Mercurial > hg > octave-nkf
annotate src/DLD-FUNCTIONS/regexp.cc @ 12464:dfeea9cae79e
require PCRE to build Octave
author | John W. Eaton <jwe@octave.org> |
---|---|
date | Mon, 21 Feb 2011 03:50:59 -0500 |
parents | e4dbfe3019b1 |
children | f96b9b9f141b |
rev | line source |
---|---|
5582 | 1 /* |
2 | |
11523 | 3 Copyright (C) 2005-2011 David Bateman |
4 Copyright (C) 2002-2005 Paul Kienzle | |
7016 | 5 |
6 This file is part of Octave. | |
5582 | 7 |
8 Octave is free software; you can redistribute it and/or modify it | |
9 under the terms of the GNU General Public License as published by the | |
7016 | 10 Free Software Foundation; either version 3 of the License, or (at your |
11 option) any later version. | |
5582 | 12 |
13 Octave is distributed in the hope that it will be useful, but WITHOUT | |
14 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or | |
15 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License | |
16 for more details. | |
17 | |
18 You should have received a copy of the GNU General Public License | |
7016 | 19 along with Octave; see the file COPYING. If not, see |
20 <http://www.gnu.org/licenses/>. | |
5582 | 21 |
22 */ | |
23 | |
24 #ifdef HAVE_CONFIG_H | |
25 #include <config.h> | |
26 #endif | |
27 | |
5773 | 28 #include <algorithm> |
5765 | 29 #include <sstream> |
30 | |
5582 | 31 #include "defun-dld.h" |
32 #include "error.h" | |
33 #include "gripes.h" | |
34 #include "oct-obj.h" | |
35 #include "utils.h" | |
36 | |
37 #include "Cell.h" | |
38 #include "oct-map.h" | |
39 #include "str-vec.h" | |
5785 | 40 #include "quit.h" |
41 #include "parse.h" | |
8377
25bc2d31e1bf
improve OCTAVE_LOCAL_BUFFER
Jaroslav Hajek <highegg@gmail.com>
parents:
8140
diff
changeset
|
42 #include "oct-locbuf.h" |
5582 | 43 |
44 #include <pcre.h> | |
45 | |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
46 // Define the maximum number of retries for a pattern that |
8140
cdd05e46f6c9
Increase pcre's match_limit for difficult regexps
Thomas Weber <thomas.weber.mail@gmail.com>
parents:
8093
diff
changeset
|
47 // possibly results in an infinite recursion. |
cdd05e46f6c9
Increase pcre's match_limit for difficult regexps
Thomas Weber <thomas.weber.mail@gmail.com>
parents:
8093
diff
changeset
|
48 #define PCRE_MATCHLIMIT_MAX 10 |
cdd05e46f6c9
Increase pcre's match_limit for difficult regexps
Thomas Weber <thomas.weber.mail@gmail.com>
parents:
8093
diff
changeset
|
49 |
5785 | 50 // The regexp is constructed as a linked list to avoid resizing the |
51 // return values in arrays at each new match. | |
52 | |
53 // FIXME don't bother collecting and composing return values the user | |
54 // doesn't want. | |
55 | |
56 class regexp_elem | |
5582 | 57 { |
5785 | 58 public: |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
59 regexp_elem (const string_vector& _named_token, const Cell& _t, |
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
60 const std::string& _m, const Matrix& _te, double _s, |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
61 double _e) : |
5785 | 62 named_token (_named_token), t (_t), m (_m), te (_te), s (_s), e (_e) { } |
63 | |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
64 regexp_elem (const regexp_elem &a) : named_token (a.named_token), t (a.t), |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
65 m (a.m), te (a.te), s (a.s), e (a.e) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
66 { } |
5785 | 67 |
68 string_vector named_token; | |
69 Cell t; | |
70 std::string m; | |
71 Matrix te; | |
72 double s; | |
73 double e; | |
74 }; | |
75 | |
76 typedef std::list<regexp_elem>::const_iterator const_iterator; | |
77 | |
8093
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
78 #define MAXLOOKBEHIND 10 |
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
79 static bool lookbehind_warned = false; |
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
80 |
5785 | 81 static int |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
82 octregexp_list (const octave_value_list &args, const std::string &nm, |
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
83 bool case_insensitive, std::list<regexp_elem> &lst, |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
84 string_vector &named, int &nopts, bool &once) |
5785 | 85 { |
86 int sz = 0; | |
12464
dfeea9cae79e
require PCRE to build Octave
John W. Eaton <jwe@octave.org>
parents:
12462
diff
changeset
|
87 |
5582 | 88 int nargin = args.length(); |
5779 | 89 bool lineanchors = false; |
90 bool dotexceptnewline = false; | |
91 bool freespacing = false; | |
5582 | 92 |
5785 | 93 nopts = nargin - 2; |
7893
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
94 once = false; |
5785 | 95 |
5582 | 96 std::string buffer = args(0).string_value (); |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
97 size_t max_length = (buffer.length () > MAXLOOKBEHIND ? |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
98 MAXLOOKBEHIND: buffer.length ()); |
8093
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
99 |
5582 | 100 if (error_state) |
101 { | |
102 gripe_wrong_type_arg (nm.c_str(), args(0)); | |
5785 | 103 return 0; |
5582 | 104 } |
105 | |
106 std::string pattern = args(1).string_value (); | |
107 if (error_state) | |
108 { | |
109 gripe_wrong_type_arg (nm.c_str(), args(1)); | |
5785 | 110 return 0; |
5582 | 111 } |
112 | |
113 for (int i = 2; i < nargin; i++) | |
114 { | |
115 std::string str = args(i).string_value(); | |
116 if (error_state) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
117 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
118 error ("%s: optional arguments must be strings", nm.c_str()); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
119 break; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
120 } |
5582 | 121 std::transform (str.begin (), str.end (), str.begin (), tolower); |
122 if (str.find("once", 0) == 0) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
123 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
124 once = true; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
125 nopts--; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
126 } |
5779 | 127 else if (str.find("matchcase", 0) == 0) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
128 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
129 case_insensitive = false; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
130 nopts--; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
131 } |
5779 | 132 else if (str.find("ignorecase", 0) == 0) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
133 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
134 case_insensitive = true; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
135 nopts--; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
136 } |
5785 | 137 else if (str.find("dotall", 0) == 0) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
138 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
139 dotexceptnewline = false; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
140 nopts--; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
141 } |
5785 | 142 else if (str.find("stringanchors", 0) == 0) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
143 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
144 lineanchors = false; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
145 nopts--; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
146 } |
5779 | 147 else if (str.find("literalspacing", 0) == 0) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
148 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
149 freespacing = false; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
150 nopts--; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
151 } |
12464
dfeea9cae79e
require PCRE to build Octave
John W. Eaton <jwe@octave.org>
parents:
12462
diff
changeset
|
152 |
5785 | 153 // Only accept these options with pcre |
154 else if (str.find("dotexceptnewline", 0) == 0) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
155 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
156 dotexceptnewline = true; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
157 nopts--; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
158 } |
5785 | 159 else if (str.find("lineanchors", 0) == 0) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
160 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
161 lineanchors = true; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
162 nopts--; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
163 } |
5785 | 164 else if (str.find("freespacing", 0) == 0) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
165 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
166 freespacing = true; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
167 nopts--; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
168 } |
5582 | 169 else if (str.find("start", 0) && str.find("end", 0) && |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
170 str.find("tokenextents", 0) && str.find("match", 0) && |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
171 str.find("tokens", 0) && str.find("names", 0)) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
172 error ("%s: unrecognized option", nm.c_str()); |
5582 | 173 } |
174 | |
175 if (!error_state) | |
176 { | |
5785 | 177 Cell t; |
178 std::string m; | |
179 double s, e; | |
5582 | 180 |
181 // named tokens "(?<name>...)" are only treated with PCRE not regex. | |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
182 |
5619 | 183 size_t pos = 0; |
184 size_t new_pos; | |
185 int nnames = 0; | |
186 int inames = 0; | |
5765 | 187 std::ostringstream buf; |
5619 | 188 Array<int> named_idx; |
5582 | 189 |
8093
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
190 while ((new_pos = pattern.find ("(?",pos)) != std::string::npos) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
191 { |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
192 if (pattern.at (new_pos + 2) == '<' && |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
193 !(pattern.at (new_pos + 3) == '=' || |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
194 pattern.at (new_pos + 3) == '!')) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
195 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
196 // The syntax of named tokens in pcre is "(?P<name>...)" while |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
197 // we need a syntax "(?<name>...)", so fix that here. Also an |
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
198 // expression like |
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
199 // "(?<first>\w+)\s+(?<last>\w+)|(?<last>\w+),\s+(?<first>\w+)" |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
200 // should be perfectly legal, while pcre does not allow the same |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
201 // named token name on both sides of the alternative. Also fix |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
202 // that here by replacing name tokens by dummy names, and dealing |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
203 // with the dummy names later. |
8093
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
204 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
205 size_t tmp_pos = pattern.find_first_of ('>',new_pos); |
8093
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
206 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
207 if (tmp_pos == std::string::npos) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
208 { |
11553
01f703952eff
Improve docstrings for functions in DLD-FUNCTIONS directory.
