 _NL_COLLATE_INDIRECTMB);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3485 idx1 = findidx (&cp);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3486 if (BE (idx1 == 0 || cp < name + strlen ((const char *) name), 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3487 /* This isn't a valid character. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3488 return REG_ECOLLATE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3489
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3490 /* Build single byte matcing table for this equivalence class. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3491 char_buf[1] = (unsigned char) '\0';
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3492 len = weights[idx1];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3493 for (ch = 0; ch < SBC_MAX; ++ch)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3494 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3495 char_buf[0] = ch;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3496 cp = char_buf;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3497 idx2 = findidx (&cp);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3498 /*
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3499 idx2 = table[ch];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3500 */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3501 if (idx2 == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3502 /* This isn't a valid character. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3503 continue;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3504 if (len == weights[idx2])
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3505 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3506 int cnt = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3507 while (cnt <= len &&
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3508 weights[idx1 + 1 + cnt] == weights[idx2 + 1 + cnt])
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3509 ++cnt;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3510
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3511 if (cnt > len)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3512 bitset_set (sbcset, ch);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3513 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3514 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3515 /* Check whether the array has enough space. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3516 if (BE (*equiv_class_alloc == mbcset->nequiv_classes, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3517 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3518 /* Not enough, realloc it. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3519 /* +1 in case of mbcset->nequiv_classes is 0. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3520 int new_equiv_class_alloc = 2 * mbcset->nequiv_classes + 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3521 /* Use realloc since the array is NULL if *alloc == 0. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3522 int32_t *new_equiv_classes = re_realloc (mbcset->equiv_classes,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3523 int32_t,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3524 new_equiv_class_alloc);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3525 if (BE (new_equiv_classes == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3526 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3527 mbcset->equiv_classes = new_equiv_classes;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3528 *equiv_class_alloc = new_equiv_class_alloc;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3529 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3530 mbcset->equiv_classes[mbcset->nequiv_classes++] = idx1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3531 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3532 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3533 #endif /* _LIBC */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3534 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3535 if (BE (strlen ((const char *) name) != 1, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3536 return REG_ECOLLATE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3537 bitset_set (sbcset, *name);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3538 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3539 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3540 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3541
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3542 /* Helper function for parse_bracket_exp.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3543 Build the character class which is represented by NAME.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3544 The result are written to MBCSET and SBCSET.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3545 CHAR_CLASS_ALLOC is the allocated size of mbcset->char_classes,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3546 is a pointer argument sinse we may update it. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3547
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3548 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3549 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3550 build_charclass (trans, sbcset, mbcset, char_class_alloc, class_name, syntax)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3551 re_charset_t *mbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3552 int *char_class_alloc;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3553 #else /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3554 build_charclass (trans, sbcset, class_name, syntax)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3555 #endif /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3556 unsigned RE_TRANSLATE_TYPE trans;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3557 re_bitset_ptr_t sbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3558 const unsigned char *class_name;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3559 reg_syntax_t syntax;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3560 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3561 int i;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3562 const char *name = (const char *) class_name;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3563
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3564 /* In case of REG_ICASE "upper" and "lower" match the both of
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3565 upper and lower cases. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3566 if ((syntax & RE_ICASE)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3567 && (strcmp (name, "upper") == 0 || strcmp (name, "lower") == 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3568 name = "alpha";
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3569
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3570 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3571 /* Check the space of the arrays. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3572 if (BE (*char_class_alloc == mbcset->nchar_classes, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3573 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3574 /* Not enough, realloc it. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3575 /* +1 in case of mbcset->nchar_classes is 0. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3576 int new_char_class_alloc = 2 * mbcset->nchar_classes + 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3577 /* Use realloc since array is NULL if *alloc == 0. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3578 wctype_t *new_char_classes = re_realloc (mbcset->char_classes, wctype_t,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3579 new_char_class_alloc);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3580 if (BE (new_char_classes == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3581 return REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3582 mbcset->char_classes = new_char_classes;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3583 *char_class_alloc = new_char_class_alloc;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3584 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3585 mbcset->char_classes[mbcset->nchar_classes++] = __wctype (name);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3586 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3587
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3588 #define BUILD_CHARCLASS_LOOP(ctype_func) \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3589 for (i = 0; i < SBC_MAX; ++i) \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3590 { \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3591 if (ctype_func (i)) \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3592 { \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3593 int ch = trans ? trans[i] : i; \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3594 bitset_set (sbcset, ch); \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3595 } \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3596 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3597
