Mercurial > octave

annotate liboctave/util/lo-regexp.cc @ 32091:d313e8ece0b8 stable

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
lo-regexp.cc: Only include PCRE2/PCRE headers if feature test succeeded (bug #64172). * liboctave/util/lo-regexp.cc: Condition inclusion of PCRE2/PCRE headers on whether the configure test for the respective libraries succeeded.
author Markus Mützel <markus.muetzel@gmx.de>
date Sat, 13 May 2023 13:57:14 +0200
parents 597f3ee61a48
children 0fe37636be04
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
27923
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
1 ////////////////////////////////////////////////////////////////////////
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
2 //
31706
597f3ee61a48 update Octave Project Developers copyright for the new year
John W. Eaton <jwe@octave.org>
parents: 31607
diff changeset
3 // Copyright (C) 2002-2023 The Octave Project Developers
27923
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
4 //
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
5 // See the file COPYRIGHT.md in the top-level directory of this
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
6 // distribution or <https://octave.org/copyright/>.
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
7 //
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
8 // This file is part of Octave.
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
9 //
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
10 // Octave is free software: you can redistribute it and/or modify it
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
11 // under the terms of the GNU General Public License as published by
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
12 // the Free Software Foundation, either version 3 of the License, or
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
13 // (at your option) any later version.
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
14 //
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
15 // Octave is distributed in the hope that it will be useful, but
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
16 // WITHOUT ANY WARRANTY; without even the implied warranty of
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
17 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
18 // GNU General Public License for more details.
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
19 //
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
20 // You should have received a copy of the GNU General Public License
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
21 // along with Octave; see the file COPYING. If not, see
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
22 // <https://www.gnu.org/licenses/>.
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
23 //
bd51beb6205e update formatting of copyright notices
John W. Eaton <jwe@octave.org>
parents: 27919
diff changeset
24 ////////////////////////////////////////////////////////////////////////
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
25
21724
aba2e6293dd8 use "#if ..." consistently instead of "#ifdef" and "#ifndef"
John W. Eaton <jwe@octave.org>
parents: 21301
diff changeset
26 #if defined (HAVE_CONFIG_H)
21301
40de9f8f23a6 Use '#include "config.h"' rather than <config.h>.
Rik <rik@octave.org>
parents: 21202
diff changeset
27 # include "config.h"
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
28 #endif
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
29
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
30 #include <list>
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
31 #include <sstream>
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
32 #include <string>
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
33 #include <vector>
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
34
32091
d313e8ece0b8 lo-regexp.cc: Only include PCRE2/PCRE headers if feature test succeeded (bug #64172).
Markus Mützel <markus.muetzel@gmx.de>
parents: 31706
diff changeset
35 #if defined (HAVE_PCRE2)
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
36 # define PCRE2_CODE_UNIT_WIDTH 8
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
37 # if defined (HAVE_PCRE2_H)
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
38 # include <pcre2.h>
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
39 # elif defined (HAVE_PCRE2_PCRE2_H)
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
40 # include <pcre2/pcre2.h>
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
41 # endif
32091
d313e8ece0b8 lo-regexp.cc: Only include PCRE2/PCRE headers if feature test succeeded (bug #64172).
Markus Mützel <markus.muetzel@gmx.de>
parents: 31706
diff changeset
42 #elif defined (HAVE_PCRE)
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
43 # if defined (HAVE_PCRE_H)
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
44 # include <pcre.h>
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
45 # elif defined (HAVE_PCRE_PCRE_H)
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
46 # include <pcre/pcre.h>
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
47 # endif
14025
9867be070ee1 use pcre/pcre.h if it is present
John W. Eaton <jwe@octave.org>
parents: 14024
diff changeset
48 #endif
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
49
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
50 #include "Matrix.h"
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
51 #include "base-list.h"
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
52 #include "lo-error.h"
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
53 #include "oct-locbuf.h"
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
54 #include "quit.h"
16940
99122191d3dd maint: Rename regexp.h to lo-regexp.h, regexp.cc to lo-regexp.cc in liboctave.
Rik <rik@octave.org>
parents: 15819
diff changeset
55 #include "lo-regexp.h"
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
56 #include "str-vec.h"
27523
19ad9150dd69 Check if input to regexp is valid UTF-8 is faster than PCRE's checks (bug #57064).
