Mercurial > hg > octave-lyh
annotate liboctave/util/regexp.cc @ 15541:9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
* NEWS: Give an example of how escape sequence processing in single-quoted
regular expressions works.
* libinterp/corefcn/regexp.cc(do_regexp_string_escapes): Rename to
do_regexp_ptn_string_escapes. Only sequence to expand is '\b' for backspace.
Others are handled by PCRE.
* libinterp/corefcn/regexp.cc(do_regexp_rep_string_escapes): New function to
do escape sequence processing for the replacement string since the sequences
to expand differ from that of the regexp pattern.
* liboctave/util/regexp.cc(regexp::replace): Process backslashes in replacement
string so that '\$1' results in '$1' rather than replacement with first
capture buffer.
author | Rik <rik@octave.org> |
---|---|
date | Wed, 17 Oct 2012 20:13:19 -0700 |
parents | 648dabbb4c6b |
children | 1e9a6285acc4 |
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 |
72c96de7a403
maint: update copyright notices for 2012
John W. Eaton <jwe@octave.org>
parents:
14057
diff
changeset
|
4 Copyright (C) 2005-2012 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 |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
10 under the terms of the GNU General Public License as published by the |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
11 Free Software Foundation; either version 3 of the License, or (at your |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
12 option) any later version. |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
13 |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
14 Octave is distributed in the hope that it will be useful, but WITHOUT |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
15 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
16 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
17 for more details. |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
18 |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
19 You should have received a copy of the GNU General Public License |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
20 along with Octave; see the file COPYING. If not, see |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
21 <http://www.gnu.org/licenses/>. |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
22 |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
23 */ |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
24 |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
25 #ifdef HAVE_CONFIG_H |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
26 #include <config.h> |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
27 #endif |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
28 |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
29 #include <list> |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
30 #include <sstream> |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
31 #include <string> |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
32 #include <vector> |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
33 |
14025
9867be070ee1
use pcre/pcre.h if it is present
John W. Eaton <jwe@octave.org>
parents:
14024
diff
changeset
|
34 #if defined (HAVE_PCRE_H) |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
35 #include <pcre.h> |
14025
9867be070ee1
use pcre/pcre.h if it is present
John W. Eaton <jwe@octave.org>
parents:
14024
diff
changeset
|
36 #elif defined (HAVE_PCRE_PCRE_H) |
9867be070ee1
use pcre/pcre.h if it is present
John W. Eaton <jwe@octave.org>
parents:
14024
diff
changeset
|
37 #include <pcre/pcre.h> |
9867be070ee1
use pcre/pcre.h if it is present
John W. Eaton <jwe@octave.org>
parents:
14024
diff
changeset
|
38 #endif |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
39 |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
40 #include "Matrix.h" |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
41 #include "base-list.h" |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
42 #include "lo-error.h" |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
43 #include "oct-locbuf.h" |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
44 #include "quit.h" |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
45 #include "regexp.h" |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
46 #include "str-vec.h" |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
47 |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
48 // Define the maximum number of retries for a pattern that possibly |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
49 // results in an infinite recursion. |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
50 #define PCRE_MATCHLIMIT_MAX 10 |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
51 |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
52 // FIXME -- should this be configurable? |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
53 #define MAXLOOKBEHIND 10 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
54 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
55 static bool lookbehind_warned = false; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
56 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
57 // FIXME -- don't bother collecting and composing return values the user |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
58 // doesn't want. |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
59 |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11570
diff
changeset
|
60 void |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
61 regexp::free (void) |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11570
diff
changeset
|
62 { |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
63 if (data) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
64 pcre_free (static_cast<pcre *> (data)); |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
65 } |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
66 |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
67 void |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
68 regexp::compile_internal (void) |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
69 { |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
70 // If we had a previously compiled pattern, release it. |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
71 free (); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
72 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
73 size_t max_length = MAXLOOKBEHIND; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
74 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
75 size_t pos = 0; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
76 size_t new_pos; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
77 int inames = 0; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
78 std::ostringstream buf; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
79 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
80 while ((new_pos = pattern.find ("(?", pos)) != std::string::npos) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
81 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
82 if (pattern.at (new_pos + 2) == '<' |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
83 && !(pattern.at (new_pos + 3) == '=' |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
84 || pattern.at (new_pos + 3) == '!')) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
85 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
86 // The syntax of named tokens in pcre is "(?P<name>...)" while |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
87 // we need a syntax "(?<name>...)", so fix that here. Also an |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
88 // expression like |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
89 // "(?<first>\w+)\s+(?<last>\w+)|(?<last>\w+),\s+(?<first>\w+)" |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
90 // should be perfectly legal, while pcre does not allow the same |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
91 // named token name on both sides of the alternative. Also fix |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
92 // that here by replacing name tokens by dummy names, and dealing |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
93 // with the dummy names later. |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
94 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
95 size_t tmp_pos = pattern.find_first_of ('>', new_pos); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
96 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
97 if (tmp_pos == std::string::npos) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
