Mercurial > octave
annotate liboctave/util/lo-regexp.cc @ 23302:78cf75ffd382
maint: tweak readability of code in lo-regexp.cc.
* lo-regexp.cc: Adjust some indentation and reword some comments
to make things clearer.
author | Rik <rik@octave.org> |
---|---|
date | Sat, 18 Mar 2017 10:15:50 -0700 |
parents | ef4d915df748 |
children | 980f39c3ab90 |
rev | line source |
---|---|
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
1 /* |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
2 |
14138
72c96de7a403
maint: update copyright notices for 2012
John W. Eaton <jwe@octave.org>
parents:
14057
diff
changeset
|
3 Copyright (C) 2012 John W. Eaton |
22323
bac0d6f07a3e
maint: Update copyright notices for 2016.
John W. Eaton <jwe@octave.org>
parents:
22022
diff
changeset
|
4 Copyright (C) 2005-2016 David Bateman |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
5 Copyright (C) 2002-2005 Paul Kienzle |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
6 |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
7 This file is part of Octave. |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
8 |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
9 Octave is free software; you can redistribute it and/or modify it |
22755
3a2b891d0b33
maint: Standardize Copyright formatting.
Rik <rik@octave.org>
parents:
22402
diff
changeset
|
10 under the terms of the GNU General Public License as published by |
3a2b891d0b33
maint: Standardize Copyright formatting.
Rik <rik@octave.org>
parents:
22402
diff
changeset
|
11 the Free Software Foundation; either version 3 of the License, or |
3a2b891d0b33
maint: Standardize Copyright formatting.
Rik <rik@octave.org>
parents:
22402
diff
changeset
|
12 (at your option) any later version. |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
13 |
22755
3a2b891d0b33
maint: Standardize Copyright formatting.
Rik <rik@octave.org>
parents:
22402
diff
changeset
|
14 Octave is distributed in the hope that it will be useful, but |
3a2b891d0b33
maint: Standardize Copyright formatting.
Rik <rik@octave.org>
parents:
22402
diff
changeset
|
15 WITHOUT ANY WARRANTY; without even the implied warranty of |
3a2b891d0b33
maint: Standardize Copyright formatting.
Rik <rik@octave.org>
parents:
22402
diff
changeset
|
16 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
3a2b891d0b33
maint: Standardize Copyright formatting.
Rik <rik@octave.org>
parents:
22402
diff
changeset
|
17 GNU General Public License for more details. |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
18 |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
19 You should have received a copy of the GNU General Public License |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
20 along with Octave; see the file COPYING. If not, see |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
21 <http://www.gnu.org/licenses/>. |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
22 |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
23 */ |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
24 |
21724
aba2e6293dd8
use "#if ..." consistently instead of "#ifdef" and "#ifndef"
John W. Eaton <jwe@octave.org>
parents:
21301
diff
changeset
|
25 #if defined (HAVE_CONFIG_H) |
21301
40de9f8f23a6
Use '#include "config.h"' rather than <config.h>.
Rik <rik@octave.org>
parents:
21202
diff
changeset
|
26 # include "config.h" |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
27 #endif |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
28 |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
29 #include <list> |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
30 #include <sstream> |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
31 #include <string> |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
32 #include <vector> |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
33 |
14025
9867be070ee1
use pcre/pcre.h if it is present
John W. Eaton <jwe@octave.org>
parents:
14024
diff
changeset
|
34 #if defined (HAVE_PCRE_H) |
21202
f7121e111991
maint: indent #ifdef blocks in liboctave and src directories.
Rik <rik@octave.org>
parents:
21136
diff
changeset
|
35 # include <pcre.h> |
14025
9867be070ee1
use pcre/pcre.h if it is present
John W. Eaton <jwe@octave.org>
parents:
14024
diff
changeset
|
36 #elif defined (HAVE_PCRE_PCRE_H) |
21202
f7121e111991
maint: indent #ifdef blocks in liboctave and src directories.