Markus Mützel <markus.muetzel@gmx.de>
parents: 27379
diff changeset
57 #include "unistr-wrappers.h"
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
58 #include "unwind-prot.h"
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
59
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
60 #if defined (HAVE_PCRE2)
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
61 typedef pcre2_code octave_pcre_code;
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
62 typedef PCRE2_SIZE OCTAVE_PCRE_SIZE;
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
63 void (*octave_pcre_code_free) (octave_pcre_code *) = pcre2_code_free;
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
64 # define OCTAVE_PCRE_CASELESS PCRE2_CASELESS
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
65 # define OCTAVE_PCRE_DOTALL PCRE2_DOTALL
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
66 # define OCTAVE_PCRE_MULTILINE PCRE2_MULTILINE
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
67 # define OCTAVE_PCRE_EXTENDED PCRE2_EXTENDED
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
68 # define OCTAVE_PCRE_UTF PCRE2_UTF
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
69 # define OCTAVE_PCRE_INFO_CAPTURECOUNT PCRE2_INFO_CAPTURECOUNT
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
70 # define OCTAVE_PCRE_INFO_NAMECOUNT PCRE2_INFO_NAMECOUNT
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
71 # define OCTAVE_PCRE_INFO_NAMEENTRYSIZE PCRE2_INFO_NAMEENTRYSIZE
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
72 # define OCTAVE_PCRE_INFO_NAMETABLE PCRE2_INFO_NAMETABLE
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
73 #elif defined (HAVE_PCRE)
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
74 typedef pcre octave_pcre_code;
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
75 typedef int OCTAVE_PCRE_SIZE;
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
76 void (*octave_pcre_code_free) (void *) = pcre_free;
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
77 # define OCTAVE_PCRE_CASELESS PCRE_CASELESS
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
78 # define OCTAVE_PCRE_DOTALL PCRE_DOTALL
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
79 # define OCTAVE_PCRE_MULTILINE PCRE_MULTILINE
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
80 # define OCTAVE_PCRE_EXTENDED PCRE_EXTENDED
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
81 # define OCTAVE_PCRE_UTF PCRE_UTF8
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
82 # define OCTAVE_PCRE_INFO_CAPTURECOUNT PCRE_INFO_CAPTURECOUNT
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
83 # define OCTAVE_PCRE_INFO_NAMECOUNT PCRE_INFO_NAMECOUNT
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
84 # define OCTAVE_PCRE_INFO_NAMEENTRYSIZE PCRE_INFO_NAMEENTRYSIZE
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
85 # define OCTAVE_PCRE_INFO_NAMETABLE PCRE_INFO_NAMETABLE
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
86 #else
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
87 # error "PCRE2 or PCRE library is required to build Octave"
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
88 #endif
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
89
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
90 static inline int
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
91 octave_pcre_pattern_info (const octave_pcre_code *code, int what, void *where)
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
92 {
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
93 #if defined (HAVE_PCRE2)
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
94 return pcre2_pattern_info (code, what, where);
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
95 #else
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
96 return pcre_fullinfo (code, nullptr, what, where);
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
97 #endif
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
98 }
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
99
31605
e88a07dec498 maint: Use macros to begin/end C++ namespaces.
Rik <rik@octave.org>
parents: 31425
diff changeset
100 OCTAVE_BEGIN_NAMESPACE(octave)
e88a07dec498 maint: Use macros to begin/end C++ namespaces.
Rik <rik@octave.org>
parents: 31425
diff changeset
101
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
102 // Define the maximum number of retries for a pattern
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
103 // that possibly results in an infinite recursion.
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
104 #define PCRE_MATCHLIMIT_MAX 10
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
105
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
106 // FIXME: should this be configurable?
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
107 #define MAXLOOKBEHIND 10
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
108
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
109 static bool lookbehind_warned = false;
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
110
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
111 // FIXME: don't bother collecting and composing return values
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
112 // the user doesn't want.
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
113
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
114 void
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
115 regexp::free (void)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
116 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
117 octave_pcre_code_free (static_cast<octave_pcre_code *> (m_code));
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
118 }
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
119
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
120 void
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
121 regexp::compile_internal (void)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
122 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
123 // If we had a previously compiled pattern, release it.
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
124 free ();
22333
2758af148ced move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents: 22323
diff changeset
125
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
126 std::size_t max_length = MAXLOOKBEHIND;
22333
2758af148ced move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents: 22323
diff changeset
127
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
128 std::size_t pos = 0;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
129 std::size_t new_pos;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
130 int inames = 0;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
131 std::ostringstream buf;
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
132
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
133 while ((new_pos = m_pattern.find ("(?", pos)) != std::string::npos)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
134 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
135 std::size_t tmp_pos;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
136 if (m_pattern.size () > new_pos + 2
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
137 && m_pattern.at (new_pos + 2) == '<'
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
138 && ! (m_pattern.size () > new_pos + 3
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
139 && (m_pattern.at (new_pos + 3) == '='
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
140 || m_pattern.at (new_pos + 3) == '!'))