98 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
99 (*current_liboctave_error_handler) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
100 ("regexp: syntax error in pattern"); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
101 return; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
102 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
103 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
104 std::string tmp_name = |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
105 pattern.substr (new_pos+3, tmp_pos-new_pos-3); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
106 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
107 bool found = false; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
108 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
109 for (int i = 0; i < nnames; i++) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
110 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
111 if (named_pats(i) == tmp_name) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
112 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
113 named_idx.resize (dim_vector (inames+1, 1)); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
114 named_idx(inames) = i; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
115 found = true; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
116 break; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
117 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
118 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
119 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
120 if (! found) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
121 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
122 named_idx.resize (dim_vector (inames+1, 1)); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
123 named_idx(inames) = nnames; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
124 named_pats.append (tmp_name); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
125 nnames++; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
126 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
127 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
128 if (new_pos - pos > 0) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
129 buf << pattern.substr (pos, new_pos-pos); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
130 if (inames < 10) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
131 buf << "(?P<n00" << inames++; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
132 else if (inames < 100) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
133 buf << "(?P<n0" << inames++; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
134 else |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
135 buf << "(?P<n" << inames++; |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
136 |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
137 pos = tmp_pos; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
138 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
139 else if (pattern.at (new_pos + 2) == '<') |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
140 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
141 // Find lookbehind operators of arbitrary length (ie like |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
142 // "(?<=[a-z]*)") and replace with a maximum length operator |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
143 // as PCRE can not yet handle arbitrary length lookahead |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
144 // operators. Use the string length as the maximum length to |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
145 // avoid issues. |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
146 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
147 int brackets = 1; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
148 size_t tmp_pos1 = new_pos + 2; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
149 size_t tmp_pos2 = tmp_pos1; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
150 |
14494
b472596cd7b9
Stop segfault when regular expression contains malformed lookbehind atom (bug #35961)
Rik <octave@nomad.inbox5.com>
parents:
14452
diff
changeset
|
151 while (tmp_pos1 < pattern.length () && brackets > 0) |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
152 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
153 char ch = pattern.at (tmp_pos1); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
154 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
155 if (ch == '(') |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
156 brackets++; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
157 else if (ch == ')') |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
158 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
159 if (brackets > 1) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
160 tmp_pos2 = tmp_pos1; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
161 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
162 brackets--; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
163 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
164 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
165 tmp_pos1++; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
166 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
167 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
168 if (brackets != 0) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
169 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
170 buf << pattern.substr (pos, new_pos - pos) << "(?"; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
171 pos = new_pos + 2; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
172 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
173 else |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
174 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
175 size_t tmp_pos3 = pattern.find_first_of ("*+", tmp_pos2); |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
176 |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
177 if (tmp_pos3 != std::string::npos && tmp_pos3 < tmp_pos1) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
178 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
179 if (!lookbehind_warned) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
180 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
181 lookbehind_warned = true; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
182 (*current_liboctave_warning_handler) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
183 ("%s: arbitrary length lookbehind patterns are only supported up to length %d", |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
184 who.c_str (), MAXLOOKBEHIND); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
185 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
186 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
187 buf << pattern.substr (pos, new_pos - pos) << "("; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
188 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
189 size_t i; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
190 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
191 if (pattern.at (tmp_pos3) == '*') |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
192 i = 0; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
193 else |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
194 i = 1; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
195 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
196 for (; i < max_length + 1; i++) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
197 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
198 buf << pattern.substr (new_pos, tmp_pos3 - new_pos) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
199 << "{" << i << "}"; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
200 buf << pattern.substr (tmp_pos3 + 1, |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
201 tmp_pos1 - tmp_pos3 - 1); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
202 if (i != max_length) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
203 buf << "|"; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
204 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
205 buf << ")"; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
206 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
207 else |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
208 buf << pattern.substr (pos, tmp_pos1 - pos); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
209 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