Rik <rik@octave.org>
parents:
21136
diff
changeset
|
37 # include <pcre/pcre.h> |
14025
9867be070ee1
use pcre/pcre.h if it is present
John W. Eaton <jwe@octave.org>
parents:
14024
diff
changeset
|
38 #endif |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
39 |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
40 #include "Matrix.h" |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
41 #include "base-list.h" |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
42 #include "lo-error.h" |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
43 #include "oct-locbuf.h" |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
44 #include "quit.h" |
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
|
45 #include "lo-regexp.h" |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
46 #include "str-vec.h" |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
47 |
17769
49a5a4be04a1
maint: Use GNU style coding conventions for code in liboctave/
Rik <rik@octave.org>
parents:
17744
diff
changeset
|
48 // Define the maximum number of retries for a pattern |
49a5a4be04a1
maint: Use GNU style coding conventions for code in liboctave/
Rik <rik@octave.org>
parents:
17744
diff
changeset
|
49 // that possibly results in an infinite recursion. |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
50 #define PCRE_MATCHLIMIT_MAX 10 |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
51 |
17769
49a5a4be04a1
maint: Use GNU style coding conventions for code in liboctave/
Rik <rik@octave.org>
parents:
17744
diff
changeset
|
52 // FIXME: should this be configurable? |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
53 #define MAXLOOKBEHIND 10 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
54 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
55 static bool lookbehind_warned = false; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
56 |
17769
49a5a4be04a1
maint: Use GNU style coding conventions for code in liboctave/
Rik <rik@octave.org>
parents:
17744
diff
changeset
|
57 // FIXME: don't bother collecting and composing return values |
49a5a4be04a1
maint: Use GNU style coding conventions for code in liboctave/
Rik <rik@octave.org>
parents:
17744
diff
changeset
|
58 // the user doesn't want. |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
59 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
60 namespace octave |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
61 { |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
62 void |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
63 regexp::free (void) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
64 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
65 if (data) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
66 pcre_free (static_cast<pcre *> (data)); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
67 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
68 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
69 void |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
70 regexp::compile_internal (void) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
71 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
72 // If we had a previously compiled pattern, release it. |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
73 free (); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
74 |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
75 size_t max_length = MAXLOOKBEHIND; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
76 |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
77 size_t pos = 0; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
78 size_t new_pos; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
79 int inames = 0; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
80 std::ostringstream buf; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
81 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
82 while ((new_pos = pattern.find ("(?", pos)) != std::string::npos) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
83 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
84 if (pattern.at (new_pos + 2) == '<' |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
85 && !(pattern.at (new_pos + 3) == '=' |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
86 || pattern.at (new_pos + 3) == '!')) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
87 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
88 // The syntax of named tokens in pcre is "(?P<name>...)" while |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
89 // we need a syntax "(?<name>...)", so fix that here. Also an |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
90 // expression like |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
91 // "(?<first>\w+)\s+(?<last>\w+)|(?<last>\w+),\s+(?<first>\w+)" |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
92 // should be perfectly legal, while pcre does not allow the same |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
93 // named token name on both sides of the alternative. Also fix |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
94 // that here by replacing name tokens by dummy names, and dealing |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
95 // with the dummy names later. |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
96 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
97 size_t tmp_pos = pattern.find_first_of ('>', new_pos); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
98 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
99 if (tmp_pos == std::string::npos) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
100 (*current_liboctave_error_handler) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
101 ("regexp: syntax error in pattern"); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
102 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
103 std::string tmp_name = |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
104 pattern.substr (new_pos+3, tmp_pos-new_pos-3); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
105 |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
106 bool found = false; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
107 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
108 for (int i = 0; i < nnames; i++) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
109 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
110 if (named_pats(i) == tmp_name) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
111 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
112 named_idx.resize (dim_vector (inames+1, 1)); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
113 named_idx(inames) = i; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
114 found = true; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
115 break; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
116 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
117 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
118 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
119 if (! found) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
120 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
121 named_idx.resize (dim_vector (inames+1, 1)); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
122 named_idx(inames) = nnames; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
123 named_pats.append (tmp_name); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
124 nnames++; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
125 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
126 |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