Rik <octave@nomad.inbox5.com>
parents:
11523
diff
changeset
|
209 error ("regexp: syntax error in pattern"); |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
210 break; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
211 } |
8093
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
212 |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
213 std::string tmp_name = |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
214 pattern.substr(new_pos+3,tmp_pos-new_pos-3); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
215 bool found = false; |
8093
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
216 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
217 for (int i = 0; i < nnames; i++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
218 if (named(i) == tmp_name) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
219 { |
12433
02669a1aa070
regexp.cc: avoid deprecated Array<T>:resize function
John W. Eaton <jwe@octave.org>
parents:
11590
diff
changeset
|
220 named_idx.resize (dim_vector (inames+1, 1)); |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
221 named_idx(inames) = i; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
222 found = true; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
223 break; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
224 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
225 if (! found) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
226 { |
12433
02669a1aa070
regexp.cc: avoid deprecated Array<T>:resize function
John W. Eaton <jwe@octave.org>
parents:
11590
diff
changeset
|
227 named_idx.resize (dim_vector (inames+1, 1)); |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
228 named_idx(inames) = nnames; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
229 named.append(tmp_name); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
230 nnames++; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
231 } |
8093
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
232 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
233 if (new_pos - pos > 0) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
234 buf << pattern.substr(pos,new_pos-pos); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
235 if (inames < 10) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
236 buf << "(?P<n00" << inames++; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
237 else if (inames < 100) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
238 buf << "(?P<n0" << inames++; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
239 else |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
240 buf << "(?P<n" << inames++; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
241 pos = tmp_pos; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
242 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
243 else if (pattern.at (new_pos + 2) == '<') |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
244 { |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
245 // Find lookbehind operators of arbitrary length (ie like |
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
246 // "(?<=[a-z]*)") and replace with a maximum length operator |
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
247 // as PCRE can not yet handle arbitrary length lookahead |
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
248 // operators. Use the string length as the maximum length to |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
249 // avoid issues. |
5582 | 250 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
251 int brackets = 1; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
252 size_t tmp_pos1 = new_pos + 2; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
253 size_t tmp_pos2 = tmp_pos1; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
254 while (tmp_pos1 <= pattern.length () && brackets > 0) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
255 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
256 char ch = pattern.at (tmp_pos1); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
257 if (ch == '(') |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
258 brackets++; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
259 else if (ch == ')') |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
260 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
261 if (brackets > 1) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
262 tmp_pos2 = tmp_pos1; |
8093
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
263 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
264 brackets--; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
265 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
266 tmp_pos1++; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
267 } |
8093
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
268 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
269 if (brackets != 0) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
270 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
271 buf << pattern.substr (pos, new_pos - pos) << "(?"; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
272 pos = new_pos + 2; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
273 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
274 else |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
275 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
276 size_t tmp_pos3 = pattern.find_first_of ("*+", tmp_pos2); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
277 if (tmp_pos3 != std::string::npos && tmp_pos3 < tmp_pos1) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
278 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
279 if (!lookbehind_warned) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
280 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
281 lookbehind_warned = true; |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
282 warning ("%s: arbitrary length lookbehind patterns are only supported up to length %d", nm.c_str(), MAXLOOKBEHIND); |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
283 } |
8093
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
284 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
285 buf << pattern.substr (pos, new_pos - pos) << "("; |
8093
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
286 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
287 size_t i; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
288 if (pattern.at (tmp_pos3) == '*') |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
289 i = 0; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
290 else |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
291 i = 1; |
8093
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
292 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
293 for (; i < max_length + 1; i++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
294 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
295 buf << pattern.substr(new_pos, tmp_pos3 - new_pos) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
296 << "{" << i << "}"; |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
297 buf << pattern.substr(tmp_pos3 + 1, |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
298 tmp_pos1 - tmp_pos3 - 1); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
299 if (i != max_length) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
300 buf << "|"; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
301 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
302 buf << ")"; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
303 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
304 else |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
305 buf << pattern.substr (pos, tmp_pos1 - pos); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
306 pos = tmp_pos1; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
307 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
308 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
309 else |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
310 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
311 buf << pattern.substr (pos, new_pos - pos) << "(?"; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
312 pos = new_pos + 2; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
313 } |
5619 | 314 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
315 } |
5619 | 316 |
5765 | 317 buf << pattern.substr(pos); |
5619 | 318 |
319 if (error_state) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
320 return 0; |
5582 | 321 |
322 // Compile expression | |
323 pcre *re; | |
324 const char *err; | |
325 int erroffset; | |
5765 | 326 std::string buf_str = buf.str (); |
327 re = pcre_compile (buf_str.c_str (), | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
328 (case_insensitive ? PCRE_CASELESS : 0) | |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
329 (dotexceptnewline ? 0 : PCRE_DOTALL) | |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
330 (lineanchors ? PCRE_MULTILINE : 0) | |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
331 (freespacing ? PCRE_EXTENDED : 0), |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
332 &err, &erroffset, 0); |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
333 |
10550 | 334 if (re == 0) |
335 { | |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
336 error("%s: %s at position %d of expression", nm.c_str(), |
10550 | 337 err, erroffset); |
338 return 0; | |
339 } | |
5582 | 340 |
341 int subpatterns; | |
342 int namecount; | |
343 int nameentrysize; | |
344 char *nametable; | |
345 int idx = 0; | |
346 | |
7520 | 347 pcre_fullinfo(re, 0, PCRE_INFO_CAPTURECOUNT, &subpatterns); |
348 pcre_fullinfo(re, 0, PCRE_INFO_NAMECOUNT, &namecount); | |
349 pcre_fullinfo(re, 0, PCRE_INFO_NAMEENTRYSIZE, &nameentrysize); | |
350 pcre_fullinfo(re, 0, PCRE_INFO_NAMETABLE, &nametable); | |
5582 | 351 |
352 OCTAVE_LOCAL_BUFFER(int, ovector, (subpatterns+1)*3); | |
353 OCTAVE_LOCAL_BUFFER(int, nidx, namecount); | |
354 | |
355 for (int i = 0; i < namecount; i++) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
356 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
357 // Index of subpattern in first two bytes MSB first of name. |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
358 // Extract index. |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
359 nidx[i] = (static_cast<int>(nametable[i*nameentrysize])) << 8 | |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
360 static_cast<int>(nametable[i*nameentrysize+1]); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
361 } |
5582 | 362 |
363 while(true) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
364 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
365 OCTAVE_QUIT; |
5785 | 366 |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
367 int matches = pcre_exec(re, 0, buffer.c_str(), |
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
368 buffer.length(), idx, |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
369 (idx ? PCRE_NOTBOL : 0), |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
370 ovector, (subpatterns+1)*3); |
5582 | 371 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
372 if (matches == PCRE_ERROR_MATCHLIMIT) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
373 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
374 // try harder; start with default value for MATCH_LIMIT and increase it |
11590
4ced6b90fffb
style fixes for warning and error messages in source files
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
375 warning ("your pattern caused PCRE to hit its MATCH_LIMIT; trying harder now, but this will be slow"); |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
376 pcre_extra pe; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
377 pcre_config(PCRE_CONFIG_MATCH_LIMIT, static_cast <void *> (&pe.match_limit)); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
378 pe.flags = PCRE_EXTRA_MATCH_LIMIT; |
8140
cdd05e46f6c9
Increase pcre's match_limit for difficult regexps
Thomas Weber <thomas.weber.mail@gmail.com>
parents:
8093
diff
changeset
|
379 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
380 int i = 0; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
381 while (matches == PCRE_ERROR_MATCHLIMIT && |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
382 i++ < PCRE_MATCHLIMIT_MAX) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
383 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
384 OCTAVE_QUIT; |
8140
cdd05e46f6c9
Increase pcre's match_limit for difficult regexps
Thomas Weber <thomas.weber.mail@gmail.com>
parents:
8093
diff
changeset
|
385 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
386 pe.match_limit *= 10; |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
387 matches = pcre_exec(re, &pe, buffer.c_str(), |
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
388 buffer.length(), idx, |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
389 (idx ? PCRE_NOTBOL : 0), |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
390 ovector, (subpatterns+1)*3); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
391 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
392 } |
8140
cdd05e46f6c9
Increase pcre's match_limit for difficult regexps
Thomas Weber <thomas.weber.mail@gmail.com>
parents:
8093
diff
changeset
|
393 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
394 if (matches < 0 && matches != PCRE_ERROR_NOMATCH) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
395 { |
11590
4ced6b90fffb
style fixes for warning and error messages in source files
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
396 error ("%s: internal error calling pcre_exec; error code from pcre_exec is %i", |
4ced6b90fffb
style fixes for warning and error messages in source files
John W. Eaton <jwe@octave.org>
parents:
11586
diff
changeset
|
397 nm.c_str(), matches); |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
398 pcre_free(re); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
399 return 0; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
400 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
401 else if (matches == PCRE_ERROR_NOMATCH) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
402 break; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
403 else if (ovector[1] <= ovector[0]) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
404 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
405 // FIXME: Zero sized match!! Is this the right thing to do? |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
406 idx = ovector[0] + 1; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
407 continue; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
408 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
409 else |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
410 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
411 int pos_match = 0; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
412 Matrix te(matches-1,2); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
413 for (int i = 1; i < matches; i++) |
10518
fcafe0e9bd58
Handle repeated matches in matches returned by pcre
David Bateman <dbateman@free.fr>
parents:
10504
diff
changeset
|
414 if (ovector[2*i] >= 0 && ovector[2*i+1] > 0) |
fcafe0e9bd58
Handle repeated matches in matches returned by pcre
David Bateman <dbateman@free.fr>
parents:
10504
diff
changeset
|
415 if (i == 1 || ovector[2*i] != ovector[2*i-2] |
fcafe0e9bd58
Handle repeated matches in matches returned by pcre
David Bateman <dbateman@free.fr>
parents:
10504
diff
changeset
|