210 pos = tmp_pos1; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
211 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
212 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
213 else |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
214 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
215 buf << pattern.substr (pos, new_pos - pos) << "(?"; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
216 pos = new_pos + 2; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
217 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
218 |
7784 | 219 } |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11570
diff
changeset
|
220 |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
221 buf << pattern.substr (pos); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
222 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
223 const char *err; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
224 int erroffset; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
225 std::string buf_str = buf.str (); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
226 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
227 int pcre_options |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
228 = ((options.case_insensitive () ? PCRE_CASELESS : 0) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
229 | (options.dotexceptnewline () ? 0 : PCRE_DOTALL) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
230 | (options.lineanchors () ? PCRE_MULTILINE : 0) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
231 | (options.freespacing () ? PCRE_EXTENDED : 0)); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
232 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
233 data = pcre_compile (buf_str.c_str (), pcre_options, &err, &erroffset, 0); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
234 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
235 if (! data) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
236 (*current_liboctave_error_handler) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
237 ("%s: %s at position %d of expression", who.c_str (), |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
238 err, erroffset); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
239 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
240 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
241 regexp::match_data |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
242 regexp::match (const std::string& buffer) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
243 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
244 regexp::match_data retval; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
245 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
246 std::list<regexp::match_element> lst; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
247 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
248 int subpatterns; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
249 int namecount; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
250 int nameentrysize; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
251 char *nametable; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
252 size_t idx = 0; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
253 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
254 pcre *re = static_cast <pcre *> (data); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
255 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
256 pcre_fullinfo (re, 0, PCRE_INFO_CAPTURECOUNT, &subpatterns); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
257 pcre_fullinfo (re, 0, PCRE_INFO_NAMECOUNT, &namecount); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
258 pcre_fullinfo (re, 0, PCRE_INFO_NAMEENTRYSIZE, &nameentrysize); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
259 pcre_fullinfo (re, 0, PCRE_INFO_NAMETABLE, &nametable); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
260 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
261 OCTAVE_LOCAL_BUFFER (int, ovector, (subpatterns+1)*3); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
262 OCTAVE_LOCAL_BUFFER (int, nidx, namecount); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
263 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
264 for (int i = 0; i < namecount; i++) |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
265 { |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
266 // Index of subpattern in first two bytes MSB first of name. |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
267 // Extract index. |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
268 nidx[i] = (static_cast<int> (nametable[i*nameentrysize])) << 8 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
269 | static_cast<int> (nametable[i*nameentrysize+1]); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
270 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
271 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
272 while (true) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
273 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
274 OCTAVE_QUIT; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
275 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
276 int matches = pcre_exec (re, 0, buffer.c_str (), |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
277 buffer.length (), idx, |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
278 (idx ? PCRE_NOTBOL : 0), |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
279 ovector, (subpatterns+1)*3); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
280 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
281 if (matches == PCRE_ERROR_MATCHLIMIT) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
282 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
283 // Try harder; start with default value for MATCH_LIMIT |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
284 // and increase it. |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
285 (*current_liboctave_warning_handler) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
286 ("your pattern caused PCRE to hit its MATCH_LIMIT; trying harder now, but this will be slow"); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
287 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
288 pcre_extra pe; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
289 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
290 pcre_config (PCRE_CONFIG_MATCH_LIMIT, |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
291 static_cast <void *> (&pe.match_limit)); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
292 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
293 pe.flags = PCRE_EXTRA_MATCH_LIMIT; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
294 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
295 int i = 0; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
296 while (matches == PCRE_ERROR_MATCHLIMIT |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
297 && i++ < PCRE_MATCHLIMIT_MAX) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
298 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
299 OCTAVE_QUIT; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
300 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
301 pe.match_limit *= 10; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
302 matches = pcre_exec (re, &pe, buffer.c_str (), |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
303 buffer.length (), idx, |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
304 (idx ? PCRE_NOTBOL : 0), |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
305 ovector, (subpatterns+1)*3); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
306 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
307 } |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
308 |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
309 if (matches < 0 && matches != PCRE_ERROR_NOMATCH) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
310 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
311 (*current_liboctave_error_handler) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
312 ("%s: internal error calling pcre_exec; error code from pcre_exec is %i", |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
313 who.c_str (), matches); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
314 return retval; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
315 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
316 else if (matches == PCRE_ERROR_NOMATCH) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
317 break; |
14536
6d5c951ec520
Add 'emptymatch', 'noemptymatch' options to regular expressions.
