Mercurial > octave-nkf

annotate liboctave/regexp.cc @ 14025:9867be070ee1

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
use pcre/pcre.h if it is present * configure.ac: Check for pcre/pcre.h. * liboctave/regexp.cc: Include either pcre.h or pcre/pcre.h depending on which is available.
author John W. Eaton <jwe@octave.org>
date Sun, 11 Dec 2011 22:35:13 -0500
parents fc9f204faea0
children 7e60182e6a1c
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
1 /*
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
2
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
3 Copyright (C) 2011 John W. Eaton
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
4 Copyright (C) 2005-2011 David Bateman
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
5 Copyright (C) 2002-2005 Paul Kienzle
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
6
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
7 This file is part of Octave.
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
8
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
9 Octave is free software; you can redistribute it and/or modify it
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
10 under the terms of the GNU General Public License as published by the
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
11 Free Software Foundation; either version 3 of the License, or (at your
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
12 option) any later version.
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
13
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
14 Octave is distributed in the hope that it will be useful, but WITHOUT
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
15 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
16 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
17 for more details.
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
18
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
19 You should have received a copy of the GNU General Public License
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
20 along with Octave; see the file COPYING. If not, see
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
21 <http://www.gnu.org/licenses/>.
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
22
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
23 */
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
24
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
25 #ifdef HAVE_CONFIG_H
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
26 #include <config.h>
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
27 #endif
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
28
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
29 #include <list>
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
30 #include <sstream>
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
31 #include <string>
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
32 #include <vector>
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
33
14025
9867be070ee1 use pcre/pcre.h if it is present
John W. Eaton <jwe@octave.org>
parents: 14024
diff changeset
34 #if defined (HAVE_PCRE_H)
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
35 #include <pcre.h>
14025
9867be070ee1 use pcre/pcre.h if it is present
John W. Eaton <jwe@octave.org>
parents: 14024
diff changeset
36 #elif defined (HAVE_PCRE_PCRE_H)
9867be070ee1 use pcre/pcre.h if it is present
John W. Eaton <jwe@octave.org>
parents: 14024
diff changeset
37 #include <pcre/pcre.h>
9867be070ee1 use pcre/pcre.h if it is present
John W. Eaton <jwe@octave.org>
parents: 14024
diff changeset
38 #endif
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
39
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
40 #include "Matrix.h"
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
41 #include "base-list.h"
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
42 #include "lo-error.h"
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
43 #include "oct-locbuf.h"
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
44 #include "quit.h"
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
45 #include "regexp.h"
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
46 #include "str-vec.h"
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
47
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
48 // Define the maximum number of retries for a pattern that possibly
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
49 // results in an infinite recursion.
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
50 #define PCRE_MATCHLIMIT_MAX 10
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
51
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
52 // FIXME -- should this be configurable?
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
53 #define MAXLOOKBEHIND 10
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
54
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
55 static bool lookbehind_warned = false;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
56
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
57 // FIXME -- don't bother collecting and composing return values the user
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
58 // doesn't want.
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
59
11586
12df7854fa7c strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents: 11570
diff changeset
60 void
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
61 regexp::free (void)
11586
12df7854fa7c strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents: 11570
diff changeset
62 {
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
63 if (data)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
64 pcre_free (static_cast<pcre *> (data));
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
65 }
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
66
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
67 void
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
68 regexp::compile_internal (void)
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
69 {
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
70 // If we had a previously compiled pattern, release it.
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
71 free ();
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
72
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
73 size_t max_length = MAXLOOKBEHIND;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
74
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
75 size_t pos = 0;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
76 size_t new_pos;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
77 int inames = 0;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
78 std::ostringstream buf;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
79
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
80 while ((new_pos = pattern.find ("(?", pos)) != std::string::npos)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
81 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
82 if (pattern.at (new_pos + 2) == '<'
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
83 && !(pattern.at (new_pos + 3) == '='
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
84 || pattern.at (new_pos + 3) == '!'))
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
85 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
86 // The syntax of named tokens in pcre is "(?P<name>...)" while
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
87 // we need a syntax "(?<name>...)", so fix that here. Also an
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
88 // expression like
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
89 // "(?<first>\w+)\s+(?<last>\w+)|(?<last>\w+),\s+(?<first>\w+)"
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
90 // should be perfectly legal, while pcre does not allow the same
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
91 // named token name on both sides of the alternative. Also fix
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
92 // that here by replacing name tokens by dummy names, and dealing
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
93 // with the dummy names later.