127 if (new_pos - pos > 0) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
128 buf << pattern.substr (pos, new_pos-pos); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
129 if (inames < 10) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
130 buf << "(?P<n00" << inames++; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
131 else if (inames < 100) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
132 buf << "(?P<n0" << inames++; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
133 else |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
134 buf << "(?P<n" << inames++; |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
135 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
136 pos = tmp_pos; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
137 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
138 else if (pattern.at (new_pos + 2) == '<') |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
139 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
140 // Find lookbehind operators of arbitrary length (ie like |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
141 // "(?<=[a-z]*)") and replace with a maximum length operator |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
142 // as PCRE can not yet handle arbitrary length lookahead |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
143 // operators. Use the string length as the maximum length to |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
144 // avoid issues. |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
145 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
146 int brackets = 1; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
147 size_t tmp_pos1 = new_pos + 2; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
148 size_t tmp_pos2 = tmp_pos1; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
149 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
150 while (tmp_pos1 < pattern.length () && brackets > 0) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
151 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
152 char ch = pattern.at (tmp_pos1); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
153 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
154 if (ch == '(') |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
155 brackets++; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
156 else if (ch == ')') |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
157 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
158 if (brackets > 1) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
159 tmp_pos2 = tmp_pos1; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
160 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
161 brackets--; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
162 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
163 |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
164 tmp_pos1++; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
165 } |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
166 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
167 if (brackets != 0) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
168 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
169 buf << pattern.substr (pos, new_pos - pos) << "(?"; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
170 pos = new_pos + 2; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
171 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
172 else |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
173 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
174 size_t tmp_pos3 = pattern.find_first_of ("*+", tmp_pos2); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
175 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
176 if (tmp_pos3 != std::string::npos && tmp_pos3 < tmp_pos1) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
177 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
178 if (! lookbehind_warned) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
179 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
180 lookbehind_warned = true; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
181 (*current_liboctave_warning_with_id_handler) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
182 ("Octave:regexp-lookbehind-limit", |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
183 "%s: arbitrary length lookbehind patterns are only supported up to length %d", |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
184 who.c_str (), MAXLOOKBEHIND); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
185 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
186 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
187 buf << pattern.substr (pos, new_pos - pos) << "("; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
188 |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
189 size_t i; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
190 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
191 if (pattern.at (tmp_pos3) == '*') |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
192 i = 0; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
193 else |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
194 i = 1; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
195 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
196 for (; i < max_length + 1; i++) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
197 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
198 buf << pattern.substr (new_pos, tmp_pos3 - new_pos) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
199 << "{" << i << "}"; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
200 buf << pattern.substr (tmp_pos3 + 1, |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
201 tmp_pos1 - tmp_pos3 - 1); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
202 if (i != max_length) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
203 buf << "|"; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
204 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
205 buf << ")"; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
206 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
207 else |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
208 buf << pattern.substr (pos, tmp_pos1 - pos); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
209 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
210 pos = tmp_pos1; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
211 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
212 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
213 else |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
214 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
215 buf << pattern.substr (pos, new_pos - pos) << "(?"; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
216 pos = new_pos + 2; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
217 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
218 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
219 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
220 |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
221 buf << pattern.substr (pos); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
222 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
223 const char *err; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
224 int erroffset; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
225 std::string buf_str = buf.str (); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
226 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
227 int pcre_options |
23302
78cf75ffd382
maint: tweak readability of code in lo-regexp.cc.