Rik <octave@nomad.inbox5.com>
parents:
14506
diff
changeset
|
318 else if (ovector[1] <= ovector[0] && ! options.emptymatch ()) |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
319 { |
14536
6d5c951ec520
Add 'emptymatch', 'noemptymatch' options to regular expressions.
Rik <octave@nomad.inbox5.com>
parents:
14506
diff
changeset
|
320 // Zero length match. Skip to next char. |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
321 idx = ovector[0] + 1; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
322 if (idx < buffer.length ()) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
323 continue; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
324 else |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
325 break; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
326 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
327 else |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
328 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
329 int pos_match = 0; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
330 Matrix token_extents (matches-1, 2); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
331 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
332 for (int i = 1; i < matches; i++) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
333 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
334 if (ovector[2*i] >= 0 && ovector[2*i+1] > 0 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
335 && (i == 1 || ovector[2*i] != ovector[2*i-2] |
14455
c57f9c911053
regexp.cc: Remove redundant condition in if statement
Rik <octave@nomad.inbox5.com>
parents:
14452
diff
changeset
|
336 || ovector[2*i-1] != ovector[2*i+1])) |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
337 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
338 token_extents(pos_match,0) = double (ovector[2*i]+1); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
339 token_extents(pos_match++,1) = double (ovector[2*i+1]); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
340 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
341 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
342 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
343 token_extents.resize (pos_match, 2); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
344 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
345 double start = double (ovector[0]+1); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
346 double end = double (ovector[1]); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
347 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
348 const char **listptr; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
349 int status = pcre_get_substring_list (buffer.c_str (), ovector, |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
350 matches, &listptr); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
351 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
352 if (status == PCRE_ERROR_NOMEMORY) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
353 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
354 (*current_liboctave_error_handler) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
355 ("%s: cannot allocate memory in pcre_get_substring_list", |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
356 who.c_str ()); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
357 return retval; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
358 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
359 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
360 string_vector tokens (pos_match); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
361 string_vector named_tokens (nnames); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
362 int pos_offset = 0; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
363 pos_match = 0; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
364 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
365 for (int i = 1; i < matches; i++) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
366 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
367 if (ovector[2*i] >= 0 && ovector[2*i+1] > 0) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
368 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
369 if (i == 1 || ovector[2*i] != ovector[2*i-2] |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
370 || ovector[2*i-1] != ovector[2*i+1]) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
371 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
372 if (namecount > 0) |
14452
721be41ea988
Stop segfault when using regular expressions with named capture buffers (bug #35683).
Rik <octave@nomad.inbox5.com>
parents:
14138
diff
changeset
|
373 { |
721be41ea988
Stop segfault when using regular expressions with named capture buffers (bug #35683).
Rik <octave@nomad.inbox5.com>
parents:
14138
diff
changeset
|
374 // FIXME: Should probably do this with a map() |
721be41ea988
Stop segfault when using regular expressions with named capture buffers (bug #35683).
Rik <octave@nomad.inbox5.com>
parents:
14138
diff
changeset
|
375 // rather than a linear search. However, |
721be41ea988
Stop segfault when using regular expressions with named capture buffers (bug #35683).
Rik <octave@nomad.inbox5.com>
parents:
14138
diff
changeset
|
376 // the number of captured, named expressions |
721be41ea988
Stop segfault when using regular expressions with named capture buffers (bug #35683).