416 || ovector[2*i-1] != ovector[2*i+1]) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
417 { |
10518
fcafe0e9bd58
Handle repeated matches in matches returned by pcre
David Bateman <dbateman@free.fr>
parents:
10504
diff
changeset
|
418 if (ovector[2*i] >= 0 && ovector[2*i+1] > 0) |
fcafe0e9bd58
Handle repeated matches in matches returned by pcre
David Bateman <dbateman@free.fr>
parents:
10504
diff
changeset
|
419 { |
fcafe0e9bd58
Handle repeated matches in matches returned by pcre
David Bateman <dbateman@free.fr>
parents:
10504
diff
changeset
|
420 te(pos_match,0) = double (ovector[2*i]+1); |
fcafe0e9bd58
Handle repeated matches in matches returned by pcre
David Bateman <dbateman@free.fr>
parents:
10504
diff
changeset
|
421 te(pos_match++,1) = double (ovector[2*i+1]); |
fcafe0e9bd58
Handle repeated matches in matches returned by pcre
David Bateman <dbateman@free.fr>
parents:
10504
diff
changeset
|
422 } |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
423 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
424 te.resize(pos_match,2); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
425 s = double (ovector[0]+1); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
426 e = double (ovector[1]); |
5582 | 427 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
428 const char **listptr; |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
429 int status = pcre_get_substring_list(buffer.c_str(), ovector, |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
430 matches, &listptr); |
5582 | 431 |
10550 | 432 if (status == PCRE_ERROR_NOMEMORY) |
433 { | |
434 error("%s: cannot allocate memory in pcre_get_substring_list", | |
435 nm.c_str()); | |
436 pcre_free(re); | |
437 return 0; | |
438 } | |
5582 | 439 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
440 Cell cell_t (dim_vector(1,pos_match)); |
10518
fcafe0e9bd58
Handle repeated matches in matches returned by pcre
David Bateman <dbateman@free.fr>
parents:
10504
diff
changeset
|
441 string_vector named_tokens(nnames); |
fcafe0e9bd58
Handle repeated matches in matches returned by pcre
David Bateman <dbateman@free.fr>
parents:
10504
diff
changeset
|
442 int pos_offset = 0; |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
443 pos_match = 0; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
444 for (int i = 1; i < matches; i++) |
10550 | 445 if (ovector[2*i] >= 0 && ovector[2*i+1] > 0) |
446 { | |
447 if (i == 1 || ovector[2*i] != ovector[2*i-2] | |
448 || ovector[2*i-1] != ovector[2*i+1]) | |
449 { | |
450 if (namecount > 0) | |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
451 named_tokens(named_idx(i-pos_offset-1)) = |
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
452 std::string(*(listptr+nidx[i-pos_offset-1])); |
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
453 cell_t(pos_match++) = |
10550 | 454 std::string(*(listptr+i)); |
455 } | |
456 else | |
457 pos_offset++; | |
10546
f5b8b28917a2
Eliminate compile warning about explicit braces
Rik <code@nomad.inbox5.com>
parents:
10518
diff
changeset
|
458 } |
5582 | 459 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
460 m = std::string(*listptr); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
461 t = cell_t; |
5785 | 462 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
463 pcre_free_substring_list(listptr); |
5582 | 464 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
465 regexp_elem new_elem (named_tokens, t, m, te, s, e); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
466 lst.push_back (new_elem); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
467 idx = ovector[1]; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
468 sz++; |
5785 | 469 |
11002
2538d03489cc
avoid infinite loop in regexp searches with PCRE
John W. Eaton <jwe@octave.org>
parents:
10846
diff
changeset
|
470 if (once || idx >= buffer.length ()) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
471 break; |
5582 | 472 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
473 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
474 } |
5582 | 475 |
476 pcre_free(re); | |
12464
dfeea9cae79e
require PCRE to build Octave
John W. Eaton <jwe@octave.org>
parents:
12462
diff
changeset
|
477 } |
8619
930a8114197b
For zero length matches in regexp, advance index by one and try again
David Bateman <dbateman@free.fr>
parents:
8477
diff
changeset
|
478 |
5785 | 479 return sz; |
480 } | |
5582 | 481 |
5785 | 482 static octave_value_list |
483 octregexp (const octave_value_list &args, int nargout, const std::string &nm, | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
484 bool case_insensitive) |
5785 | 485 { |
486 octave_value_list retval; | |
487 int nargin = args.length(); | |
488 std::list<regexp_elem> lst; | |
489 string_vector named; | |
490 int nopts; | |
7893
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
491 bool once; |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
492 int sz = octregexp_list (args, nm, case_insensitive, lst, named, nopts, once); |
5785 | 493 |
494 if (! error_state) | |
495 { | |
496 // Converted the linked list in the correct form for the return values | |
497 | |
498 octave_idx_type i = 0; | |
11045
cc3aad9dd3ef
dispatch.cc, fltk_backend.cc, regexp.cc: use octave_scalar_map instead of Octave_map
John W. Eaton <jwe@octave.org>
parents:
11032
diff
changeset
|
499 octave_scalar_map nmap; |
12464
dfeea9cae79e
require PCRE to build Octave
John W. Eaton <jwe@octave.org>
parents:
12462
diff
changeset
|
500 |
5785 | 501 if (sz == 1) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
502 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
503 for (int j = 0; j < named.length(); j++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
504 nmap.assign (named(j), lst.begin()->named_token(j)); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
505 retval(5) = nmap; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
506 } |
5785 | 507 else |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
508 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
509 for (int j = 0; j < named.length (); j++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
510 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
511 i = 0; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
512 Cell tmp(dim_vector (1, sz)); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
513 for (const_iterator p = lst.begin(); p != lst.end(); p++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
514 tmp(i++) = p->named_token(j); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
515 nmap.assign (named(j), octave_value (tmp)); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
516 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
517 retval(5) = nmap; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
518 } |
5785 | 519 |
7893
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
520 if (once) |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
521 retval(4) = sz ? lst.front ().t : Cell(); |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
522 else |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
523 { |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
524 Cell t (dim_vector(1, sz)); |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
525 i = 0; |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
526 for (const_iterator p = lst.begin(); p != lst.end(); p++) |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
527 t(i++) = p->t; |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
528 retval(4) = t; |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
529 } |
5785 | 530 |
7893
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
531 if (once) |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
532 retval(3) = sz ? lst.front ().m : std::string(); |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
533 else |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
534 { |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
535 Cell m (dim_vector(1, sz)); |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
536 i = 0; |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
537 for (const_iterator p = lst.begin(); p != lst.end(); p++) |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
538 m(i++) = p->m; |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
539 retval(3) = m; |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
540 } |
5785 | 541 |
7893
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
542 if (once) |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
543 retval(2) = sz ? lst.front ().te : Matrix(); |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
544 else |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
545 { |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
546 Cell te (dim_vector(1, sz)); |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
547 i = 0; |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
548 for (const_iterator p = lst.begin(); p != lst.end(); p++) |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
549 te(i++) = p->te; |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
550 retval(2) = te; |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
551 } |
5785 | 552 |
7893
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
553 if (once) |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
554 { |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
555 if (sz) |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
556 retval(1) = lst.front ().e; |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
557 else |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
558 retval(1) = Matrix(); |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
559 } |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
560 else |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
561 { |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
562 NDArray e (dim_vector(1, sz)); |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
563 i = 0; |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
564 for (const_iterator p = lst.begin(); p != lst.end(); p++) |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
565 e(i++) = p->e; |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
566 retval(1) = e; |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
567 } |
5785 | 568 |
7893
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
569 if (once) |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
570 { |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
571 if (sz) |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
572 retval(0) = lst.front ().s; |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
573 else |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
574 retval(0) = Matrix(); |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
575 } |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
576 else |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
577 { |
10551 | 578 NDArray s (dim_vector(1, sz)); |
579 | |
580 i = 0; | |
581 for (const_iterator p = lst.begin(); p != lst.end(); p++) | |
582 s(i++) = p->s; | |
583 retval(0) = s; | |
7893
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
584 } |
5582 | 585 |
586 // Alter the order of the output arguments | |
587 if (nopts > 0) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
588 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
589 int n = 0; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
590 octave_value_list new_retval; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
591 new_retval.resize(nargout); |
5582 | 592 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
593 OCTAVE_LOCAL_BUFFER (int, arg_used, 6); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
594 for (int j = 0; j < 6; j++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
595 arg_used[j] = false; |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
596 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
597 for (int j = 2; j < nargin; j++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
598 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
599 int k = 0; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
600 std::string str = args(j).string_value(); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
601 std::transform (str.begin (), str.end (), str.begin (), tolower); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
602 if (str.find("once", 0) == 0 |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
603 || str.find("stringanchors", 0) == 0 |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
604 || str.find("lineanchors", 0) == 0 |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
605 || str.find("matchcase", 0) == 0 |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
606 || str.find("ignorecase", 0) == 0 |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
607 || str.find("dotall", 0) == 0 |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
608 || str.find("dotexceptnewline", 0) == 0 |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
609 || str.find("literalspacing", 0) == 0 |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
610 || str.find("freespacing", 0) == 0 |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
611 ) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
612 continue; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
613 else if (str.find("start", 0) == 0) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
614 k = 0; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
615 else if (str.find("end", 0) == 0) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
616 k = 1; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
617 else if (str.find("tokenextents", 0) == 0) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
618 k = 2; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
619 else if (str.find("match", 0) == 0) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
620 k = 3; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
621 else if (str.find("tokens", 0) == 0) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
622 k = 4; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
623 else if (str.find("names", 0) == 0) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
624 k = 5; |
5582 | 625 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
626 new_retval(n++) = retval(k); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
627 arg_used[k] = true; |
5582 | 628 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
629 if (n == nargout) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
630 break; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
631 } |
5582 | 632 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
633 // Fill in the rest of the arguments |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
634 if (n < nargout) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
635 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
636 for (int j = 0; j < 6; j++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
637 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
638 if (! arg_used[j]) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
639 new_retval(n++) = retval(j); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
640 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
641 } |
5582 | 642 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
643 retval = new_retval; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
644 } |
5582 | 645 } |
646 | |
647 return retval; | |
648 } | |
649 | |
6361 | 650 static octave_value_list |