Rik <rik@octave.org>
parents:
23084
diff
changeset
|
228 = ( (options.case_insensitive () ? PCRE_CASELESS : 0) |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
229 | (options.dotexceptnewline () ? 0 : PCRE_DOTALL) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
230 | (options.lineanchors () ? PCRE_MULTILINE : 0) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
231 | (options.freespacing () ? PCRE_EXTENDED : 0)); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
232 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
233 data = pcre_compile (buf_str.c_str (), pcre_options, &err, &erroffset, 0); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
234 |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
235 if (! data) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
236 (*current_liboctave_error_handler) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
237 ("%s: %s at position %d of expression", who.c_str (), err, erroffset); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
238 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
239 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
240 regexp::match_data |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
241 regexp::match (const std::string& buffer) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
242 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
243 regexp::match_data retval; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
244 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
245 std::list<regexp::match_element> lst; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
246 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
247 int subpatterns; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
248 int namecount; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
249 int nameentrysize; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
250 char *nametable; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
251 size_t idx = 0; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
252 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
253 pcre *re = static_cast<pcre *> (data); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
254 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
255 pcre_fullinfo (re, 0, PCRE_INFO_CAPTURECOUNT, &subpatterns); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
256 pcre_fullinfo (re, 0, PCRE_INFO_NAMECOUNT, &namecount); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
257 pcre_fullinfo (re, 0, PCRE_INFO_NAMEENTRYSIZE, &nameentrysize); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
258 pcre_fullinfo (re, 0, PCRE_INFO_NAMETABLE, &nametable); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
259 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
260 OCTAVE_LOCAL_BUFFER (int, ovector, (subpatterns+1)*3); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
261 OCTAVE_LOCAL_BUFFER (int, nidx, namecount); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
262 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
263 for (int i = 0; i < namecount; i++) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
264 { |
23302
78cf75ffd382
maint: tweak readability of code in lo-regexp.cc.
Rik <rik@octave.org>
parents:
23084
diff
changeset
|
265 // Index of subpattern in first two bytes of name (MSB first). |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
266 // Extract index. |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
267 nidx[i] = (static_cast<int> (nametable[i*nameentrysize])) << 8 |
22402
4caa7b28d183
maint: Style check C++ code in liboctave/
Rik <rik@octave.org>
parents:
22333
diff
changeset
|
268 | static_cast<int> (nametable[i*nameentrysize+1]); |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
269 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
270 |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
271 while (true) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
272 { |
22860
0b1e25cc4457
eliminate use of OCTAVE_QUIT macro in C++ sources
John W. Eaton <jwe@octave.org>
parents:
22755
diff
changeset
|
273 octave_quit (); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
274 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
275 int matches = pcre_exec (re, 0, buffer.c_str (), |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
276 buffer.length (), idx, |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
277 (idx ? PCRE_NOTBOL : 0), |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
278 ovector, (subpatterns+1)*3); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
279 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
280 if (matches == PCRE_ERROR_MATCHLIMIT) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
281 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
282 // Try harder; start with default value for MATCH_LIMIT |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
283 // and increase it. |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
284 (*current_liboctave_warning_with_id_handler) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
285 ("Octave:regexp-match-limit", |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
286 "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
|
287 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
288 pcre_extra pe; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
289 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
290 pcre_config (PCRE_CONFIG_MATCH_LIMIT, |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
291 static_cast<void *> (&pe.match_limit)); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
292 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
293 pe.flags = PCRE_EXTRA_MATCH_LIMIT; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
294 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
295 int i = 0; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
296 while (matches == PCRE_ERROR_MATCHLIMIT |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
297 && i++ < PCRE_MATCHLIMIT_MAX) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
298 { |
22860
0b1e25cc4457
eliminate use of OCTAVE_QUIT macro in C++ sources
John W. Eaton <jwe@octave.org>
parents:
22755
diff
changeset
|
299 octave_quit (); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
300 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
301 pe.match_limit *= 10; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
302 matches = pcre_exec (re, &pe, buffer.c_str (), |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
303 buffer.length (), idx, |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
304 (idx ? PCRE_NOTBOL : 0), |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
305 ovector, (subpatterns+1)*3); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
306 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
307 } |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
308 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
309 if (matches < 0 && matches != PCRE_ERROR_NOMATCH) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
310 (*current_liboctave_error_handler) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
311 ("%s: internal error calling pcre_exec; " |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
312 "error code from pcre_exec is %i", 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
|
313 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
314 if (matches == PCRE_ERROR_NOMATCH) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
315 break; |
23302
78cf75ffd382
maint: tweak readability of code in lo-regexp.cc.