Rik <octave@nomad.inbox5.com>
parents:
14138
diff
changeset
|
377 // is usually pretty small (< 4) |
721be41ea988
Stop segfault when using regular expressions with named capture buffers (bug #35683).
Rik <octave@nomad.inbox5.com>
parents:
14138
diff
changeset
|
378 for (int j = 0; j < namecount; j++) |
721be41ea988
Stop segfault when using regular expressions with named capture buffers (bug #35683).
Rik <octave@nomad.inbox5.com>
parents:
14138
diff
changeset
|
379 { |
721be41ea988
Stop segfault when using regular expressions with named capture buffers (bug #35683).
Rik <octave@nomad.inbox5.com>
parents:
14138
diff
changeset
|
380 if (nidx[j] == i) |
721be41ea988
Stop segfault when using regular expressions with named capture buffers (bug #35683).
Rik <octave@nomad.inbox5.com>
parents:
14138
diff
changeset
|
381 { |
721be41ea988
Stop segfault when using regular expressions with named capture buffers (bug #35683).
Rik <octave@nomad.inbox5.com>
parents:
14138
diff
changeset
|
382 named_tokens(named_idx(j)) = |
721be41ea988
Stop segfault when using regular expressions with named capture buffers (bug #35683).
Rik <octave@nomad.inbox5.com>
parents:
14138
diff
changeset
|
383 std::string (*(listptr+i-pos_offset)); |
721be41ea988
Stop segfault when using regular expressions with named capture buffers (bug #35683).
Rik <octave@nomad.inbox5.com>
parents:
14138
diff
changeset
|
384 break; |
721be41ea988
Stop segfault when using regular expressions with named capture buffers (bug #35683).
Rik <octave@nomad.inbox5.com>
parents:
14138
diff
changeset
|
385 } |
721be41ea988
Stop segfault when using regular expressions with named capture buffers (bug #35683).
Rik <octave@nomad.inbox5.com>
parents:
14138
diff
changeset
|
386 } |
721be41ea988
Stop segfault when using regular expressions with named capture buffers (bug #35683).
Rik <octave@nomad.inbox5.com>
parents:
14138
diff
changeset
|
387 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
388 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
389 tokens(pos_match++) = std::string (*(listptr+i)); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
390 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
391 else |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
392 pos_offset++; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
393 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
394 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
395 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
396 std::string match_string = std::string (*listptr); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
397 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
398 pcre_free_substring_list (listptr); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
399 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
400 regexp::match_element new_elem (named_tokens, tokens, match_string, |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
401 token_extents, start, end); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
402 lst.push_back (new_elem); |
14536
6d5c951ec520
Add 'emptymatch', 'noemptymatch' options to regular expressions.
Rik <octave@nomad.inbox5.com>
parents:
14506
diff
changeset
|
403 |
6d5c951ec520
Add 'emptymatch', 'noemptymatch' options to regular expressions.
Rik <octave@nomad.inbox5.com>
parents:
14506
diff
changeset
|
404 if (ovector[1] <= ovector[0]) |
6d5c951ec520
Add 'emptymatch', 'noemptymatch' options to regular expressions.
Rik <octave@nomad.inbox5.com>
parents:
14506
diff
changeset
|
405 { |
6d5c951ec520
Add 'emptymatch', 'noemptymatch' options to regular expressions.
Rik <octave@nomad.inbox5.com>
parents:
14506
diff
changeset
|
406 // Zero length match. Skip to next char. |
6d5c951ec520
Add 'emptymatch', 'noemptymatch' options to regular expressions.
Rik <octave@nomad.inbox5.com>
parents:
14506
diff
changeset
|
407 idx = ovector[0] + 1; |
6d5c951ec520
Add 'emptymatch', 'noemptymatch' options to regular expressions.
Rik <octave@nomad.inbox5.com>
parents:
14506
diff
changeset
|
408 if (idx <= buffer.length ()) |
6d5c951ec520
Add 'emptymatch', 'noemptymatch' options to regular expressions.
Rik <octave@nomad.inbox5.com>
parents:
14506
diff
changeset
|
409 continue; |
6d5c951ec520
Add 'emptymatch', 'noemptymatch' options to regular expressions.