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
141 && (tmp_pos = m_pattern.find_first_of ('>', new_pos))
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
142 != std::string::npos
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
143 && m_pattern.find_first_of (')', tmp_pos) != std::string::npos)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
144 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
145 // The syntax of named tokens in pcre is "(?P<name>...)" while
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
146 // we need a syntax "(?<name>...)", so fix that here. Also an
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
147 // expression like
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
148 // "(?<first>\w+)\s+(?<last>\w+)|(?<last>\w+),\s+(?<first>\w+)"
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
149 // should be perfectly legal, while pcre does not allow the same
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
150 // named token name on both sides of the alternative. Also fix
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
151 // that here by replacing name tokens by dummy names, and dealing
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
152 // with the dummy names later.
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
153
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
154 std::string tmp_name
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
155 = m_pattern.substr (new_pos+3, tmp_pos-new_pos-3);
22333
2758af148ced move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents: 22323
diff changeset
156
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
157 bool found = false;
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
158
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
159 for (int i = 0; i < m_names; i++)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
160 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
161 if (m_named_pats(i) == tmp_name)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
162 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
163 m_named_idx.resize (dim_vector (inames+1, 1));
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
164 m_named_idx(inames) = i;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
165 found = true;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
166 break;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
167 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
168 }
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
169
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
170 if (! found)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
171 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
172 m_named_idx.resize (dim_vector (inames+1, 1));
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
173 m_named_idx(inames) = m_names;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
174 m_named_pats.append (tmp_name);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
175 m_names++;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
176 }
22333
2758af148ced move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents: 22323
diff changeset
177
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
178 if (new_pos - pos > 0)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
179 buf << m_pattern.substr (pos, new_pos-pos);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
180 if (inames < 10)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
181 buf << "(?P<n00" << inames++;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
182 else if (inames < 100)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
183 buf << "(?P<n0" << inames++;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
184 else
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
185 buf << "(?P<n" << inames++;
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
186
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
187 pos = tmp_pos;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
188 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
189 else if (m_pattern.size () > new_pos + 2
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
190 && m_pattern.at (new_pos + 2) == '<')
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
191 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
192 // Find lookbehind operators of arbitrary length (ie like
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
193 // "(?<=[a-z]*)") and replace with a maximum length operator
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
194 // as PCRE can not yet handle arbitrary length lookahead
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
195 // operators. Use the string length as the maximum length to
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
196 // avoid issues.
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
197
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
198 int brackets = 1;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
199 std::size_t tmp_pos1 = new_pos + 2;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
200 std::size_t tmp_pos2 = tmp_pos1;
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
201
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
202 while (tmp_pos1 < m_pattern.length () && brackets > 0)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
203 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
204 char ch = m_pattern.at (tmp_pos1);
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
205
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
206 if (ch == '(')
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
207 brackets++;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
208 else if (ch == ')')
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
209 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
210 if (brackets > 1)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
211 tmp_pos2 = tmp_pos1;
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
212
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
213 brackets--;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
214 }
22333
2758af148ced move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents: 22323
diff changeset
215
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
216 tmp_pos1++;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
217 }
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
218
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
219 if (brackets != 0)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
220 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
221 buf << m_pattern.substr (pos, new_pos - pos) << "(?";
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
222 pos = new_pos + 2;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
223 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
224 else
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
225 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
226 std::size_t tmp_pos3 = m_pattern.find_first_of ("*+", tmp_pos2);
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
227
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
228 if (tmp_pos3 != std::string::npos && tmp_pos3 < tmp_pos1)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
229 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
230 if (! lookbehind_warned)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
231 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
232 lookbehind_warned = true;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
233 (*current_liboctave_warning_with_id_handler)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
234 ("Octave:regexp-lookbehind-limit",
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
235 "%s: arbitrary length lookbehind patterns are only supported up to length %d",
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
236 m_who.c_str (), MAXLOOKBEHIND);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
237 }
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
238
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
239 buf << m_pattern.substr (pos, new_pos - pos) << '(';
22333
2758af148ced move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents: 22323
diff changeset
240
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
241 std::size_t i;
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
242
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
243 if (m_pattern.at (tmp_pos3) == '*')
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
244 i = 0;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
245 else
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
246 i = 1;
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
247
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
248 for (; i < max_length + 1; i++)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
249 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
250 buf << m_pattern.substr (new_pos, tmp_pos3 - new_pos)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
251 << '{' << i << '}';
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
252 buf << m_pattern.substr (tmp_pos3 + 1,
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
253 tmp_pos1 - tmp_pos3 - 1);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
254 if (i != max_length)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
255 buf << '|';
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
256 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
257 buf << ')';
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
258 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
259 else
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
260 buf << m_pattern.substr (pos, tmp_pos1 - pos);
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
261
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
262 pos = tmp_pos1;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
263 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
264 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
265 else
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
266 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
267 buf << m_pattern.substr (pos, new_pos - pos) << "(?";
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
268 pos = new_pos + 2;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
269 }
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
270
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
271 }
22333
2758af148ced move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents: 22323
diff changeset
272
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
273 buf << m_pattern.substr (pos);
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
274
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
275 // Replace NULLs with escape sequence because conversion function c_str()
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
276 // will terminate string early at embedded NULLs.