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
94
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
95 size_t tmp_pos = pattern.find_first_of ('>', new_pos);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
96
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
97 if (tmp_pos == std::string::npos)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
98 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
99 (*current_liboctave_error_handler)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
100 ("regexp: syntax error in pattern");
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
101 return;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
102 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
103
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
104 std::string tmp_name =
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
105 pattern.substr (new_pos+3, tmp_pos-new_pos-3);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
106
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
107 bool found = false;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
108
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
109 for (int i = 0; i < nnames; i++)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
110 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
111 if (named_pats(i) == tmp_name)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
112 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
113 named_idx.resize (dim_vector (inames+1, 1));
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
114 named_idx(inames) = i;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
115 found = true;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
116 break;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
117 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
118 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
119
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
120 if (! found)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
121 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
122 named_idx.resize (dim_vector (inames+1, 1));
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
123 named_idx(inames) = nnames;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
124 named_pats.append (tmp_name);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
125 nnames++;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
126 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
127
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
128 if (new_pos - pos > 0)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
129 buf << pattern.substr (pos, new_pos-pos);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
130 if (inames < 10)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
131 buf << "(?P<n00" << inames++;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
132 else if (inames < 100)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
133 buf << "(?P<n0" << inames++;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
134 else
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
135 buf << "(?P<n" << inames++;
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
136
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
137 pos = tmp_pos;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
138 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
139 else if (pattern.at (new_pos + 2) == '<')
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
140 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
141 // Find lookbehind operators of arbitrary length (ie like
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
142 // "(?<=[a-z]*)") and replace with a maximum length operator
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
143 // as PCRE can not yet handle arbitrary length lookahead
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
144 // operators. Use the string length as the maximum length to
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
145 // avoid issues.
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
146
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
147 int brackets = 1;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
148 size_t tmp_pos1 = new_pos + 2;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
149 size_t tmp_pos2 = tmp_pos1;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
150
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
151 while (tmp_pos1 <= pattern.length () && brackets > 0)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
152 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
153 char ch = pattern.at (tmp_pos1);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
154
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
155 if (ch == '(')
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
156 brackets++;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
157 else if (ch == ')')
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
158 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
159 if (brackets > 1)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
160 tmp_pos2 = tmp_pos1;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
161
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
162 brackets--;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
163 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
164
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
165 tmp_pos1++;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
166 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
167
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
168 if (brackets != 0)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
169 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
170 buf << pattern.substr (pos, new_pos - pos) << "(?";
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
171 pos = new_pos + 2;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
172 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
173 else
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
174 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
175 size_t tmp_pos3 = pattern.find_first_of ("*+", tmp_pos2);
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
176
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
177 if (tmp_pos3 != std::string::npos && tmp_pos3 < tmp_pos1)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
178 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
179 if (!lookbehind_warned)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
180 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
181 lookbehind_warned = true;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
182 (*current_liboctave_warning_handler)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
183 ("%s: arbitrary length lookbehind patterns are only supported up to length %d",
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
184 who.c_str (), MAXLOOKBEHIND);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
185 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
186
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
187 buf << pattern.substr (pos, new_pos - pos) << "(";
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
188
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
189 size_t i;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
190
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
191 if (pattern.at (tmp_pos3) == '*')
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
192 i = 0;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
193 else
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
194 i = 1;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
195
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
196 for (; i < max_length + 1; i++)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
197 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
198 buf << pattern.substr (new_pos, tmp_pos3 - new_pos)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
199 << "{" << i << "}";
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
200 buf << pattern.substr (tmp_pos3 + 1,
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
201 tmp_pos1 - tmp_pos3 - 1);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
202 if (i != max_length)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
203 buf << "|";
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
204 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
205 buf << ")";
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
206 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
207 else
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
208 buf << pattern.substr (pos, tmp_pos1 - pos);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
209
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
210 pos = tmp_pos1;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