Rik <rik@octave.org>
parents:
23084
diff
changeset
|
316 else if (ovector[0] >= ovector[1] && ! options.emptymatch ()) |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
317 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
318 // Zero length match. Skip to next char. |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
319 idx = ovector[0] + 1; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
320 if (idx < buffer.length ()) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
321 continue; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
322 else |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
323 break; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
324 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
325 else |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
326 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
327 int pos_match = 0; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
328 Matrix token_extents (matches-1, 2); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
329 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
330 for (int i = 1; i < matches; i++) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
331 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
332 if (ovector[2*i] >= 0 && ovector[2*i+1] > 0 |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
333 && (i == 1 || ovector[2*i] != ovector[2*i-2] |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
334 || ovector[2*i-1] != ovector[2*i+1])) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
335 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
336 token_extents(pos_match,0) = double (ovector[2*i]+1); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
337 token_extents(pos_match++,1) = double (ovector[2*i+1]); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
338 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
339 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
340 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
341 token_extents.resize (pos_match, 2); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
342 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
343 double start = double (ovector[0]+1); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
344 double end = double (ovector[1]); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
345 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
346 const char **listptr; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
347 int status = pcre_get_substring_list (buffer.c_str (), ovector, |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
348 matches, &listptr); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
349 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
350 if (status == PCRE_ERROR_NOMEMORY) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
351 (*current_liboctave_error_handler) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
352 ("%s: cannot allocate memory in pcre_get_substring_list", |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
353 who.c_str ()); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
354 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
355 string_vector tokens (pos_match); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
356 string_vector named_tokens (nnames); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
357 int pos_offset = 0; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
358 pos_match = 0; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
359 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
360 for (int i = 1; i < matches; i++) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
361 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
362 if (ovector[2*i] >= 0 && ovector[2*i+1] > 0) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
363 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
364 if (i == 1 || ovector[2*i] != ovector[2*i-2] |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
365 || ovector[2*i-1] != ovector[2*i+1]) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
366 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
367 if (namecount > 0) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
368 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
369 // FIXME: Should probably do this with a map() |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
370 // rather than a linear search. However, |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
371 // the number of captured, named expressions |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
372 // is usually pretty small (< 4) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
373 for (int j = 0; j < namecount; j++) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
374 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
375 if (nidx[j] == i) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
376 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
377 named_tokens(named_idx(j)) = |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
378 std::string (*(listptr+i-pos_offset)); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
379 break; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
380 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
381 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
382 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
383 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
384 tokens(pos_match++) = std::string (*(listptr+i)); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
385 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
386 else |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
387 pos_offset++; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
388 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
389 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
390 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
391 std::string match_string = std::string (*listptr); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
392 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
393 pcre_free_substring_list (listptr); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
394 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
395 regexp::match_element new_elem (named_tokens, tokens, match_string, |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
396 token_extents, start, end); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
397 lst.push_back (new_elem); |
14536
6d5c951ec520
Add 'emptymatch', 'noemptymatch' options to regular expressions.