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
277 std::string buf_str = buf.str ();
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
278 while ((pos = buf_str.find ('\0')) != std::string::npos)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
279 buf_str.replace (pos, 1, "\\000");
24741
00dfa167c1fe Fix handling of NULL character in regular expressions for Matlab compatibility.
Rik <rik@octave.org>
parents: 24607
diff changeset
280
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
281 int pcre_options
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
282 = ( (m_options.case_insensitive () ? OCTAVE_PCRE_CASELESS : 0)
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
283 | (m_options.dotexceptnewline () ? 0 : OCTAVE_PCRE_DOTALL)
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
284 | (m_options.lineanchors () ? OCTAVE_PCRE_MULTILINE : 0)
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
285 | (m_options.freespacing () ? OCTAVE_PCRE_EXTENDED : 0)
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
286 | OCTAVE_PCRE_UTF);
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
287
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
288 #if defined (HAVE_PCRE2)
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
289 PCRE2_SIZE erroffset;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
290 int errnumber;
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
291
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
292 m_code = pcre2_compile (reinterpret_cast<PCRE2_SPTR> (buf_str.c_str ()),
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
293 PCRE2_ZERO_TERMINATED, pcre_options,
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
294 &errnumber, &erroffset, nullptr);
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
295
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
296 if (! m_code)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
297 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
298 // PCRE docs say:
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
299 //
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
300 // If the buffer is too small, the message is truncated (but
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
301 // still with a trailing zero), and the negative error code
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
302 // PCRE2_ERROR_NOMEMORY is returned. None of the messages are
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
303 // very long; a buffer size of 120 code units is ample.
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
304 //
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
305 // so we assume that 256 will be large enough to avoid truncated
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
306 // messages.
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
307
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
308 PCRE2_UCHAR err [256];
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
309 pcre2_get_error_message (errnumber, err, sizeof (err));
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
310 (*current_liboctave_error_handler)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
311 ("%s: %s at position %zu of expression", m_who.c_str (), err,
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
312 erroffset);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
313 }
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
314 #else
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
315 const char *err;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
316 int erroffset;
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
317
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
318 m_code = pcre_compile (buf_str.c_str (), pcre_options,
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
319 &err, &erroffset, nullptr);
22333
2758af148ced move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents: 22323
diff changeset
320
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
321 if (! m_code)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
322 (*current_liboctave_error_handler)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
323 ("%s: %s at position %d of expression", m_who.c_str (), err, erroffset);
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
324 #endif
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
325 }
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
326
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
327 regexp::match_data
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
328 regexp::match (const std::string& buffer) const
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
329 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
330 // check if input is valid utf-8
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
331 const uint8_t *buf_str = reinterpret_cast<const uint8_t *> (buffer.c_str ());
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
332 if (octave_u8_check_wrapper (buf_str, buffer.length ()))
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
333 (*current_liboctave_error_handler)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
334 ("%s: the input string is invalid UTF-8", m_who.c_str ());
27523
19ad9150dd69 Check if input to regexp is valid UTF-8 is faster than PCRE's checks (bug #57064).