211 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
212 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
213 else
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
214 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
215 buf << pattern.substr (pos, new_pos - pos) << "(?";
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
216 pos = new_pos + 2;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
217 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
218
7784
d3a7882fa0b3 style fixes
John W. Eaton <jwe@octave.org>
parents: 7779
diff changeset
219 }
11586
12df7854fa7c strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents: 11570
diff changeset
220
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
221 buf << pattern.substr (pos);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
222
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
223 const char *err;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
224 int erroffset;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
225 std::string buf_str = buf.str ();
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
226
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
227 int pcre_options
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
228 = ((options.case_insensitive () ? PCRE_CASELESS : 0)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
229 | (options.dotexceptnewline () ? 0 : PCRE_DOTALL)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
230 | (options.lineanchors () ? PCRE_MULTILINE : 0)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
231 | (options.freespacing () ? PCRE_EXTENDED : 0));
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
232
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
233 data = pcre_compile (buf_str.c_str (), pcre_options, &err, &erroffset, 0);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
234
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
235 if (! data)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
236 (*current_liboctave_error_handler)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
237 ("%s: %s at position %d of expression", who.c_str (),
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
238 err, erroffset);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
239 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
240
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
241 regexp::match_data
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
242 regexp::match (const std::string& buffer)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
243 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
244 regexp::match_data retval;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
245
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
246 std::list<regexp::match_element> lst;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
247
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
248 int subpatterns;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
249 int namecount;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
250 int nameentrysize;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
251 char *nametable;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
252 size_t idx = 0;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
253
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
254 pcre *re = static_cast <pcre *> (data);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
255
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
256 pcre_fullinfo (re, 0, PCRE_INFO_CAPTURECOUNT, &subpatterns);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
257 pcre_fullinfo (re, 0, PCRE_INFO_NAMECOUNT, &namecount);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
258 pcre_fullinfo (re, 0, PCRE_INFO_NAMEENTRYSIZE, &nameentrysize);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
259 pcre_fullinfo (re, 0, PCRE_INFO_NAMETABLE, &nametable);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
260
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
261 OCTAVE_LOCAL_BUFFER (int, ovector, (subpatterns+1)*3);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
262 OCTAVE_LOCAL_BUFFER (int, nidx, namecount);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
263
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
264 for (int i = 0; i < namecount; i++)
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
265 {
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
266 // Index of subpattern in first two bytes MSB first of name.
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
267 // Extract index.
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
268 nidx[i] = (static_cast<int> (nametable[i*nameentrysize])) << 8
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
269 | static_cast<int> (nametable[i*nameentrysize+1]);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
270 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
271
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
272 while (true)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
273 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
274 OCTAVE_QUIT;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
275
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
276 int matches = pcre_exec (re, 0, buffer.c_str (),
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
277 buffer.length (), idx,
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
278 (idx ? PCRE_NOTBOL : 0),
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
279 ovector, (subpatterns+1)*3);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
280
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
281 if (matches == PCRE_ERROR_MATCHLIMIT)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
282 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
283 // Try harder; start with default value for MATCH_LIMIT
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
284 // and increase it.
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
285 (*current_liboctave_warning_handler)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
286 ("your pattern caused PCRE to hit its MATCH_LIMIT; trying harder now, but this will be slow");
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
287
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
288 pcre_extra pe;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
289
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
290 pcre_config (PCRE_CONFIG_MATCH_LIMIT,
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
291 static_cast <void *> (&pe.match_limit));
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
292
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
293 pe.flags = PCRE_EXTRA_MATCH_LIMIT;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
294
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
295 int i = 0;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
296 while (matches == PCRE_ERROR_MATCHLIMIT
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
297 && i++ < PCRE_MATCHLIMIT_MAX)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
298 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
299 OCTAVE_QUIT;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
300
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
301 pe.match_limit *= 10;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
302 matches = pcre_exec (re, &pe, buffer.c_str (),
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
303 buffer.length (), idx,
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
304 (idx ? PCRE_NOTBOL : 0),
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
305 ovector, (subpatterns+1)*3);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
306 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
307 }
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
308
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
309 if (matches < 0 && matches != PCRE_ERROR_NOMATCH)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
310 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
311 (*current_liboctave_error_handler)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
312 ("%s: internal error calling pcre_exec; error code from pcre_exec is %i",
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
313 who.c_str (), matches);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
314 return retval;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
315 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
316 else if (matches == PCRE_ERROR_NOMATCH)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
317 break;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
318 else if (ovector[1] <= ovector[0])
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
319 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
320 // Zero sized match. Skip to next char.