Rik <octave@nomad.inbox5.com>
parents:
14506
diff
changeset
|
410 } |
6d5c951ec520
Add 'emptymatch', 'noemptymatch' options to regular expressions.
Rik <octave@nomad.inbox5.com>
parents:
14506
diff
changeset
|
411 else |
6d5c951ec520
Add 'emptymatch', 'noemptymatch' options to regular expressions.
Rik <octave@nomad.inbox5.com>
parents:
14506
diff
changeset
|
412 idx = ovector[1]; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
413 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
414 if (options.once () || idx >= buffer.length ()) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
415 break; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
416 } |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
417 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
418 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
419 retval = regexp::match_data (lst, named_pats); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
420 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
421 return retval; |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
422 } |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
423 |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
424 bool |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
425 regexp::is_match (const std::string& buffer) |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
426 { |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
427 regexp::match_data rx_lst = match (buffer); |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
428 |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
429 regexp::match_data::const_iterator p = rx_lst.begin (); |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
430 |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
431 std::string match_string = p->match_string (); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
432 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
433 return ! match_string.empty (); |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
434 } |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
435 |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
436 Array<bool> |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
437 regexp::is_match (const string_vector& buffer) |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
438 { |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
439 octave_idx_type len = buffer.length (); |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
440 |
14057
7e60182e6a1c
avoid deprecated Array constructor
John W. Eaton <jwe@octave.org>
parents:
14025
diff
changeset
|
441 Array<bool> retval (dim_vector (len, 1)); |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
442 |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
443 for (octave_idx_type i = 0; i < buffer.length (); i++) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
444 retval(i) = is_match (buffer(i)); |
7779
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
445 |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
446 return retval; |
791231dac333
Add regexp matching to Fwho and Fclear
David Bateman <dbateman@free.fr>
parents:
diff
changeset
|
447 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
448 |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
449 // Declare rep_token_t used in processing replacement string |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
450 typedef struct |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
451 { |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
452 size_t pos; |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
453 int num; |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
454 } rep_token_t; |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
455 |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
456 |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
457 std::string |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
458 regexp::replace (const std::string& buffer, const std::string& replacement) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
459 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
460 std::string retval; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
461 |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
462 regexp::match_data rx_lst = match (buffer); |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
463 |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
464 size_t num_matches = rx_lst.size (); |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
465 |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
466 if (num_matches == 0) |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
467 { |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
468 retval = buffer; |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
469 return retval; |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
470 } |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
471 |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
472 // Identify replacement tokens; build a vector of group numbers in |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
473 // the replacement string so that we can quickly calculate the size |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
474 // of the replacement. |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
475 |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
476 // FIXME: All code assumes that only 10 tokens ($0-$9) exist. |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
477 // $11 represents $1 followed by the character '1' rather than |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
478 // the eleventh capture buffer. |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
479 |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
480 std::string repstr = replacement; |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
481 std::vector<rep_token_t> tokens; |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
482 tokens.reserve (5); // Reserve memory for 5 pattern replacements |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
483 |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
484 for (size_t i=0; i < repstr.size (); i++) |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
485 { |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
486 if (repstr[i] == '\\') |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
487 { |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
488 if (i < repstr.size () - 1 && repstr[i+1] == '$') |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
489 { |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
490 repstr.erase (i,1); // erase backslash |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
491 i++; // skip over '$' |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
492 continue; |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
493 } |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
494 if (i < repstr.size () - 1 && repstr[i+1] == '\\') |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
495 { |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
496 repstr.erase (i,1); // erase 1st backslash |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
497 continue; |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
498 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
499 } |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
500 else if (repstr[i] == '$') |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
501 { |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
502 if (i < repstr.size () - 1 && isdigit (repstr[i+1])) |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
503 { |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
504 rep_token_t tmp_token; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
505 |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
506 tmp_token.pos = i; |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
507 tmp_token.num = repstr[i+1]-'0'; |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
508 tokens.push_back (tmp_token); |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
509 } |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
510 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
511 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
512 |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
513 std::string rep; |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
514 int num_tokens = tokens.size (); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
515 |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
516 if (num_tokens > 0) |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