Markus Mützel <markus.muetzel@gmx.de>
parents: 27379
diff changeset
335
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
336 regexp::match_data retval;
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
337
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
338 std::list<regexp::match_element> lst;
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
339
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
340 int subpatterns;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
341 int namecount;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
342 int nameentrysize;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
343 char *nametable;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
344 std::size_t idx = 0;
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
345
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
346 octave_pcre_code *re = static_cast<octave_pcre_code *> (m_code);
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
347
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
348 octave_pcre_pattern_info (re, OCTAVE_PCRE_INFO_CAPTURECOUNT, &subpatterns);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
349 octave_pcre_pattern_info (re, OCTAVE_PCRE_INFO_NAMECOUNT, &namecount);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
350 octave_pcre_pattern_info (re, OCTAVE_PCRE_INFO_NAMEENTRYSIZE, &nameentrysize);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
351 octave_pcre_pattern_info (re, OCTAVE_PCRE_INFO_NAMETABLE, &nametable);
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
352
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
353 #if defined (HAVE_PCRE)
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
354 OCTAVE_LOCAL_BUFFER (OCTAVE_PCRE_SIZE, ovector, (subpatterns+1)*3);
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
355 #endif
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
356
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
357 OCTAVE_LOCAL_BUFFER (int, nidx, namecount);
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
358
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
359 for (int i = 0; i < namecount; i++)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
360 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
361 // Index of subpattern in first two bytes of name (MSB first).
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
362 // Extract index.
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
363 nidx[i] = (static_cast<int> (nametable[i*nameentrysize])) << 8
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
364 | static_cast<int> (nametable[i*nameentrysize+1]);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
365 }
22333
2758af148ced move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents: 22323
diff changeset
366
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
367 while (true)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
368 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
369 octave_quit ();
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
370
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
371 #if defined (HAVE_PCRE2)
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
372 pcre2_match_data *m_data
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
373 = pcre2_match_data_create_from_pattern (re, nullptr);
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
374
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
375 unwind_action cleanup_match_data
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
376 ([=] () { pcre2_match_data_free (m_data); });
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
377
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
378 int matches = pcre2_match (re, reinterpret_cast<PCRE2_SPTR> (buffer.c_str ()),
22333
2758af148ced move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents: 22323
diff changeset
379 buffer.length (), idx,
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
380 PCRE2_NO_UTF_CHECK | (idx ? PCRE2_NOTBOL : 0),
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
381 m_data, nullptr);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
382
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
383 if (matches < 0 && matches != PCRE2_ERROR_NOMATCH)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
384 (*current_liboctave_error_handler)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
385 ("%s: internal error calling pcre2_match; "
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
386 "error code from pcre2_match is %i", m_who.c_str (), matches);
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
387
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
388 if (matches == PCRE2_ERROR_NOMATCH)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
389 break;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
390
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
391 OCTAVE_PCRE_SIZE *ovector = pcre2_get_ovector_pointer (m_data);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
392 #else
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
393 int matches = pcre_exec (re, nullptr, buffer.c_str (),
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
394 buffer.length (), idx,
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
395 PCRE_NO_UTF8_CHECK | (idx ? PCRE_NOTBOL : 0),
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
396 ovector, (subpatterns+1)*3);
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
397
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
398 if (matches == PCRE_ERROR_MATCHLIMIT)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
399 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
400 // Try harder; start with default value for MATCH_LIMIT
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
401 // and increase it.
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
402 (*current_liboctave_warning_with_id_handler)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
403 ("Octave:regexp-match-limit",
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
404 "your pattern caused PCRE to hit its MATCH_LIMIT; trying harder now, but this will be slow");
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
405
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
406 pcre_extra pe;
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
407
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
408 pcre_config (PCRE_CONFIG_MATCH_LIMIT,
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
409 static_cast<void *> (&pe.match_limit));
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
410
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
411 pe.flags = PCRE_EXTRA_MATCH_LIMIT;
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
412
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
413 int i = 0;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
414 while (matches == PCRE_ERROR_MATCHLIMIT
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
415 && i++ < PCRE_MATCHLIMIT_MAX)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
416 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
417 octave_quit ();
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
418
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
419 pe.match_limit *= 10;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
420 matches = pcre_exec (re, &pe, buffer.c_str (),
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
421 buffer.length (), idx,
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
422 PCRE_NO_UTF8_CHECK
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
423 | (idx ? PCRE_NOTBOL : 0),
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
424 ovector, (subpatterns+1)*3);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
425 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
426 }
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
427
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
428 if (matches < 0 && matches != PCRE_ERROR_NOMATCH)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
429 (*current_liboctave_error_handler)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
430 ("%s: internal error calling pcre_exec; "
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
431 "error code from pcre_exec is %i", m_who.c_str (), matches);
21136
7cac4e7458f2 maint: clean up code around calls to current_liboctave_error_handler.
Rik <rik@octave.org>
parents: 20955
diff changeset
432
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
433 if (matches == PCRE_ERROR_NOMATCH)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
434 break;
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
435 #endif
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
436 if (ovector[0] >= ovector[1] && ! m_options.emptymatch ())
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
437 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
438 // Zero length match. Skip to next char.