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
321 idx = ovector[0] + 1;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
322 if (idx < buffer.length ())
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
323 continue;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
324 else
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
325 break;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
326 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
327 else
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
328 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
329 int pos_match = 0;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
330 Matrix token_extents (matches-1, 2);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
331
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
332 for (int i = 1; i < matches; i++)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
333 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
334 if (ovector[2*i] >= 0 && ovector[2*i+1] > 0
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
335 && (i == 1 || ovector[2*i] != ovector[2*i-2]
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
336 || ovector[2*i-1] != ovector[2*i+1])
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
337 && ovector[2*i] >= 0 && ovector[2*i+1] > 0)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
338 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
339 token_extents(pos_match,0) = double (ovector[2*i]+1);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
340 token_extents(pos_match++,1) = double (ovector[2*i+1]);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
341 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
342 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
343
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
344 token_extents.resize (pos_match, 2);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
345
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
346 double start = double (ovector[0]+1);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
347 double end = double (ovector[1]);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
348
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
349 const char **listptr;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
350 int status = pcre_get_substring_list (buffer.c_str (), ovector,
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
351 matches, &listptr);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
352
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
353 if (status == PCRE_ERROR_NOMEMORY)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
354 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
355 (*current_liboctave_error_handler)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
356 ("%s: cannot allocate memory in pcre_get_substring_list",
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
357 who.c_str ());
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
358 return retval;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
359 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
360
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
361 string_vector tokens (pos_match);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
362 string_vector named_tokens (nnames);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
363 int pos_offset = 0;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
364 pos_match = 0;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
365
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
366 for (int i = 1; i < matches; i++)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
367 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
368 if (ovector[2*i] >= 0 && ovector[2*i+1] > 0)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
369 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
370 if (i == 1 || ovector[2*i] != ovector[2*i-2]
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
371 || ovector[2*i-1] != ovector[2*i+1])
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
372 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
373 if (namecount > 0)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
374 named_tokens(named_idx(i-pos_offset-1)) =
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
375 std::string (*(listptr+nidx[i-pos_offset-1]));
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
376
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
377 tokens(pos_match++) = std::string (*(listptr+i));
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
378 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
379 else
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
380 pos_offset++;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
381 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
382 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
383
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
384 std::string match_string = std::string (*listptr);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
385
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
386 pcre_free_substring_list (listptr);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
387
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
388 regexp::match_element new_elem (named_tokens, tokens, match_string,
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
389 token_extents, start, end);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
390 lst.push_back (new_elem);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
391 idx = ovector[1];
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
392
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
393 if (options.once () || idx >= buffer.length ())
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
394 break;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
395 }
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
396 }
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
397
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
398 retval = regexp::match_data (lst, named_pats);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
399
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
400 return retval;
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
401 }
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
402
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
403 bool
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
404 regexp::is_match (const std::string& buffer)
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
405 {
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
406 regexp::match_data rx_lst = match (buffer);
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
407
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
408 regexp::match_data::const_iterator p = rx_lst.begin ();
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
409
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
410 std::string match_string = p->match_string ();
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
411
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
412 return ! match_string.empty ();
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
413 }
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
414
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
415 Array<bool>
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
416 regexp::is_match (const string_vector& buffer)
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
417 {
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
418 octave_idx_type len = buffer.length ();
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
419
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
420 Array<bool> retval (len, 1);
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
421
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
422 for (octave_idx_type i = 0; i < buffer.length (); i++)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
423 retval(i) = is_match (buffer(i));
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
424
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
425 return retval;
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
426 }
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
427
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
428 std::string
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
429 regexp::replace (const std::string& buffer, const std::string& replacement)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
430 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
431 std::string retval;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
432
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
433 // Identify replacement tokens; build a vector of group numbers in
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
434 // the replacement string so that we can quickly calculate the size
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
435 // of the replacement.