651 octcellregexp (const octave_value_list &args, int nargout, const std::string &nm, | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
652 bool case_insensitive) |
6361 | 653 { |
654 octave_value_list retval; | |
655 | |
656 if (args(0).is_cell()) | |
657 { | |
658 OCTAVE_LOCAL_BUFFER (Cell, newretval, nargout); | |
659 octave_value_list new_args = args; | |
660 Cell cellstr = args(0).cell_value(); | |
661 if (args(1).is_cell()) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
662 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
663 Cell cellpat = args(1).cell_value(); |
6361 | 664 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
665 if (cellpat.numel() == 1) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
666 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
667 for (int j = 0; j < nargout; j++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
668 newretval[j].resize(cellstr.dims()); |
6361 | 669 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
670 new_args(1) = cellpat(0); |
6361 | 671 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
672 for (octave_idx_type i = 0; i < cellstr.numel (); i++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
673 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
674 new_args(0) = cellstr(i); |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
675 octave_value_list tmp = octregexp (new_args, nargout, nm, |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
676 case_insensitive); |
6361 | 677 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
678 if (error_state) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
679 break; |
6361 | 680 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
681 for (int j = 0; j < nargout; j++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
682 newretval[j](i) = tmp(j); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
683 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
684 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
685 else if (cellstr.numel() == 1) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
686 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
687 for (int j = 0; j < nargout; j++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
688 newretval[j].resize(cellpat.dims()); |
6361 | 689 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
690 new_args(0) = cellstr(0); |
6361 | 691 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
692 for (octave_idx_type i = 0; i < cellpat.numel (); i++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
693 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
694 new_args(1) = cellpat(i); |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
695 octave_value_list tmp = octregexp (new_args, nargout, nm, |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
696 case_insensitive); |
6361 | 697 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
698 if (error_state) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
699 break; |
6361 | 700 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
701 for (int j = 0; j < nargout; j++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
702 newretval[j](i) = tmp(j); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
703 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
704 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
705 else if (cellstr.numel() == cellpat.numel()) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
706 { |
6361 | 707 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
708 if (cellstr.dims() != cellpat.dims()) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
709 error ("%s: Inconsistent cell array dimensions", nm.c_str()); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
710 else |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
711 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
712 for (int j = 0; j < nargout; j++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
713 newretval[j].resize(cellstr.dims()); |
6361 | 714 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
715 for (octave_idx_type i = 0; i < cellstr.numel (); i++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
716 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
717 new_args(0) = cellstr(i); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
718 new_args(1) = cellpat(i); |
6361 | 719 |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
720 octave_value_list tmp = octregexp (new_args, nargout, nm, |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
721 case_insensitive); |
6361 | 722 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
723 if (error_state) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
724 break; |
6361 | 725 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
726 for (int j = 0; j < nargout; j++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
727 newretval[j](i) = tmp(j); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
728 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
729 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
730 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
731 else |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
732 error ("regexp: cell array arguments must be scalar or equal size"); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
733 } |
6361 | 734 else |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
735 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
736 for (int j = 0; j < nargout; j++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
737 newretval[j].resize(cellstr.dims()); |
6361 | 738 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
739 for (octave_idx_type i = 0; i < cellstr.numel (); i++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
740 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
741 new_args(0) = cellstr(i); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
742 octave_value_list tmp = octregexp (new_args, nargout, nm, case_insensitive); |
6361 | 743 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
744 if (error_state) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
745 break; |
6361 | 746 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
747 for (int j = 0; j < nargout; j++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
748 newretval[j](i) = tmp(j); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
749 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
750 } |
6361 | 751 |
752 if (!error_state) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
753 for (int j = 0; j < nargout; j++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
754 retval(j) = octave_value (newretval[j]); |
6361 | 755 } |
756 else if (args(1).is_cell()) | |
757 { | |
758 OCTAVE_LOCAL_BUFFER (Cell, newretval, nargout); | |
759 octave_value_list new_args = args; | |
760 Cell cellpat = args(1).cell_value(); | |
761 | |
762 for (int j = 0; j < nargout; j++) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
763 newretval[j].resize(cellpat.dims()); |
6361 | 764 |
765 for (octave_idx_type i = 0; i < cellpat.numel (); i++) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
766 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
767 new_args(1) = cellpat(i); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
768 octave_value_list tmp = octregexp (new_args, nargout, nm, case_insensitive); |
6361 | 769 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
770 if (error_state) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
771 break; |
6361 | 772 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
773 for (int j = 0; j < nargout; j++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
774 newretval[j](i) = tmp(j); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
775 } |
6361 | 776 |
777 if (!error_state) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
778 for (int j = 0; j < nargout; j++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
779 retval(j) = octave_value (newretval[j]); |
6361 | 780 } |
781 else | |
782 retval = octregexp (args, nargout, nm, case_insensitive); | |
783 | |
784 return retval; | |
785 | |
786 } | |
787 | |
5582 | 788 DEFUN_DLD (regexp, args, nargout, |
789 "-*- texinfo -*-\n\ | |
10840 | 790 @deftypefn {Loadable Function} {[@var{s}, @var{e}, @var{te}, @var{m}, @var{t}, @var{nm}] =} regexp (@var{str}, @var{pat})\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
791 @deftypefnx {Loadable Function} {[@dots{}] =} regexp (@var{str}, @var{pat}, \"@var{opt1}\", @dots{})\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
792 Regular expression string matching. Search for @var{pat} in @var{str} and\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
793 return the positions and substrings of any matches, or empty values if there\n\ |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
794 are none.\n\ |
5582 | 795 \n\ |
796 The matched pattern @var{pat} can include any of the standard regex\n\ | |
797 operators, including:\n\ | |
798 \n\ | |
799 @table @code\n\ | |
800 @item .\n\ | |
801 Match any character\n\ | |
10840 | 802 \n\ |
5582 | 803 @item * + ? @{@}\n\ |
804 Repetition operators, representing\n\ | |
805 @table @code\n\ | |
806 @item *\n\ | |
807 Match zero or more times\n\ | |
10840 | 808 \n\ |
5582 | 809 @item +\n\ |
810 Match one or more times\n\ | |
10840 | 811 \n\ |
5582 | 812 @item ?\n\ |
813 Match zero or one times\n\ | |
10840 | 814 \n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
815 @item @{@var{n}@}\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
816 Match exactly @var{n} times\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
817 \n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
818 @item @{@var{n},@}\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
819 Match @var{n} or more times\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
820 \n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
821 @item @{@var{m},@var{n}@}\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
822 Match between @var{m} and @var{n} times\n\ |
5582 | 823 @end table\n\ |
10840 | 824 \n\ |
5582 | 825 @item [@dots{}] [^@dots{}]\n\ |
10840 | 826 \n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
827 List operators. The pattern will match any character listed between \"[\"\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
828 and \"]\". If the first character is \"^\" then the pattern is inverted and\n\ |
11032
c9b0a75b02e8
Make all regexp in Octave compatible with both POSIX and PCRE.
Rik <octave@nomad.inbox5.com>
parents:
11025
diff
changeset
|
829 any character except those listed between brackets will match.\n\ |
c9b0a75b02e8
Make all regexp in Octave compatible with both POSIX and PCRE.
Rik <octave@nomad.inbox5.com>
parents:
11025
diff
changeset
|
830 \n\ |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
831 Escape sequences defined below can also be used inside list\n\ |
11032
c9b0a75b02e8
Make all regexp in Octave compatible with both POSIX and PCRE.
Rik <octave@nomad.inbox5.com>
parents:
11025
diff
changeset
|
832 operators. For example, a template for a floating point number might be\n\ |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
833 @code{[-+.\\d]+}.\n\ |
10840 | 834 \n\ |
5582 | 835 @item ()\n\ |
836 Grouping operator\n\ | |
10840 | 837 \n\ |
5582 | 838 @item |\n\ |
9036
58604c45ca74
Cleanup of data types related documentation
Rik <rdrider0-list@yahoo.com>
parents:
8920
diff
changeset
|
839 Alternation operator. Match one of a choice of regular expressions. The\n\ |
11032
c9b0a75b02e8
Make all regexp in Octave compatible with both POSIX and PCRE.
Rik <octave@nomad.inbox5.com>
parents:
11025
diff
changeset
|
840 alternatives must be delimited by the grouping operator @code{()} above.\n\ |
10840 | 841 \n\ |
5582 | 842 @item ^ $\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
843 Anchoring operators. Requires pattern to occur at the start (@code{^}) or\n\ |
11032
c9b0a75b02e8
Make all regexp in Octave compatible with both POSIX and PCRE.
Rik <octave@nomad.inbox5.com>
parents:
11025
diff
changeset
|
844 end (@code{$}) of the string.\n\ |
5582 | 845 @end table\n\ |
846 \n\ | |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
847 In addition, the following escaped characters have special meaning. Note,\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
848 it is recommended to quote @var{pat} in single quotes, rather than double\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
849 quotes, to avoid the escape sequences being interpreted by Octave before\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
850 being passed to @code{regexp}.\n\ |
5582 | 851 \n\ |
852 @table @code\n\ | |
853 @item \\b\n\ | |
854 Match a word boundary\n\ | |
10840 | 855 \n\ |
5582 | 856 @item \\B\n\ |
857 Match within a word\n\ | |
10840 | 858 \n\ |
5582 | 859 @item \\w\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
860 Match any word character\n\ |
10840 | 861 \n\ |
5582 | 862 @item \\W\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
863 Match any non-word character\n\ |
10840 | 864 \n\ |
5582 | 865 @item \\<\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
866 Match the beginning of a word\n\ |
10840 | 867 \n\ |
5582 | 868 @item \\>\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
869 Match the end of a word\n\ |
10840 | 870 \n\ |
5582 | 871 @item \\s\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
872 Match any whitespace character\n\ |
10840 | 873 \n\ |
5582 | 874 @item \\S\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
875 Match any non-whitespace character\n\ |
10840 | 876 \n\ |
5582 | 877 @item \\d\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
878 Match any digit\n\ |
10840 | 879 \n\ |
5582 | 880 @item \\D\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
881 Match any non-digit\n\ |
5582 | 882 @end table\n\ |
883 \n\ | |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
884 The outputs of @code{regexp} default to the order given below\n\ |
5582 | 885 \n\ |
11032
c9b0a75b02e8
Make all regexp in Octave compatible with both POSIX and PCRE.
Rik <octave@nomad.inbox5.com>
parents:
11025
diff
changeset
|
886 @table @var\n\ |
c9b0a75b02e8
Make all regexp in Octave compatible with both POSIX and PCRE.
Rik <octave@nomad.inbox5.com>
parents:
11025
diff
changeset
|
887 @item s\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
888 The start indices of each matching substring\n\ |
5582 | 889 \n\ |
11032
c9b0a75b02e8
Make all regexp in Octave compatible with both POSIX and PCRE.
Rik <octave@nomad.inbox5.com>
parents:
11025
diff
changeset
|
890 @item e\n\ |
5582 | 891 The end indices of each matching substring\n\ |
892 \n\ | |
11032
c9b0a75b02e8
Make all regexp in Octave compatible with both POSIX and PCRE.
Rik <octave@nomad.inbox5.com>
parents:
11025
diff
changeset
|
893 @item te\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
894 The extents of each matched token surrounded by @code{(@dots{})} in\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
895 @var{pat}\n\ |
5582 | 896 \n\ |
11032
c9b0a75b02e8
Make all regexp in Octave compatible with both POSIX and PCRE.
Rik <octave@nomad.inbox5.com>
parents:
11025
diff
changeset
|
897 @item m\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
898 A cell array of the text of each match\n\ |
5582 | 899 \n\ |
11032
c9b0a75b02e8
Make all regexp in Octave compatible with both POSIX and PCRE.
Rik <octave@nomad.inbox5.com>
parents:
11025
diff
changeset
|
900 @item t\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
901 A cell array of the text of each token matched\n\ |
5582 | 902 \n\ |
11032
c9b0a75b02e8
Make all regexp in Octave compatible with both POSIX and PCRE.