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
439 idx = ovector[0] + 1;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
440 if (idx < buffer.length ())
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
441 continue;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
442 else
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
443 break;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
444 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
445 else
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
446 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
447 int pos_match = 0;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
448 Matrix token_extents (matches-1, 2);
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
449
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
450 for (int i = 1; i < matches; i++)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
451 {
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
452 #if defined (HAVE_PCRE2)
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
453 if (ovector[2*i] != PCRE2_SIZE_MAX
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
454 #else
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
455 if (ovector[2*i] >= 0
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
456 #endif
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
457 && ovector[2*i+1] > 0
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
458 && (i == 1 || ovector[2*i] != ovector[2*i-2]
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
459 || ovector[2*i-1] != ovector[2*i+1]))
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
460 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
461 token_extents(pos_match, 0) = double (ovector[2*i]+1);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
462 token_extents(pos_match++, 1) = double (ovector[2*i+1]);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
463 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
464 }
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
465
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
466 token_extents.resize (pos_match, 2);
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
467
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
468 OCTAVE_PCRE_SIZE start = ovector[0] + 1;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
469 OCTAVE_PCRE_SIZE end = ovector[1];
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
470
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
471 #if defined (HAVE_PCRE2)
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
472 // Must use explicit length constructor as match can contain '\0'.
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
473 std::string match_string = std::string (buffer.c_str() + start - 1,
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
474 end - start + 1);
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
475 #else
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
476 const char **listptr;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
477 int status = pcre_get_substring_list (buffer.c_str (), ovector,
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
478 matches, &listptr);
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
479
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
480 if (status == PCRE_ERROR_NOMEMORY)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
481 (*current_liboctave_error_handler)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
482 ("%s: cannot allocate memory in pcre_get_substring_list",
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
483 m_who.c_str ());
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
484
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
485 // Must use explicit length constructor as match can contain '\0'.
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
486 std::string match_string = std::string (*listptr, end - start + 1);
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
487 #endif
24741
00dfa167c1fe Fix handling of NULL character in regular expressions for Matlab compatibility.
Rik <rik@octave.org>
parents: 24607
diff changeset
488
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
489 string_vector tokens (pos_match);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
490 string_vector named_tokens (m_names);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
491 int pos_offset = 0;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
492 pos_match = 0;
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
493
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
494 for (int i = 1; i < matches; i++)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
495 {
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
496 #if defined (HAVE_PCRE2)
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
497 if (ovector[2*i] != PCRE2_SIZE_MAX
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
498 #else
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
499 if (ovector[2*i] >= 0
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
500 #endif
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
501 && ovector[2*i+1] > 0)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
502 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
503 if (i == 1 || ovector[2*i] != ovector[2*i-2]
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
504 || ovector[2*i-1] != ovector[2*i+1])
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
505 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
506 if (namecount > 0)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
507 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
508 // FIXME: Should probably do this with a map()
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
509 // rather than a linear search. However,
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
510 // the number of captured, named expressions
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
511 // is usually pretty small (< 4)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
512 for (int j = 0; j < namecount; j++)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
513 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
514 if (nidx[j] == i)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
515 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
516 std::size_t len = ovector[2*i+1] - ovector[2*i];
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
517 named_tokens(m_named_idx(j))
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
518 #if defined (HAVE_PCRE2)
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
519 = std::string (buffer.c_str () + ovector[2*i], len);
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
520 #else
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
521 = std::string (*(listptr+i-pos_offset), len);
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
522 #endif
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
523 break;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
524 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
525 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
526 }
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
527
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
528 std::size_t len = ovector[2*i+1] - ovector[2*i];
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
529 #if defined (HAVE_PCRE2)
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
530 tokens(pos_match++) = std::string (buffer.c_str() + ovector[2*i], len);
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
531 #else
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
532 tokens(pos_match++) = std::string (*(listptr+i), len);
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
533 #endif
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
534 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
535 else
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
536 pos_offset++;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
537 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
538 }
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
539
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
540 #if ! defined (HAVE_PCRE2)
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
541 pcre_free_substring_list (listptr);
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
542 #endif
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
543
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
544 // FIXME: MATCH_ELEMENT uses double values for these,
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
545 // presumably because that is what the Octave interpreter
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
546 // uses. Should we check that the values don't exceed
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
547 // flintmax here? It seems unlikely that it would happen,
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
548 // but...