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
436
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
437 int tokens = 0;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
438 for (size_t i=1; i < replacement.size (); i++)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
439 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
440 if (replacement[i-1]=='$' && isdigit (replacement[i]))
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
441 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
442 tokens++;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
443 i++;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
444 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
445 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
446 std::vector<int> token (tokens);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
447
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
448 int kk = 0;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
449 for (size_t i = 1; i < replacement.size (); i++)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
450 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
451 if (replacement[i-1]=='$' && isdigit (replacement[i]))
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
452 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
453 token[kk++] = replacement[i]-'0';
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
454 i++;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
455 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
456 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
457
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
458 regexp::match_data rx_lst = match (buffer);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
459
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
460 size_t sz = rx_lst.size ();
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
461
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
462 if (sz == 0)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
463 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
464 retval = buffer;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
465 return retval;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
466 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
467
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
468 std::string rep;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
469
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
470 if (tokens > 0)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
471 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
472 // Determine replacement length
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
473 const size_t replen = replacement.size () - 2*tokens;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
474 int delta = 0;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
475 regexp::match_data::const_iterator p = rx_lst.begin ();
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
476 for (size_t i = 0; i < sz; i++)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
477 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
478 OCTAVE_QUIT;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
479
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
480 double start = p->start ();
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
481 double end = p->end ();
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
482
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
483 const Matrix pairs (p->token_extents ());
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
484 size_t pairlen = 0;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
485 for (int j = 0; j < tokens; j++)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
486 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
487 if (token[j] == 0)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
488 pairlen += static_cast<size_t> (end - start) + 1;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
489 else if (token[j] <= pairs.rows ())
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
490 pairlen += static_cast<size_t> (pairs(token[j]-1,1)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
491 - pairs(token[j]-1,0)) + 1;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
492 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
493 delta += (static_cast<int> (replen + pairlen)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
494 - static_cast<int> (end - start + 1));
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
495 p++;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
496 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
497
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
498 // Build replacement string
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
499 rep.reserve (buffer.size () + delta);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
500 size_t from = 0;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
501 p = rx_lst.begin ();
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
502 for (size_t i = 0; i < sz; i++)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
503 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
504 OCTAVE_QUIT;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
505
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
506 double start = p->start ();
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
507 double end = p->end ();
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
508
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
509 const Matrix pairs (p->token_extents ());
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
510 rep.append (&buffer[from], static_cast<size_t> (start - 1) - from);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
511 from = static_cast<size_t> (end - 1) + 1;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
512
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
513 for (size_t j = 1; j < replacement.size (); j++)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
514 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
515 if (replacement[j-1]=='$' && isdigit (replacement[j]))
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
516 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
517 int k = replacement[j]-'0';
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
518 if (k == 0)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
519 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
520 // replace with entire match
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
521 rep.append (&buffer[static_cast<size_t> (end - 1)],
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
522 static_cast<size_t> (end - start) + 1);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
523 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
524 else if (k <= pairs.rows ())
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
525 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
526 // replace with group capture
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
527 rep.append (&buffer[static_cast<size_t> (pairs(k-1,0)-1)],
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
528 static_cast<size_t> (pairs(k-1,1)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
529 - pairs(k-1,0)) + 1);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
530 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
531 else
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
532 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
533 // replace with nothing
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
534 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
535 j++;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
536 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
537 else
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
538 rep.append (1, replacement[j-1]);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
539
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
540 if (j+1 == replacement.size ())
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
541 rep.append (1, replacement[j]);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
542 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
543 p++;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
544 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
545 rep.append (&buffer[from], buffer.size () - from);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
546 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
547 else
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
548 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
549 // Determine replacement length
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
550 const size_t replen = replacement.size ();
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
551 int delta = 0;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
552 regexp::match_data::const_iterator p = rx_lst.begin ();
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
553 for (size_t i = 0; i < sz; i++)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
554 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
555 OCTAVE_QUIT;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
556 delta += static_cast<int> (replen)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
557 - static_cast<int> (p->end () - p->start () + 1);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
558 p++;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
559 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
560
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
561 // Build replacement string
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
562 rep.reserve (buffer.size () + delta);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
563 size_t from = 0;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
564 p = rx_lst.begin ();
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
565 for (size_t i = 0; i < sz; i++)
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
566 {
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
567 OCTAVE_QUIT;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
568 rep.append (&buffer[from],
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
569 static_cast<size_t> (p->start () - 1) - from);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
570 from = static_cast<size_t> (p->end () - 1) + 1;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
571 rep.append (replacement);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
572 p++;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
573 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
574 rep.append (&buffer[from], buffer.size () - from);
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
575 }
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
576
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
577 retval = rep;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
578 return retval;
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
579 }