Rik <octave@nomad.inbox5.com>
parents:
11025
diff
changeset
|
903 @item nm\n\ |
5582 | 904 A structure containing the text of each matched named token, with the name\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
905 being used as the fieldname. A named token is denoted by\n\ |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
906 @code{(?<name>@dots{})}.\n\ |
5582 | 907 @end table\n\ |
908 \n\ | |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
909 Particular output arguments, or the order of the output arguments, can be\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
910 selected by additional @var{opt} arguments. These are strings and the\n\ |
5582 | 911 correspondence between the output arguments and the optional argument\n\ |
912 are\n\ | |
913 \n\ | |
914 @multitable @columnfractions 0.2 0.3 0.3 0.2\n\ | |
915 @item @tab 'start' @tab @var{s} @tab\n\ | |
916 @item @tab 'end' @tab @var{e} @tab\n\ | |
917 @item @tab 'tokenExtents' @tab @var{te} @tab\n\ | |
918 @item @tab 'match' @tab @var{m} @tab\n\ | |
919 @item @tab 'tokens' @tab @var{t} @tab\n\ | |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
920 @item @tab 'names' @tab @var{nm} @tab\n\ |
5582 | 921 @end multitable\n\ |
922 \n\ | |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
923 Additional arguments are summarized below.\n\ |
5779 | 924 \n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
925 @table @samp\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
926 @item once\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
927 Return only the first occurrence of the pattern.\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
928 \n\ |
5779 | 929 @item matchcase\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
930 Make the matching case sensitive. (default)\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
931 \n\ |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
932 Alternatively, use (?-i) in the pattern.\n\ |
10840 | 933 \n\ |
5779 | 934 @item ignorecase\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
935 Ignore case when matching the pattern to the string.\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
936 \n\ |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
937 Alternatively, use (?i) in the pattern.\n\ |
10840 | 938 \n\ |
5779 | 939 @item stringanchors\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
940 Match the anchor characters at the beginning and end of the string. \n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
941 (default)\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
942 \n\ |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
943 Alternatively, use (?-m) in the pattern.\n\ |
10840 | 944 \n\ |
5779 | 945 @item lineanchors\n\ |
946 Match the anchor characters at the beginning and end of the line.\n\ | |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
947 \n\ |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
948 Alternatively, use (?m) in the pattern.\n\ |
10840 | 949 \n\ |
5779 | 950 @item dotall\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
951 The pattern @code{.} matches all characters including the newline character.\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
952 (default)\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
953 \n\ |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
954 Alternatively, use (?s) in the pattern.\n\ |
10840 | 955 \n\ |
5779 | 956 @item dotexceptnewline\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
957 The pattern @code{.} matches all characters except the newline character.\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
958 \n\ |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
959 Alternatively, use (?-s) in the pattern.\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
960 \n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
961 @item literalspacing\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
962 All characters in the pattern, including whitespace, are significant and are\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
963 used in pattern matching. (default)\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
964 \n\ |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
965 Alternatively, use (?-x) in the pattern.\n\ |
10840 | 966 \n\ |
5779 | 967 @item freespacing\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
968 The pattern may include arbitrary whitespace and also comments beginning with\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
969 the character @samp{#}.\n\ |
10840 | 970 \n\ |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
971 Alternatively, use (?x) in the pattern.\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
972 \n\ |
5779 | 973 @end table\n\ |
11572
7d6d8c1e471f
Grammarcheck Texinfo for files in src directory.
Rik <octave@nomad.inbox5.com>
parents:
11553
diff
changeset
|
974 @seealso{regexpi, strfind, regexprep}\n\ |
5582 | 975 @end deftypefn") |
976 { | |
6361 | 977 octave_value_list retval; |
978 int nargin = args.length(); | |
979 | |
980 if (nargin < 2) | |
981 print_usage (); | |
982 else if (args(0).is_cell() || args(1).is_cell()) | |
983 retval = octcellregexp (args, nargout, "regexp", false); | |
984 else | |
985 retval = octregexp (args, nargout, "regexp", false); | |
986 | |
987 return retval; | |
5582 | 988 } |
989 | |
990 /* | |
991 | |
8140
cdd05e46f6c9
Increase pcre's match_limit for difficult regexps
Thomas Weber <thomas.weber.mail@gmail.com>
parents:
8093
diff
changeset
|
992 ## PCRE_ERROR_MATCHLIMIT test |
cdd05e46f6c9
Increase pcre's match_limit for difficult regexps
Thomas Weber <thomas.weber.mail@gmail.com>
parents:
8093
diff
changeset
|
993 %!test |
cdd05e46f6c9
Increase pcre's match_limit for difficult regexps
Thomas Weber <thomas.weber.mail@gmail.com>
parents:
8093
diff
changeset
|
994 %! s=sprintf('\t4\n0000\t-0.00\t-0.0000\t4\t-0.00\t-0.0000\t4\n0000\t-0.00\t-0.0000\t0\t-0.00\t-'); |
cdd05e46f6c9
Increase pcre's match_limit for difficult regexps
Thomas Weber <thomas.weber.mail@gmail.com>
parents:
8093
diff
changeset
|
995 %! ws = warning("query"); |
cdd05e46f6c9
Increase pcre's match_limit for difficult regexps
Thomas Weber <thomas.weber.mail@gmail.com>
parents:
8093
diff
changeset
|
996 %! unwind_protect |
cdd05e46f6c9
Increase pcre's match_limit for difficult regexps
Thomas Weber <thomas.weber.mail@gmail.com>
parents:
8093
diff
changeset
|
997 %! warning("off"); |
cdd05e46f6c9
Increase pcre's match_limit for difficult regexps
Thomas Weber <thomas.weber.mail@gmail.com>
parents:
8093
diff
changeset
|
998 %! regexp(s, '(\s*-*\d+[.]*\d*\s*)+\n'); |
cdd05e46f6c9
Increase pcre's match_limit for difficult regexps
Thomas Weber <thomas.weber.mail@gmail.com>
parents:
8093
diff
changeset
|
999 %! unwind_protect_cleanup |
cdd05e46f6c9
Increase pcre's match_limit for difficult regexps
Thomas Weber <thomas.weber.mail@gmail.com>
parents:
8093
diff
changeset
|
1000 %! warning(ws); |
cdd05e46f6c9
Increase pcre's match_limit for difficult regexps
Thomas Weber <thomas.weber.mail@gmail.com>
parents:
8093
diff
changeset
|
1001 %! end_unwind_protect |
cdd05e46f6c9
Increase pcre's match_limit for difficult regexps
Thomas Weber <thomas.weber.mail@gmail.com>
parents:
8093
diff
changeset
|
1002 |
5582 | 1003 ## seg-fault test |
1004 %!assert(regexp("abcde","."),[1,2,3,4,5]) | |
1005 | |
1006 ## Check that anchoring of pattern works correctly | |
1007 %!assert(regexp('abcabc','^abc'),1); | |
1008 %!assert(regexp('abcabc','abc$'),4); | |
5785 | 1009 %!assert(regexp('abcabc','^abc$'),zeros(1,0)); |
5582 | 1010 |
1011 %!test | |
1012 %! [s, e, te, m, t] = regexp(' No Match ', 'f(.*)uck'); | |
5785 | 1013 %! assert (s,zeros(1,0)) |
1014 %! assert (e,zeros(1,0)) | |
1015 %! assert (te,cell(1,0)) | |
1016 %! assert (m, cell(1,0)) | |
1017 %! assert (t, cell(1,0)) | |
5582 | 1018 |
1019 %!test | |
1020 %! [s, e, te, m, t] = regexp(' FiRetrUck ', 'f(.*)uck'); | |
5785 | 1021 %! assert (s,zeros(1,0)) |
1022 %! assert (e,zeros(1,0)) | |
1023 %! assert (te,cell(1,0)) | |
1024 %! assert (m, cell(1,0)) | |
1025 %! assert (t, cell(1,0)) | |
5582 | 1026 |
1027 %!test | |
1028 %! [s, e, te, m, t] = regexp(' firetruck ', 'f(.*)uck'); | |
1029 %! assert (s,2) | |
1030 %! assert (e,10) | |
1031 %! assert (te{1},[3,7]) | |
1032 %! assert (m{1}, 'firetruck') | |
1033 %! assert (t{1}{1}, 'iretr') | |
1034 | |
1035 %!test | |
1036 %! [s, e, te, m, t] = regexp('short test string','\w*r\w*'); | |
1037 %! assert (s,[1,12]) | |
1038 %! assert (e,[5,17]) | |
1039 %! assert (size(te), [1,2]) | |
1040 %! assert (isempty(te{1})) | |
1041 %! assert (isempty(te{2})) | |
1042 %! assert (m{1},'short') | |
1043 %! assert (m{2},'string') | |
1044 %! assert (size(t), [1,2]) | |
1045 %! assert (isempty(t{1})) | |
1046 %! assert (isempty(t{2})) | |
1047 | |
1048 %!test | |
1049 %! [s, e, te, m, t] = regexp('short test string','\w*r\w*','once'); | |
1050 %! assert (s,1) | |
1051 %! assert (e,5) | |
7893
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
1052 %! assert (isempty(te)) |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
1053 %! assert (m,'short') |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
1054 %! assert (isempty(t)) |
5582 | 1055 |
1056 %!test | |
1057 %! [m, te, e, s, t] = regexp('short test string','\w*r\w*','once', 'match', 'tokenExtents', 'end', 'start', 'tokens'); | |
1058 %! assert (s,1) | |
1059 %! assert (e,5) | |
7893
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
1060 %! assert (isempty(te)) |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
1061 %! assert (m,'short') |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
1062 %! assert (isempty(t)) |
5582 | 1063 |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
1064 %!test |
7242 | 1065 %! [s, e, te, m, t, nm] = regexp('short test string','(?<word1>\w*t)\s*(?<word2>\w*t)'); |
1066 %! assert (s,1) | |
1067 %! assert (e,10) | |
1068 %! assert (size(te), [1,1]) | |
1069 %! assert (te{1}, [1 5; 7, 10]) | |
1070 %! assert (m{1},'short test') | |
1071 %! assert (size(t),[1,1]) | |
1072 %! assert (t{1}{1},'short') | |
1073 %! assert (t{1}{2},'test') | |
1074 %! assert (size(nm), [1,1]) | |
1075 %! assert (!isempty(fieldnames(nm))) | |
1076 %! assert (sort(fieldnames(nm)),{'word1';'word2'}) | |
1077 %! assert (nm.word1,'short') | |
1078 %! assert (nm.word2,'test') | |
5582 | 1079 |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
1080 %!test |
7242 | 1081 %! [nm, m, te, e, s, t] = regexp('short test string','(?<word1>\w*t)\s*(?<word2>\w*t)', 'names', 'match', 'tokenExtents', 'end', 'start', 'tokens'); |
1082 %! assert (s,1) | |
1083 %! assert (e,10) | |
1084 %! assert (size(te), [1,1]) | |
1085 %! assert (te{1}, [1 5; 7, 10]) | |
1086 %! assert (m{1},'short test') | |
1087 %! assert (size(t),[1,1]) | |
1088 %! assert (t{1}{1},'short') | |
1089 %! assert (t{1}{2},'test') | |
1090 %! assert (size(nm), [1,1]) | |
1091 %! assert (!isempty(fieldnames(nm))) | |
1092 %! assert (sort(fieldnames(nm)),{'word1';'word2'}) | |
1093 %! assert (nm.word1,'short') | |
1094 %! assert (nm.word2,'test') | |
5619 | 1095 |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
1096 %!test |
7242 | 1097 %! [t, nm] = regexp("John Davis\nRogers, James",'(?<first>\w+)\s+(?<last>\w+)|(?<last>\w+),\s+(?<first>\w+)','tokens','names'); |
1098 %! assert (size(t), [1,2]); | |
1099 %! assert (t{1}{1},'John'); | |
1100 %! assert (t{1}{2},'Davis'); | |
1101 %! assert (t{2}{1},'Rogers'); | |
1102 %! assert (t{2}{2},'James'); | |
1103 %! assert (size(nm), [1,1]); | |
1104 %! assert (nm.first{1},'John'); | |
1105 %! assert (nm.first{2},'James'); | |
1106 %! assert (nm.last{1},'Davis'); | |
1107 %! assert (nm.last{2},'Rogers'); | |
5582 | 1108 |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
1109 %!test |
10518
fcafe0e9bd58
Handle repeated matches in matches returned by pcre
David Bateman <dbateman@free.fr>
parents:
10504
diff
changeset
|
1110 %! # Parenthesis in named token (ie (int)) causes a problem |
fcafe0e9bd58