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3598 if (strcmp (name, "alnum") == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3599 BUILD_CHARCLASS_LOOP (isalnum)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3600 else if (strcmp (name, "cntrl") == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3601 BUILD_CHARCLASS_LOOP (iscntrl)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3602 else if (strcmp (name, "lower") == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3603 BUILD_CHARCLASS_LOOP (islower)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3604 else if (strcmp (name, "space") == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3605 BUILD_CHARCLASS_LOOP (isspace)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3606 else if (strcmp (name, "alpha") == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3607 BUILD_CHARCLASS_LOOP (isalpha)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3608 else if (strcmp (name, "digit") == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3609 BUILD_CHARCLASS_LOOP (isdigit)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3610 else if (strcmp (name, "print") == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3611 BUILD_CHARCLASS_LOOP (isprint)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3612 else if (strcmp (name, "upper") == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3613 BUILD_CHARCLASS_LOOP (isupper)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3614 else if (strcmp (name, "blank") == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3615 BUILD_CHARCLASS_LOOP (isblank)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3616 else if (strcmp (name, "graph") == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3617 BUILD_CHARCLASS_LOOP (isgraph)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3618 else if (strcmp (name, "punct") == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3619 BUILD_CHARCLASS_LOOP (ispunct)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3620 else if (strcmp (name, "xdigit") == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3621 BUILD_CHARCLASS_LOOP (isxdigit)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3622 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3623 return REG_ECTYPE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3624
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3625 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3626 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3627
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3628 static bin_tree_t *
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3629 build_charclass_op (dfa, trans, class_name, extra, non_match, err)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3630 re_dfa_t *dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3631 unsigned RE_TRANSLATE_TYPE trans;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3632 const unsigned char *class_name;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3633 const unsigned char *extra;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3634 int non_match;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3635 reg_errcode_t *err;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3636 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3637 re_bitset_ptr_t sbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3638 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3639 re_charset_t *mbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3640 int alloc = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3641 #endif /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3642 reg_errcode_t ret;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3643 re_token_t br_token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3644 bin_tree_t *tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3645
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3646 sbcset = (re_bitset_ptr_t) calloc (sizeof (unsigned int), BITSET_UINTS);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3647 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3648 mbcset = (re_charset_t *) calloc (sizeof (re_charset_t), 1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3649 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3650
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3651 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3652 if (BE (sbcset == NULL || mbcset == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3653 #else /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3654 if (BE (sbcset == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3655 #endif /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3656 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3657 *err = REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3658 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3659 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3660
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3661 if (non_match)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3662 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3663 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3664 /*
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3665 if (syntax & RE_HAT_LISTS_NOT_NEWLINE)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3666 bitset_set(cset->sbcset, '\0');
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3667 */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3668 mbcset->non_match = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3669 #endif /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3670 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3671
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3672 /* We don't care the syntax in this case. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3673 ret = build_charclass (trans, sbcset,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3674 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3675 mbcset, &alloc,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3676 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3677 class_name, 0);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3678
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3679 if (BE (ret != REG_NOERROR, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3680 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3681 re_free (sbcset);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3682 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3683 free_charset (mbcset);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3684 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3685 *err = ret;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3686 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3687 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3688 /* \w match '_' also. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3689 for (; *extra; extra++)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3690 bitset_set (sbcset, *extra);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3691
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3692 /* If it is non-matching list. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3693 if (non_match)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3694 bitset_not (sbcset);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3695
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3696 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3697 /* Ensure only single byte characters are set. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3698 if (dfa->mb_cur_max > 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3699 bitset_mask (sbcset, dfa->sb_char);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3700 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3701
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3702 /* Build a tree for simple bracket. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3703 br_token.type = SIMPLE_BRACKET;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3704 br_token.opr.sbcset = sbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3705 tree = create_token_tree (dfa, NULL, NULL, &br_token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3706 if (BE (tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3707 goto build_word_op_espace;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3708
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3709 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3710 if (dfa->mb_cur_max > 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3711 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3712 bin_tree_t *mbc_tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3713 /* Build a tree for complex bracket. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3714 br_token.type = COMPLEX_BRACKET;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3715 br_token.opr.mbcset = mbcset;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3716 dfa->has_mb_node = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3717 mbc_tree = create_token_tree (dfa, NULL, NULL, &br_token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3718 if (BE (mbc_tree == NULL, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3719 goto build_word_op_espace;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3720 /* Then join them by ALT node. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3721 tree = create_tree (dfa, tree, mbc_tree, OP_ALT);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3722 if (BE (mbc_tree != NULL, 1))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3723 return tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3724 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3725 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3726 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3727 free_charset (mbcset);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3728 return tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3729 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3730 #else /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3731 return tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3732 #endif /* not RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3733
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3734 build_word_op_espace:
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3735 re_free (sbcset);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3736 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3737 free_charset (mbcset);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3738 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3739 *err = REG_ESPACE;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3740 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3741 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3742
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3743 /* This is intended for the expressions like "a{1,3}".
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3744 Fetch a number from `input', and return the number.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3745 Return -1, if the number field is empty like "{,1}".