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
549
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
550 double dstart = static_cast<double> (start);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
551 double dend = static_cast<double> (end);
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
552
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
553 regexp::match_element new_elem (named_tokens, tokens, match_string,
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
554 token_extents,
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
555 dstart, dend);
31424
d1165473e4b0 allow Octave to use PCRE2 (bug #61542)
Rafael Laboissiere <rafael@laboissiere.net>
parents: 31423
diff changeset
556
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
557 lst.push_back (new_elem);
14536
6d5c951ec520 Add 'emptymatch', 'noemptymatch' options to regular expressions.
Rik <octave@nomad.inbox5.com>
parents: 14506
diff changeset
558
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
559 if (ovector[1] <= ovector[0])
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
560 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
561 // Zero length match. Skip to next char.
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
562 idx = ovector[0] + 1;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
563 if (idx <= buffer.length ())
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
564 continue;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
565 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
566 else
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
567 idx = ovector[1];
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
568
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
569 if (m_options.once () || idx >= buffer.length ())
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
570 break;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
571 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
572 }
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
573
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
574 retval = regexp::match_data (lst, m_named_pats);
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
575
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
576 return retval;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
577 }
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
578
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
579 bool
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
580 regexp::is_match (const std::string& buffer) const
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
581 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
582 regexp::match_data rx_lst = match (buffer);
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
583
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
584 return rx_lst.size () > 0;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
585 }
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
586
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
587 Array<bool>
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
588 regexp::is_match (const string_vector& buffer) const
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
589 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
590 octave_idx_type len = buffer.numel ();
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
591
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
592 Array<bool> retval (dim_vector (len, 1));
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
593
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
594 for (octave_idx_type i = 0; i < buffer.numel (); i++)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
595 retval(i) = is_match (buffer(i));
7779
791231dac333 Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff changeset
596
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
597 return retval;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
598 }
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
599
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
600 // Declare rep_token_t used in processing replacement string
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
601 struct rep_token_t
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
602 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
603 std::size_t pos;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
604 int num;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
605 };
15541
9db32cabeacf Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents: 15271
diff changeset
606
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
607 std::string
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
608 regexp::replace (const std::string& buffer,
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
609 const std::string& replacement) const
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
610 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
611 std::string retval;
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
612
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
613 const regexp::match_data rx_lst = match (buffer);
15541
9db32cabeacf Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents: 15271
diff changeset
614
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
615 std::size_t num_matches = rx_lst.size ();
15541
9db32cabeacf Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents: 15271
diff changeset
616
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
617 if (num_matches == 0)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
618 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
619 retval = buffer;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
620 return retval;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
621 }
15541
9db32cabeacf Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents: 15271
diff changeset
622
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
623 // Identify replacement tokens; build a vector of group numbers in
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
624 // the replacement string so that we can quickly calculate the size
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
625 // of the replacement.
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
626
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
627 // FIXME: All code assumes that only 10 tokens ($0-$9) exist.
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
628 // $11 represents $1 followed by the character '1' rather than
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
629 // the eleventh capture buffer.
15541
9db32cabeacf Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents: 15271
diff changeset
630
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
631 std::string repstr = replacement;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
632 std::vector<rep_token_t> tokens;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
633 tokens.reserve (5); // Reserve memory for 5 pattern replacements
15541
9db32cabeacf Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents: 15271
diff changeset
634
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
635 for (std::size_t i=0; i < repstr.size (); i++)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
636 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
637 if (repstr[i] == '\\')
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
638 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
639 if (i < repstr.size () - 1 && repstr[i+1] == '$')
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
640 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
641 repstr.erase (i, 1); // erase backslash
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
642 i++; // skip over '$'
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
643 continue;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
644 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
645 if (i < repstr.size () - 1 && repstr[i+1] == '\\')
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
646 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
647 repstr.erase (i, 1); // erase 1st backslash
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
648 continue;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
649 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
650 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
651 else if (repstr[i] == '$')
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
652 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
653 if (i < repstr.size () - 1 && isdigit (repstr[i+1]))
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
654 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
655 rep_token_t tmp_token;
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
656
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
657 tmp_token.pos = i;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
658 tmp_token.num = repstr[i+1]-'0';
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
659 tokens.push_back (tmp_token);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
660 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
661 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
662 }
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
663
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
664 std::string rep;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
665 int num_tokens = tokens.size ();
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