Rik <octave@nomad.inbox5.com>
parents:
14506
diff
changeset
|
398 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
399 if (ovector[1] <= ovector[0]) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
400 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
401 // Zero length match. Skip to next char. |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
402 idx = ovector[0] + 1; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
403 if (idx <= buffer.length ()) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
404 continue; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
405 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
406 else |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
407 idx = ovector[1]; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
408 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
409 if (options.once () || idx >= buffer.length ()) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
410 break; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
411 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
412 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
413 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
414 retval = regexp::match_data (lst, named_pats); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
415 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
416 return retval; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
417 } |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
418 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
419 bool |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
420 regexp::is_match (const std::string& buffer) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
421 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
422 regexp::match_data rx_lst = match (buffer); |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
423 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
424 return rx_lst.size () > 0; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
425 } |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
426 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
427 Array<bool> |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
428 regexp::is_match (const string_vector& buffer) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
429 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
430 octave_idx_type len = buffer.numel (); |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
431 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
432 Array<bool> retval (dim_vector (len, 1)); |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
433 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
434 for (octave_idx_type i = 0; i < buffer.numel (); i++) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
435 retval(i) = is_match (buffer(i)); |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
436 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
437 return retval; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
438 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
439 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
440 // Declare rep_token_t used in processing replacement string |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
441 typedef struct |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
442 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
443 size_t pos; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
444 int num; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
445 } rep_token_t; |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
446 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
447 std::string |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
448 regexp::replace (const std::string& buffer, const std::string& replacement) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
449 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
450 std::string retval; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
451 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
452 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
|
453 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
454 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
|
455 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
456 if (num_matches == 0) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
457 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
458 retval = buffer; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
459 return retval; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
460 } |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
461 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
462 // Identify replacement tokens; build a vector of group numbers in |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
463 // the replacement string so that we can quickly calculate the size |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
464 // of the replacement. |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
465 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
466 // FIXME: All code assumes that only 10 tokens ($0-$9) exist. |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
467 // $11 represents $1 followed by the character '1' rather than |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
468 // the eleventh capture buffer. |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
469 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
470 std::string repstr = replacement; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
471 std::vector<rep_token_t> tokens; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
472 tokens.reserve (5); // Reserve memory for 5 pattern replacements |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
473 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
474 for (size_t i=0; i < repstr.size (); i++) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
475 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
476 if (repstr[i] == '\\') |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
477 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
478 if (i < repstr.size () - 1 && repstr[i+1] == '$') |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
479 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
480 repstr.erase (i,1); // erase backslash |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
481 i++; // skip over '$' |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
482 continue; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
483 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
484 if (i < repstr.size () - 1 && repstr[i+1] == '\\') |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
485 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
486 repstr.erase (i,1); // erase 1st backslash |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
487 continue; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
488 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
489 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
490 else if (repstr[i] == '$') |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
491 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
492 if (i < repstr.size () - 1 && isdigit (repstr[i+1])) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
493 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
494 rep_token_t tmp_token; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
495 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
496 tmp_token.pos = i; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
497 tmp_token.num = repstr[i+1]-'0'; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
498 tokens.push_back (tmp_token); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
499 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
500 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
501 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
502 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
503 std::string rep; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
504 int num_tokens = tokens.size (); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
505 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
506 if (num_tokens > 0) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
507 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
508 // Determine replacement length |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
509 const size_t replen = repstr.size () - 2*num_tokens; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
510 int delta = 0; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
511 regexp::match_data::const_iterator p = rx_lst.begin (); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