Handle repeated matches in matches returned by pcre
David Bateman <dbateman@free.fr>
parents:
10504
diff
changeset
|
1111 %! assert (regexp('qwe int asd', ['(?<typestr>(int))'], 'names'), struct ('typestr', 'int')); |
fcafe0e9bd58
Handle repeated matches in matches returned by pcre
David Bateman <dbateman@free.fr>
parents:
10504
diff
changeset
|
1112 |
5779 | 1113 %!assert(regexp("abc\nabc",'.'),[1:7]) |
1114 %!assert(regexp("abc\nabc",'.','dotall'),[1:7]) | |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
1115 %!test |
7242 | 1116 %! assert(regexp("abc\nabc",'(?s).'),[1:7]) |
1117 %! assert(regexp("abc\nabc",'.','dotexceptnewline'),[1,2,3,5,6,7]) | |
1118 %! assert(regexp("abc\nabc",'(?-s).'),[1,2,3,5,6,7]) | |
5779 | 1119 |
1120 %!assert(regexp("caseCaSe",'case'),1) | |
1121 %!assert(regexp("caseCaSe",'case',"matchcase"),1) | |
1122 %!assert(regexp("caseCaSe",'case',"ignorecase"),[1,5]) | |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
1123 %!test |
7242 | 1124 %! assert(regexp("caseCaSe",'(?-i)case'),1) |
1125 %! assert(regexp("caseCaSe",'(?i)case'),[1,5]) | |
5779 | 1126 |
1127 %!assert (regexp("abc\nabc",'c$'),7) | |
1128 %!assert (regexp("abc\nabc",'c$',"stringanchors"),7) | |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
1129 %!test |
7242 | 1130 %! assert (regexp("abc\nabc",'(?-m)c$'),7) |
1131 %! assert (regexp("abc\nabc",'c$',"lineanchors"),[3,7]) | |
1132 %! assert (regexp("abc\nabc",'(?m)c$'),[3,7]) | |
5779 | 1133 |
1134 %!assert (regexp("this word",'s w'),4) | |
1135 %!assert (regexp("this word",'s w','literalspacing'),4) | |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
1136 %!test |
7242 | 1137 %! assert (regexp("this word",'(?-x)s w','literalspacing'),4) |
1138 %! assert (regexp("this word",'s w','freespacing'),zeros(1,0)) | |
1139 %! assert (regexp("this word",'(?x)s w'),zeros(1,0)) | |
5779 | 1140 |
5582 | 1141 %!error regexp('string', 'tri', 'BadArg'); |
1142 %!error regexp('string'); | |
1143 | |
6361 | 1144 %!assert(regexp({'asdfg-dfd';'-dfd-dfd-';'qasfdfdaq'},'-'),{6;[1,5,9];zeros(1,0)}) |
1145 %!assert(regexp({'asdfg-dfd','-dfd-dfd-','qasfdfdaq'},'-'),{6,[1,5,9],zeros(1,0)}) | |
1146 %!assert(regexp({'asdfg-dfd';'-dfd-dfd-';'qasfdfdaq'},{'-';'f';'q'}),{6;[3,7];[1,9]}) | |
1147 %!assert(regexp('Strings',{'t','s'}),{2,7}) | |
1148 | |
8093
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
1149 ## Test case for lookaround operators |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
1150 %!test |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1151 %! assert(regexp('Iraq','q(?!u)'),4) |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1152 %! assert(regexp('quit','q(?!u)'), zeros(1,0)) |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1153 %! assert(regexp('quit','q(?=u)','match'), {'q'}) |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1154 %! assert(regexp("quit",'q(?=u+)','match'), {'q'}) |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1155 %! assert(regexp("qit",'q(?=u+)','match'), cell(1,0)) |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1156 %! assert(regexp("qit",'q(?=u*)','match'), {'q'}) |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1157 %! assert(regexp('thingamabob','(?<=a)b'), 9) |
8093
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
1158 |
5582 | 1159 */ |
1160 | |
6549 | 1161 DEFUN_DLD (regexpi, args, nargout, |
5582 | 1162 "-*- texinfo -*-\n\ |
10840 | 1163 @deftypefn {Loadable Function} {[@var{s}, @var{e}, @var{te}, @var{m}, @var{t}, @var{nm}] =} regexpi (@var{str}, @var{pat})\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1164 @deftypefnx {Loadable Function} {[@dots{}] =} regexpi (@var{str}, @var{pat}, \"@var{opt1}\", @dots{})\n\ |
5582 | 1165 \n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1166 Case insensitive regular expression string matching. Search for @var{pat} in\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1167 @var{str} and return the positions and substrings of any matches, or empty\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1168 values if there are none. @xref{doc-regexp,,regexp}, for details on the\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1169 syntax of the search pattern.\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1170 @seealso{regexp}\n\ |
5582 | 1171 @end deftypefn") |
1172 { | |
6361 | 1173 octave_value_list retval; |
1174 int nargin = args.length(); | |
1175 | |
1176 if (nargin < 2) | |
1177 print_usage (); | |
1178 else if (args(0).is_cell() || args(1).is_cell()) | |
1179 retval = octcellregexp (args, nargout, "regexpi", true); | |
1180 else | |
1181 retval = octregexp (args, nargout, "regexpi", true); | |
1182 | |
1183 return retval; | |
5582 | 1184 } |
1185 | |
1186 /* | |
1187 | |
1188 ## seg-fault test | |
1189 %!assert(regexpi("abcde","."),[1,2,3,4,5]) | |
1190 | |
1191 ## Check that anchoring of pattern works correctly | |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1192 %!assert(regexpi('abcabc','^ABC'),1); |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1193 %!assert(regexpi('abcabc','ABC$'),4); |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1194 %!assert(regexpi('abcabc','^ABC$'),zeros(1,0)); |
5582 | 1195 |
1196 %!test | |
1197 %! [s, e, te, m, t] = regexpi(' No Match ', 'f(.*)uck'); | |
5785 | 1198 %! assert (s,zeros(1,0)) |
1199 %! assert (e,zeros(1,0)) | |
1200 %! assert (te,cell(1,0)) | |
1201 %! assert (m, cell(1,0)) | |
1202 %! assert (t, cell(1,0)) | |
5582 | 1203 |
1204 %!test | |
1205 %! [s, e, te, m, t] = regexpi(' FiRetrUck ', 'f(.*)uck'); | |
1206 %! assert (s,2) | |
1207 %! assert (e,10) | |
1208 %! assert (te{1},[3,7]) | |
1209 %! assert (m{1}, 'FiRetrUck') | |
1210 %! assert (t{1}{1}, 'iRetr') | |
1211 | |
1212 %!test | |
1213 %! [s, e, te, m, t] = regexpi(' firetruck ', 'f(.*)uck'); | |
1214 %! assert (s,2) | |
1215 %! assert (e,10) | |
1216 %! assert (te{1},[3,7]) | |
1217 %! assert (m{1}, 'firetruck') | |
1218 %! assert (t{1}{1}, 'iretr') | |
1219 | |
1220 %!test | |
1221 %! [s, e, te, m, t] = regexpi('ShoRt Test String','\w*r\w*'); | |
1222 %! assert (s,[1,12]) | |
1223 %! assert (e,[5,17]) | |
1224 %! assert (size(te), [1,2]) | |
1225 %! assert (isempty(te{1})) | |
1226 %! assert (isempty(te{2})) | |
1227 %! assert (m{1},'ShoRt') | |
1228 %! assert (m{2},'String') | |
1229 %! assert (size(t), [1,2]) | |
1230 %! assert (isempty(t{1})) | |
1231 %! assert (isempty(t{2})) | |
1232 | |
1233 %!test | |
1234 %! [s, e, te, m, t] = regexpi('ShoRt Test String','\w*r\w*','once'); | |
1235 %! assert (s,1) | |
1236 %! assert (e,5) | |
7893
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
1237 %! assert (isempty(te)) |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
1238 %! assert (m,'ShoRt') |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
1239 %! assert (isempty(t)) |
5582 | 1240 |
1241 %!test | |
1242 %! [m, te, e, s, t] = regexpi('ShoRt Test String','\w*r\w*','once', 'match', 'tokenExtents', 'end', 'start', 'tokens'); | |
1243 %! assert (s,1) | |
1244 %! assert (e,5) | |
7893
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
1245 %! assert (isempty(te)) |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
1246 %! assert (m,'ShoRt') |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
1247 %! assert (isempty(t)) |
5582 | 1248 |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
1249 %!test |
7242 | 1250 %! [s, e, te, m, t, nm] = regexpi('ShoRt Test String','(?<word1>\w*t)\s*(?<word2>\w*t)'); |
1251 %! assert (s,1) | |
1252 %! assert (e,10) | |
1253 %! assert (size(te), [1,1]) | |
1254 %! assert (te{1}, [1 5; 7, 10]) | |
1255 %! assert (m{1},'ShoRt Test') | |
1256 %! assert (size(t),[1,1]) | |
1257 %! assert (t{1}{1},'ShoRt') | |
1258 %! assert (t{1}{2},'Test') | |
1259 %! assert (size(nm), [1,1]) | |
1260 %! assert (!isempty(fieldnames(nm))) | |
1261 %! assert (sort(fieldnames(nm)),{'word1';'word2'}) | |
1262 %! assert (nm.word1,'ShoRt') | |
1263 %! assert (nm.word2,'Test') | |
5582 | 1264 |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
1265 %!test |
7242 | 1266 %! [nm, m, te, e, s, t] = regexpi('ShoRt Test String','(?<word1>\w*t)\s*(?<word2>\w*t)', 'names', 'match', 'tokenExtents', 'end', 'start', 'tokens'); |
1267 %! assert (s,1) | |
1268 %! assert (e,10) | |
1269 %! assert (size(te), [1,1]) | |
1270 %! assert (te{1}, [1 5; 7, 10]) | |
1271 %! assert (m{1},'ShoRt Test') | |
1272 %! assert (size(t),[1,1]) | |
1273 %! assert (t{1}{1},'ShoRt') | |
1274 %! assert (t{1}{2},'Test') | |
1275 %! assert (size(nm), [1,1]) | |
1276 %! assert (!isempty(fieldnames(nm))) | |
1277 %! assert (sort(fieldnames(nm)),{'word1';'word2'}) | |
1278 %! assert (nm.word1,'ShoRt') | |
1279 %! assert (nm.word2,'Test') | |
5582 | 1280 |
5779 | 1281 %!assert(regexpi("abc\nabc",'.'),[1:7]) |
1282 %!assert(regexpi("abc\nabc",'.','dotall'),[1:7]) | |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
1283 %!test |
7242 | 1284 %! assert(regexpi("abc\nabc",'(?s).'),[1:7]) |
1285 %! assert(regexpi("abc\nabc",'.','dotexceptnewline'),[1,2,3,5,6,7]) | |
1286 %! assert(regexpi("abc\nabc",'(?-s).'),[1,2,3,5,6,7]) | |
5779 | 1287 |
1288 %!assert(regexpi("caseCaSe",'case'),[1,5]) | |
1289 %!assert(regexpi("caseCaSe",'case',"matchcase"),1) | |
1290 %!assert(regexpi("caseCaSe",'case',"ignorecase"),[1,5]) | |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
1291 %!test |
7242 | 1292 %! assert(regexpi("caseCaSe",'(?-i)case'),1) |
1293 %! assert(regexpi("caseCaSe",'(?i)case'),[1,5]) | |
5779 | 1294 |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1295 %!assert (regexpi("abc\nabc",'C$'),7) |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1296 %!assert (regexpi("abc\nabc",'C$',"stringanchors"),7) |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
1297 %!test |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1298 %! assert (regexpi("abc\nabc",'(?-m)C$'),7) |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1299 %! assert (regexpi("abc\nabc",'C$',"lineanchors"),[3,7]) |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1300 %! assert (regexpi("abc\nabc",'(?m)C$'),[3,7]) |
5779 | 1301 |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1302 %!assert (regexpi("this word",'S w'),4) |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1303 %!assert (regexpi("this word",'S w','literalspacing'),4) |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
1304 %!test |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1305 %! assert (regexpi("this word",'(?-x)S w','literalspacing'),4) |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1306 %! assert (regexpi("this word",'S w','freespacing'),zeros(1,0)) |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1307 %! assert (regexpi("this word",'(?x)S w'),zeros(1,0)) |
5779 | 1308 |
5582 | 1309 %!error regexpi('string', 'tri', 'BadArg'); |
1310 %!error regexpi('string'); | |
1311 | |
6361 | 1312 %!assert(regexpi({'asdfg-dfd';'-dfd-dfd-';'qasfdfdaq'},'-'),{6;[1,5,9];zeros(1,0)}) |
1313 %!assert(regexpi({'asdfg-dfd','-dfd-dfd-','qasfdfdaq'},'-'),{6,[1,5,9],zeros(1,0)}) | |
1314 %!assert(regexpi({'asdfg-dfd';'-dfd-dfd-';'qasfdfdaq'},{'-';'f';'q'}),{6;[3,7];[1,9]}) | |
1315 %!assert(regexpi('Strings',{'t','s'}),{2,[1,7]}) | |
1316 | |
5582 | 1317 */ |
1318 | |
6361 | 1319 |
1320 static octave_value | |
1321 octregexprep (const octave_value_list &args, const std::string &nm) | |
5785 | 1322 { |
6361 | 1323 octave_value retval; |
5785 | 1324 int nargin = args.length(); |
1325 | |
1326 // Make sure we have string,pattern,replacement | |
1327 const std::string buffer = args(0).string_value (); | |
1328 if (error_state) return retval; | |
1329 const std::string pattern = args(1).string_value (); | |
1330 if (error_state) return retval; | |
1331 const std::string replacement = args(2).string_value (); | |
1332 if (error_state) return retval; | |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1333 |
5785 | 1334 // Pack options excluding 'tokenize' and various output |
1335 // reordering strings into regexp arg list | |