517 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
518 // Determine replacement length |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
519 const size_t replen = repstr.size () - 2*num_tokens; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
520 int delta = 0; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
521 regexp::match_data::const_iterator p = rx_lst.begin (); |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
522 for (size_t i = 0; i < num_matches; i++) |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
523 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
524 OCTAVE_QUIT; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
525 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
526 double start = p->start (); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
527 double end = p->end (); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
528 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
529 const Matrix pairs (p->token_extents ()); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
530 size_t pairlen = 0; |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
531 for (int j = 0; j < num_tokens; j++) |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
532 { |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
533 if (tokens[j].num == 0) |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
534 pairlen += static_cast<size_t> (end - start) + 1; |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
535 else if (tokens[j].num <= pairs.rows ()) |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
536 pairlen += static_cast<size_t> (pairs(tokens[j].num-1,1) |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
537 - pairs(tokens[j].num-1,0)) + 1; |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
538 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
539 delta += (static_cast<int> (replen + pairlen) |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
540 - static_cast<int> (end - start + 1)); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
541 p++; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
542 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
543 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
544 // Build replacement string |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
545 rep.reserve (buffer.size () + delta); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
546 size_t from = 0; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
547 p = rx_lst.begin (); |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
548 for (size_t i = 0; i < num_matches; i++) |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
549 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
550 OCTAVE_QUIT; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
551 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
552 double start = p->start (); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
553 double end = p->end (); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
554 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
555 const Matrix pairs (p->token_extents ()); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
556 rep.append (&buffer[from], static_cast<size_t> (start - 1) - from); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
557 from = static_cast<size_t> (end - 1) + 1; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
558 |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
559 size_t cur_pos = 0; |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
560 |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
561 for (int j = 0; j < num_tokens; j++) |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
562 { |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
563 rep.append (&repstr[cur_pos], (tokens[j].pos) - cur_pos); |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
564 cur_pos = tokens[j].pos+2; |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
565 |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
566 int k = tokens[j].num; |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
567 if (k == 0) |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
568 { |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
569 // replace with entire match |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
570 rep.append (&buffer[static_cast<size_t> (end - 1)], |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
571 static_cast<size_t> (end - start) + 1); |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
572 } |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
573 else if (k <= pairs.rows ()) |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
574 { |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
575 // replace with group capture |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
576 rep.append (&buffer[static_cast<size_t> (pairs(k-1,0)-1)], |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
577 static_cast<size_t> (pairs(k-1,1) |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
578 - pairs(k-1,0)) + 1); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
579 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
580 else |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
581 { |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
582 // replace with nothing |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
583 } |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
584 } |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
585 if (cur_pos < repstr.size ()) |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
586 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
|
587 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
588 p++; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
589 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
590 rep.append (&buffer[from], buffer.size () - from); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
591 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
592 else |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
593 { |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
594 // Determine repstr length |
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
595 const size_t replen = repstr.size (); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
596 int delta = 0; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
597 regexp::match_data::const_iterator p = rx_lst.begin (); |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
598 for (size_t i = 0; i < num_matches; i++) |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
599 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
600 OCTAVE_QUIT; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
601 delta += static_cast<int> (replen) |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
602 - static_cast<int> (p->end () - p->start () + 1); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
603 p++; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
604 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
605 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
606 // Build replacement string |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
607 rep.reserve (buffer.size () + delta); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
608 size_t from = 0; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
609 p = rx_lst.begin (); |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
610 for (size_t i = 0; i < num_matches; i++) |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
611 { |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
612 OCTAVE_QUIT; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
613 rep.append (&buffer[from], |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
614 static_cast<size_t> (p->start () - 1) - from); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
615 from = static_cast<size_t> (p->end () - 1) + 1; |
15541
9db32cabeacf
Fix backslash handling in regexp pattern (Bug #37092)
Rik <rik@octave.org>
parents:
15271
diff
changeset
|
616 rep.append (repstr); |
14024
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
617 p++; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
618 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
619 rep.append (&buffer[from], buffer.size () - from); |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
620 } |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
621 |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
622 retval = rep; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
623 return retval; |
fc9f204faea0
refactor regexp (bug #34440)
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
624 } |