512 for (size_t i = 0; i < num_matches; i++) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
513 { |
22860
0b1e25cc4457
eliminate use of OCTAVE_QUIT macro in C++ sources
John W. Eaton <jwe@octave.org>
parents:
22755
diff
changeset
|
514 octave_quit (); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
515 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
516 double start = p->start (); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
517 double end = p->end (); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
518 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
519 const Matrix pairs (p->token_extents ()); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
520 size_t pairlen = 0; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
521 for (int j = 0; j < num_tokens; j++) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
522 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
523 if (tokens[j].num == 0) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
524 pairlen += static_cast<size_t> (end - start) + 1; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
525 else if (tokens[j].num <= pairs.rows ()) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
526 pairlen += static_cast<size_t> (pairs(tokens[j].num-1,1) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
527 - pairs(tokens[j].num-1,0)) + 1; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
528 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
529 delta += (static_cast<int> (replen + pairlen) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
530 - static_cast<int> (end - start + 1)); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
531 p++; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
532 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
533 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
534 // Build replacement string |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
535 rep.reserve (buffer.size () + delta); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
536 size_t from = 0; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
537 p = rx_lst.begin (); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
538 for (size_t i = 0; i < num_matches; i++) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
539 { |
22860
0b1e25cc4457
eliminate use of OCTAVE_QUIT macro in C++ sources
John W. Eaton <jwe@octave.org>
parents:
22755
diff
changeset
|
540 octave_quit (); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
541 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
542 double start = p->start (); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
543 double end = p->end (); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
544 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
545 const Matrix pairs (p->token_extents ()); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
546 rep.append (&buffer[from], static_cast<size_t> (start - 1) - from); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
547 from = static_cast<size_t> (end); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
548 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
549 size_t cur_pos = 0; |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
550 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
551 for (int j = 0; j < num_tokens; j++) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
552 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
553 rep.append (&repstr[cur_pos], (tokens[j].pos) - cur_pos); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
554 cur_pos = tokens[j].pos+2; |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
555 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
556 int k = tokens[j].num; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
557 if (k == 0) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
558 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
559 // replace with entire match |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
560 rep.append (&buffer[static_cast<size_t> (end - 1)], |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
561 static_cast<size_t> (end - start) + 1); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
562 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
563 else if (k <= pairs.rows ()) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
564 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
565 // replace with group capture |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
566 rep.append (&buffer[static_cast<size_t> (pairs(k-1,0)-1)], |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
567 static_cast<size_t> (pairs(k-1,1) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
568 - pairs(k-1,0)) + 1); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
569 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
570 else |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
571 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
572 // replace with nothing |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
573 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
574 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
575 if (cur_pos < repstr.size ()) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
576 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
|
577 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
578 p++; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
579 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
580 rep.append (&buffer[from], buffer.size () - from); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
581 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
582 else |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
583 { |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
584 // Determine repstr length |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
585 const size_t replen = repstr.size (); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
586 int delta = 0; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
587 regexp::match_data::const_iterator p = rx_lst.begin (); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
588 for (size_t i = 0; i < num_matches; i++) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
589 { |
22860
0b1e25cc4457
eliminate use of OCTAVE_QUIT macro in C++ sources
John W. Eaton <jwe@octave.org>
parents:
22755
diff
changeset
|
590 octave_quit (); |
0b1e25cc4457
eliminate use of OCTAVE_QUIT macro in C++ sources
John W. Eaton <jwe@octave.org>
parents:
22755
diff
changeset
|
591 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
592 delta += static_cast<int> (replen) |
22402
4caa7b28d183
maint: Style check C++ code in liboctave/
Rik <rik@octave.org>
parents:
22333
diff
changeset
|
593 - static_cast<int> (p->end () - p->start () + 1); |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
594 p++; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
595 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
596 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
597 // Build replacement string |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
598 rep.reserve (buffer.size () + delta); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
599 size_t from = 0; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
600 p = rx_lst.begin (); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
601 for (size_t i = 0; i < num_matches; i++) |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
602 { |
22860
0b1e25cc4457
eliminate use of OCTAVE_QUIT macro in C++ sources
John W. Eaton <jwe@octave.org>
parents:
22755
diff
changeset
|
603 octave_quit (); |
0b1e25cc4457
eliminate use of OCTAVE_QUIT macro in C++ sources
John W. Eaton <jwe@octave.org>
parents:
22755
diff
changeset
|
604 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
605 rep.append (&buffer[from], |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
606 static_cast<size_t> (p->start () - 1) - from); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
607 from = static_cast<size_t> (p->end ()); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
608 rep.append (repstr); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
609 p++; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
610 } |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
611 rep.append (&buffer[from], buffer.size () - from); |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
612 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
613 |
22333
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
614 retval = rep; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
615 return retval; |
2758af148ced
move base_list and regexp classes to octave namespace
John W. Eaton <jwe@octave.org>
parents:
22323
diff
changeset
|
616 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
617 } |