1336 octave_value_list regexpargs(nargin-1,octave_value()); | |
1337 regexpargs(0) = args(0); | |
1338 regexpargs(1) = args(1); | |
1339 int len=2; | |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1340 for (int i = 3; i < nargin; i++) |
5785 | 1341 { |
1342 const std::string opt = args(i).string_value(); | |
1343 if (opt != "tokenize" && opt != "start" && opt != "end" | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1344 && opt != "tokenextents" && opt != "match" && opt != "tokens" |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1345 && opt != "names" && opt != "warnings") |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1346 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1347 regexpargs(len++) = args(i); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1348 } |
5785 | 1349 } |
1350 regexpargs.resize(len); | |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1351 |
5785 | 1352 // Identify replacement tokens; build a vector of group numbers in |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1353 // the replacement string so that we can quickly calculate the size |
5785 | 1354 // of the replacement. |
1355 int tokens = 0; | |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1356 for (size_t i=1; i < replacement.size(); i++) |
5785 | 1357 { |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1358 if (replacement[i-1]=='$' && isdigit(replacement[i])) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1359 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1360 tokens++, i++; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1361 } |
5785 | 1362 } |
1363 std::vector<int> token(tokens); | |
1364 int kk = 0; | |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1365 for (size_t i = 1; i < replacement.size(); i++) |
5785 | 1366 { |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1367 if (replacement[i-1]=='$' && isdigit(replacement[i])) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1368 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1369 token[kk++] = replacement[i]-'0'; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1370 i++; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1371 } |
5785 | 1372 } |
1373 | |
1374 // Perform replacement | |
1375 std::string rep; | |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1376 if (tokens > 0) |
5785 | 1377 { |
1378 std::list<regexp_elem> lst; | |
1379 string_vector named; | |
1380 int nopts; | |
7893
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
1381 bool once; |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
1382 int sz = octregexp_list (regexpargs, nm , false, lst, named, nopts, once); |
5785 | 1383 |
1384 if (error_state) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1385 return retval; |
5785 | 1386 if (sz == 0) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1387 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1388 retval = args(0); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1389 return retval; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1390 } |
5785 | 1391 |
1392 // Determine replacement length | |
1393 const size_t replen = replacement.size() - 2*tokens; | |
1394 int delta = 0; | |
1395 const_iterator p = lst.begin(); | |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1396 for (int i = 0; i < sz; i++) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1397 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1398 OCTAVE_QUIT; |
5785 | 1399 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1400 const Matrix pairs(p->te); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1401 size_t pairlen = 0; |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1402 for (int j = 0; j < tokens; j++) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1403 { |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1404 if (token[j] == 0) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1405 pairlen += static_cast<size_t>(p->e - p->s) + 1; |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1406 else if (token[j] <= pairs.rows()) |
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1407 pairlen += static_cast<size_t>(pairs(token[j]-1,1) - |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1408 pairs(token[j]-1,0)) + 1; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1409 } |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1410 delta += static_cast<int>(replen + pairlen) - |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1411 static_cast<int>(p->e - p->s + 1); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1412 p++; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1413 } |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1414 |
5785 | 1415 // Build replacement string |
1416 rep.reserve(buffer.size()+delta); | |
1417 size_t from = 0; | |
1418 p = lst.begin(); | |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1419 for (int i=0; i < sz; i++) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1420 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1421 OCTAVE_QUIT; |
5785 | 1422 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1423 const Matrix pairs(p->te); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1424 rep.append(&buffer[from], static_cast<size_t>(p->s - 1) - from); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1425 from = static_cast<size_t>(p->e - 1) + 1; |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1426 for (size_t j = 1; j < replacement.size(); j++) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1427 { |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1428 if (replacement[j-1]=='$' && isdigit(replacement[j])) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1429 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1430 int k = replacement[j]-'0'; |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1431 if (k == 0) |
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1432 { |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1433 // replace with entire match |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1434 rep.append(&buffer[static_cast<size_t>(p->e - 1)], |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1435 static_cast<size_t>(p->e - p->s) + 1); |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1436 } |
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1437 else if (k <= pairs.rows()) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1438 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1439 // replace with group capture |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1440 rep.append(&buffer[static_cast<size_t>(pairs(k-1,0)-1)], |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1441 static_cast<size_t>(pairs(k-1,1) - |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1442 pairs(k-1,0))+1); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1443 } |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1444 else |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1445 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1446 // replace with nothing |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1447 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1448 j++; |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1449 } |
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1450 else |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1451 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1452 rep.append(1,replacement[j-1]); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1453 } |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1454 if (j+1 == replacement.size()) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1455 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1456 rep.append(1,replacement[j]); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1457 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1458 } |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1459 p++; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1460 } |
5785 | 1461 rep.append(&buffer[from],buffer.size()-from); |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1462 } |
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1463 else |
5785 | 1464 { |
1465 std::list<regexp_elem> lst; | |
1466 string_vector named; | |
1467 int nopts; | |
7893
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
1468 bool once; |
eb9ccb44ea41
make regexp(...,'once') matlab compatible
Jaroslav Hajek <highegg@gmail.com>
parents:
7520
diff
changeset
|
1469 int sz = octregexp_list (regexpargs, nm, false, lst, named, nopts, once); |
5785 | 1470 |
1471 if (error_state) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1472 return retval; |
5785 | 1473 if (sz == 0) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1474 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1475 retval = args(0); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1476 return retval; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1477 } |
5785 | 1478 |
1479 // Determine replacement length | |
1480 const size_t replen = replacement.size(); | |
1481 int delta = 0; | |
1482 const_iterator p = lst.begin(); | |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1483 for (int i = 0; i < sz; i++) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1484 { |
5785 | 1485 OCTAVE_QUIT; |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1486 delta += static_cast<int>(replen) - |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1487 static_cast<int>(p->e - p->s + 1); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1488 p++; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1489 } |
5785 | 1490 |
1491 // Build replacement string | |
1492 rep.reserve(buffer.size()+delta); | |
1493 size_t from = 0; | |
1494 p = lst.begin(); | |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1495 for (int i=0; i < sz; i++) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1496 { |
5785 | 1497 OCTAVE_QUIT; |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1498 rep.append(&buffer[from], static_cast<size_t>(p->s - 1) - from); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1499 from = static_cast<size_t>(p->e - 1) + 1; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1500 rep.append(replacement); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1501 p++; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1502 } |
5785 | 1503 rep.append(&buffer[from],buffer.size()-from); |
1504 } | |
11586
12df7854fa7c
strip trailing whitespace from source files
John W. Eaton <jwe@octave.org>
parents:
11572
diff
changeset
|
1505 |
6361 | 1506 retval = rep; |
1507 return retval; | |
1508 } | |
1509 | |
6549 | 1510 DEFUN_DLD (regexprep, args, , |
6361 | 1511 "-*- texinfo -*-\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1512 @deftypefn {Loadable Function} {@var{outstr} =} regexprep (@var{string}, @var{pat}, @var{repstr})\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1513 @deftypefnx {Loadable Function} {@var{outstr} =} regexprep (@var{string}, @var{pat}, @var{repstr}, \"@var{opt1}\", @dots{})\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1514 Replace occurrences of pattern @var{pat} in @var{string} with @var{repstr}.\n\ |
6361 | 1515 \n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1516 The pattern is a regular expression as documented for @code{regexp}.\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1517 @xref{doc-regexp,,regexp}.\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1518 \n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1519 The replacement string may contain @code{$i}, which substitutes\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1520 for the ith set of parentheses in the match string. For example,\n\ |
10840 | 1521 \n\ |
6361 | 1522 @example\n\ |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1523 regexprep(\"Bill Dunn\",'(\\w+) (\\w+)','$2, $1')\n\ |
6361 | 1524 @end example\n\ |
10840 | 1525 \n\ |
10846
a4f482e66b65
Grammarcheck more of the documentation.