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3746 Return -2, If an error is occured. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3747
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3748 static int
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3749 fetch_number (input, token, syntax)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3750 re_string_t *input;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3751 re_token_t *token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3752 reg_syntax_t syntax;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3753 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3754 int num = -1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3755 unsigned char c;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3756 while (1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3757 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3758 fetch_token (token, input, syntax);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3759 c = token->opr.c;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3760 if (BE (token->type == END_OF_RE, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3761 return -2;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3762 if (token->type == OP_CLOSE_DUP_NUM || c == ',')
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3763 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3764 num = ((token->type != CHARACTER || c < '0' || '9' < c || num == -2)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3765 ? -2 : ((num == -1) ? c - '0' : num * 10 + c - '0'));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3766 num = (num > RE_DUP_MAX) ? -2 : num;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3767 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3768 return num;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3769 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3770
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3771 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3772 static void
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3773 free_charset (re_charset_t *cset)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3774 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3775 re_free (cset->mbchars);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3776 # ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3777 re_free (cset->coll_syms);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3778 re_free (cset->equiv_classes);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3779 re_free (cset->range_starts);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3780 re_free (cset->range_ends);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3781 # endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3782 re_free (cset->char_classes);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3783 re_free (cset);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3784 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3785 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3786
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3787 /* Functions for binary tree operation. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3788
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3789 /* Create a tree node. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3790
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3791 static bin_tree_t *
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3792 create_tree (dfa, left, right, type)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3793 re_dfa_t *dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3794 bin_tree_t *left;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3795 bin_tree_t *right;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3796 re_token_type_t type;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3797 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3798 re_token_t t;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3799 t.type = type;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3800 return create_token_tree (dfa, left, right, &t);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3801 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3802
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3803 static bin_tree_t *
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3804 create_token_tree (dfa, left, right, token)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3805 re_dfa_t *dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3806 bin_tree_t *left;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3807 bin_tree_t *right;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3808 const re_token_t *token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3809 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3810 bin_tree_t *tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3811 if (BE (dfa->str_tree_storage_idx == BIN_TREE_STORAGE_SIZE, 0))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3812 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3813 bin_tree_storage_t *storage = re_malloc (bin_tree_storage_t, 1);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3814
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3815 if (storage == NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3816 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3817 storage->next = dfa->str_tree_storage;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3818 dfa->str_tree_storage = storage;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3819 dfa->str_tree_storage_idx = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3820 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3821 tree = &dfa->str_tree_storage->data[dfa->str_tree_storage_idx++];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3822
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3823 tree->parent = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3824 tree->left = left;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3825 tree->right = right;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3826 tree->token = *token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3827 tree->token.duplicated = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3828 tree->token.opt_subexp = 0;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3829 tree->first = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3830 tree->next = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3831 tree->node_idx = -1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3832
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3833 if (left != NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3834 left->parent = tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3835 if (right != NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3836 right->parent = tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3837 return tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3838 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3839
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3840 /* Mark the tree SRC as an optional subexpression.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3841 To be called from preorder or postorder. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3842
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3843 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3844 mark_opt_subexp (extra, node)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3845 void *extra;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3846 bin_tree_t *node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3847 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3848 int idx = (int) (long) extra;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3849 if (node->token.type == SUBEXP && node->token.opr.idx == idx)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3850 node->token.opt_subexp = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3851
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3852 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3853 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3854
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3855 /* Free the allocated memory inside NODE. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3856
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3857 static void
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3858 free_token (re_token_t *node)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3859 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3860 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3861 if (node->type == COMPLEX_BRACKET && node->duplicated == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3862 free_charset (node->opr.mbcset);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3863 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3864 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3865 if (node->type == SIMPLE_BRACKET && node->duplicated == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3866 re_free (node->opr.sbcset);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3867 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3868
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3869 /* Worker function for tree walking. Free the allocated memory inside NODE
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3870 and its children. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3871
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3872 static reg_errcode_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3873 free_tree (void *extra, bin_tree_t *node)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3874 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3875 free_token (&node->token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3876 return REG_NOERROR;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3877 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3878
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3879
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3880 /* Duplicate the node SRC, and return new node. This is a preorder
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3881 visit similar to the one implemented by the generic visitor, but
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3882 we need more infrastructure to maintain two parallel trees --- so,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3883 it's easier to duplicate. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3884
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3885 static bin_tree_t *
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3886 duplicate_tree (root, dfa)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3887 const bin_tree_t *root;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3888 re_dfa_t *dfa;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3889 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3890 const bin_tree_t *node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3891 bin_tree_t *dup_root;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3892 bin_tree_t **p_new = &dup_root, *dup_node = root->parent;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3893
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3894 for (node = root; ; )
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3895 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3896 /* Create a new tree and link it back to the current parent. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3897 *p_new = create_token_tree (dfa, NULL, NULL, &node->token);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3898 if (*p_new == NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3899 return NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3900 (*p_new)->parent = dup_node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3901 (*p_new)->token.duplicated = 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3902 dup_node = *p_new;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3903
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3904 /* Go to the left node, or up and to the right. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3905 if (node->left)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3906 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3907 node = node->left;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3908 p_new = &dup_node->left;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3909 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3910 else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3911 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3912 const bin_tree_t *prev = NULL;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3913 while (node->right == prev || node->right == NULL)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3914 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3915 prev = node;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3916 node = node->parent;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3917 dup_node = dup_node->parent;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3918 if (!node)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3919 return dup_root;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3920 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3921 node = node->right;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3922 p_new = &dup_node->right;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3923 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3924 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3925 }