666
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
667 if (num_tokens > 0)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
668 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
669 // Determine replacement length
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
670 const std::size_t replen = repstr.size () - 2*num_tokens;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
671 int delta = 0;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
672 auto p = rx_lst.begin ();
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
673 for (std::size_t i = 0; i < num_matches; i++)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
674 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
675 octave_quit ();
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
676
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
677 double start = p->start ();
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
678 double end = p->end ();
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
679
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
680 const Matrix pairs (p->token_extents ());
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
681 std::size_t pairlen = 0;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
682 for (int j = 0; j < num_tokens; j++)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
683 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
684 if (tokens[j].num == 0)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
685 pairlen += static_cast<std::size_t> (end - start + 1);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
686 else if (tokens[j].num <= pairs.rows ())
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
687 pairlen += static_cast<std::size_t> (pairs(tokens[j].num-1, 1)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
688 - pairs(tokens[j].num-1, 0)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
689 + 1);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
690 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
691 delta += (static_cast<int> (replen + pairlen)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
692 - static_cast<int> (end - start + 1));
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
693 p++;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
694 }
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
695
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
696 // Build replacement string
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
697 rep.reserve (buffer.size () + delta);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
698 std::size_t from = 0;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
699 p = rx_lst.begin ();
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
700 for (std::size_t i = 0; i < num_matches; i++)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
701 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
702 octave_quit ();
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
703
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
704 double start = p->start ();
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
705 double end = p->end ();
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
706
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
707 const Matrix pairs (p->token_extents ());
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
708 rep.append (&buffer[from], static_cast<std::size_t> (start - 1 - from));
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
709 from = static_cast<std::size_t> (end);
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
710
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
711 std::size_t cur_pos = 0;
15541
9db32cabeacf Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents: 15271
diff changeset
712
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
713 for (int j = 0; j < num_tokens; j++)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
714 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
715 rep.append (&repstr[cur_pos], (tokens[j].pos) - cur_pos);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
716 cur_pos = tokens[j].pos+2;
15541
9db32cabeacf Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents: 15271
diff changeset
717
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
718 int k = tokens[j].num;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
719 if (k == 0)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
720 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
721 // replace with entire match
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
722 rep.append (&buffer[static_cast<std::size_t> (end - 1)],
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
723 static_cast<std::size_t> (end - start + 1));
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
724 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
725 else if (k <= pairs.rows ())
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
726 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
727 // replace with group capture
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
728 rep.append (&buffer[static_cast<std::size_t> (pairs(k-1, 0)-1)],
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
729 static_cast<std::size_t> (pairs(k-1, 1)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
730 - pairs(k-1, 0) + 1));
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
731 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
732 else
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
733 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
734 // replace with nothing
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
735 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
736 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
737 if (cur_pos < repstr.size ())
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
738 rep.append (&repstr[cur_pos], repstr.size () - cur_pos);
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
739
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
740 p++;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
741 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
742 rep.append (&buffer[from], buffer.size () - from);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
743 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
744 else
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
745 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
746 // Determine repstr length
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
747 const std::size_t replen = repstr.size ();
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
748 int delta = 0;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
749 auto p = rx_lst.begin ();
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
750 for (std::size_t i = 0; i < num_matches; i++)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
751 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
752 octave_quit ();
22860
0b1e25cc4457 eliminate use of OCTAVE_QUIT macro in C++ sources
John W. Eaton <jwe@octave.org>
parents: 22755
diff changeset
753
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
754 delta += static_cast<int> (replen)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
755 - static_cast<int> (p->end () - p->start () + 1);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
756 p++;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
757 }
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
758
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
759 // Build replacement string
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
760 rep.reserve (buffer.size () + delta);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
761 std::size_t from = 0;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
762 p = rx_lst.begin ();
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
763 for (std::size_t i = 0; i < num_matches; i++)
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
764 {
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
765 octave_quit ();
22860
0b1e25cc4457 eliminate use of OCTAVE_QUIT macro in C++ sources
John W. Eaton <jwe@octave.org>
parents: 22755
diff changeset
766
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
767 rep.append (&buffer[from],
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
768 static_cast<std::size_t> (p->start () - 1 - from));
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
769 from = static_cast<std::size_t> (p->end ());
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
770 rep.append (repstr);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
771 p++;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
772 }
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
773 rep.append (&buffer[from], buffer.size () - from);
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
774 }
14024
fc9f204faea0 refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents: 11586
diff changeset
775
31607
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
776 retval = rep;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
777 return retval;
aac27ad79be6 maint: Re-indent code after switch to using namespace macros.
Rik <rik@octave.org>
parents: 31605
diff changeset
778 }
31605
e88a07dec498 maint: Use macros to begin/end C++ namespaces.
Rik <rik@octave.org>
parents: 31425
diff changeset
779
e88a07dec498 maint: Use macros to begin/end C++ namespaces.
Rik <rik@octave.org>
parents: 31425
diff changeset
780 OCTAVE_END_NAMESPACE(octave)