Rik <octave@nomad.inbox5.com>
parents:
10840
diff
changeset
|
1526 @noindent\n\ |
6361 | 1527 returns \"Dunn, Bill\"\n\ |
1528 \n\ | |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1529 Options in addition to those of @code{regexp} are\n\ |
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1530 \n\ |
6361 | 1531 @table @samp\n\ |
1532 \n\ | |
1533 @item once\n\ | |
7001 | 1534 Replace only the first occurrence of @var{pat} in the result.\n\ |
6361 | 1535 \n\ |
1536 @item warnings\n\ | |
1537 This option is present for compatibility but is ignored.\n\ | |
1538 \n\ | |
1539 @end table\n\ | |
11572
7d6d8c1e471f
Grammarcheck Texinfo for files in src directory.
Rik <octave@nomad.inbox5.com>
parents:
11553
diff
changeset
|
1540 @seealso{regexp, regexpi, strrep}\n\ |
6361 | 1541 @end deftypefn") |
1542 { | |
1543 octave_value_list retval; | |
1544 int nargin = args.length(); | |
1545 | |
1546 if (nargin < 3) | |
1547 { | |
1548 print_usage (); | |
1549 return retval; | |
1550 } | |
1551 | |
1552 if (args(0).is_cell() || args(1).is_cell() || args(2).is_cell()) | |
1553 { | |
1554 Cell str; | |
1555 Cell pat; | |
1556 Cell rep; | |
6495 | 1557 dim_vector dv0; |
1558 dim_vector dv1(1,1); | |
6361 | 1559 |
1560 if (args(0).is_cell()) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1561 str = args(0).cell_value(); |
6361 | 1562 else |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1563 str = Cell (args(0)); |
6361 | 1564 |
1565 if (args(1).is_cell()) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1566 pat = args(1).cell_value(); |
6361 | 1567 else |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1568 pat = Cell (args(1)); |
6361 | 1569 |
1570 if (args(2).is_cell()) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1571 rep = args(2).cell_value(); |
6361 | 1572 else |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1573 rep = Cell (args(2)); |
6361 | 1574 |
6495 | 1575 dv0 = str.dims(); |
1576 if (pat.numel() != 1) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1577 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1578 dv1 = pat.dims(); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1579 if (rep.numel() != 1 && dv1 != rep.dims()) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1580 error ("regexprep: Inconsistent cell array dimensions"); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1581 } |
6361 | 1582 else if (rep.numel() != 1) |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1583 dv1 = rep.dims(); |
6361 | 1584 |
1585 if (!error_state) | |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1586 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1587 Cell ret (dv0); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1588 octave_value_list new_args = args; |
6361 | 1589 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1590 for (octave_idx_type i = 0; i < dv0.numel(); i++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1591 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1592 new_args(0) = str(i); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1593 if (pat.numel() == 1) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1594 new_args(1) = pat(0); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1595 if (rep.numel() == 1) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1596 new_args(2) = rep(0); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1597 for (octave_idx_type j = 0; j < dv1.numel(); j++) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1598 { |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1599 if (pat.numel() != 1) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1600 new_args(1) = pat(j); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1601 if (rep.numel() != 1) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1602 new_args(2) = rep(j); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1603 new_args(0) = octregexprep (new_args, "regexprep"); |
6361 | 1604 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1605 if (error_state) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1606 break; |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1607 } |
6361 | 1608 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1609 if (error_state) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1610 break; |
6495 | 1611 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1612 ret(i) = new_args(0); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1613 } |
6361 | 1614 |
10154
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1615 if (!error_state) |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1616 retval = octave_value (ret); |
40dfc0c99116
DLD-FUNCTIONS/*.cc: untabify
John W. Eaton <jwe@octave.org>
parents:
9064
diff
changeset
|
1617 } |
6361 | 1618 } |
1619 else | |
1620 retval = octregexprep (args, "regexprep"); | |
1621 | |
5785 | 1622 return retval; |
1623 } | |
1624 | |
1625 /* | |
1626 %!test # Replace with empty | |
1627 %! xml = '<!-- This is some XML --> <tag v="hello">some stuff<!-- sample tag--></tag>'; | |
1628 %! t = regexprep(xml,'<[!?][^>]*>',''); | |
1629 %! assert(t,' <tag v="hello">some stuff</tag>') | |
1630 | |
1631 %!test # Replace with non-empty | |
1632 %! xml = '<!-- This is some XML --> <tag v="hello">some stuff<!-- sample tag--></tag>'; | |
1633 %! t = regexprep(xml,'<[!?][^>]*>','?'); | |
1634 %! assert(t,'? <tag v="hello">some stuff?</tag>') | |
1635 | |
1636 %!test # Check that 'tokenize' is ignored | |
1637 %! xml = '<!-- This is some XML --> <tag v="hello">some stuff<!-- sample tag--></tag>'; | |
1638 %! t = regexprep(xml,'<[!?][^>]*>','','tokenize'); | |
1639 %! assert(t,' <tag v="hello">some stuff</tag>') | |
1640 | |
11032
c9b0a75b02e8
Make all regexp in Octave compatible with both POSIX and PCRE.
Rik <octave@nomad.inbox5.com>
parents:
11025
diff
changeset
|
1641 ## Test capture replacement |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
1642 %!test |
7242 | 1643 %! data = "Bob Smith\nDavid Hollerith\nSam Jenkins"; |
1644 %! result = "Smith, Bob\nHollerith, David\nJenkins, Sam"; | |
1645 %! t = regexprep(data,'(?m)^(\w+)\s+(\w+)$','$2, $1'); | |
1646 %! assert(t,result) | |
5785 | 1647 |
11032
c9b0a75b02e8
Make all regexp in Octave compatible with both POSIX and PCRE.
Rik <octave@nomad.inbox5.com>
parents:
11025
diff
changeset
|
1648 ## Return the original if no match |
5785 | 1649 %!assert(regexprep('hello','world','earth'),'hello') |
1650 | |
1651 ## Test a general replacement | |
1652 %!assert(regexprep("a[b]c{d}e-f=g", "[^A-Za-z0-9_]", "_"), "a_b_c_d_e_f_g"); | |
1653 | |
1654 ## Make sure it works at the beginning and end | |
1655 %!assert(regexprep("a[b]c{d}e-f=g", "a", "_"), "_[b]c{d}e-f=g"); | |
1656 %!assert(regexprep("a[b]c{d}e-f=g", "g", "_"), "a[b]c{d}e-f=_"); | |
1657 | |
1658 ## Options | |
1659 %!assert(regexprep("a[b]c{d}e-f=g", "[^A-Za-z0-9_]", "_", "once"), "a_b]c{d}e-f=g"); | |
1660 %!assert(regexprep("a[b]c{d}e-f=g", "[^A-Z0-9_]", "_", "ignorecase"), "a_b_c_d_e_f_g"); | |
1661 | |
1662 ## Option combinations | |
1663 %!assert(regexprep("a[b]c{d}e-f=g", "[^A-Z0-9_]", "_", "once", "ignorecase"), "a_b]c{d}e-f=g"); | |
1664 | |
1665 ## End conditions on replacement | |
1666 %!assert(regexprep("abc","(b)",".$1"),"a.bc"); | |
1667 %!assert(regexprep("abc","(b)","$1"),"abc"); | |
1668 %!assert(regexprep("abc","(b)","$1."),"ab.c"); | |
1669 %!assert(regexprep("abc","(b)","$1.."),"ab..c"); | |
1670 | |
6361 | 1671 ## Test cell array arguments |
6503 | 1672 %!assert(regexprep("abc",{"b","a"},"?"),{"??c"}) |
6361 | 1673 %!assert(regexprep({"abc","cba"},"b","?"),{"a?c","c?a"}) |
6503 | 1674 %!assert(regexprep({"abc","cba"},{"b","a"},{"?","!"}),{"!?c","c?!"}) |
6361 | 1675 |
8093
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
1676 # Nasty lookbehind expression |
12462
e4dbfe3019b1
Use PCRE regular expressions throughout Octave.
Rik <octave@nomad.inbox5.com>
parents:
12433
diff
changeset
|
1677 %!test |
11025
df2152514429
Update docstrings for regular expression functions
Rik <octave@nomad.inbox5.com>
parents:
11018
diff
changeset
|
1678 %! assert(regexprep('x^(-1)+y(-1)+z(-1)=0','(?<=[a-z]+)\(\-[1-9]*\)','_minus1'),'x^(-1)+y_minus1+z_minus1=0') |
8093
dcc31f473596
Treat PCRE lookbehind operators in a manner that is approximately correct
David Bateman <dbateman@free.fr>
parents:
8021
diff
changeset
|
1679 |
5785 | 1680 */ |