/[pcre]/code/trunk/doc/pcre.txt
ViewVC logotype

Contents of /code/trunk/doc/pcre.txt

Parent Directory Parent Directory | Revision Log Revision Log


Revision 96 - (show annotations) (download)
Fri Mar 2 13:10:43 2007 UTC (7 years, 8 months ago) by nigel
File MIME type: text/plain
File size: 263470 byte(s)
 r6896@hex:  nm | 2007-03-02 13:09:14 +0000
 Added EOL and keywork properties throughout

1 -----------------------------------------------------------------------------
2 This file contains a concatenation of the PCRE man pages, converted to plain
3 text format for ease of searching with a text editor, or for use on systems
4 that do not have a man page processor. The small individual files that give
5 synopses of each function in the library have not been included. There are
6 separate text files for the pcregrep and pcretest commands.
7 -----------------------------------------------------------------------------
8
9
10 PCRE(3) PCRE(3)
11
12
13 NAME
14 PCRE - Perl-compatible regular expressions
15
16
17 INTRODUCTION
18
19 The PCRE library is a set of functions that implement regular expres-
20 sion pattern matching using the same syntax and semantics as Perl, with
21 just a few differences. (Certain features that appeared in Python and
22 PCRE before they appeared in Perl are also available using the Python
23 syntax.)
24
25 The current implementation of PCRE (release 7.x) corresponds approxi-
26 mately with Perl 5.10, including support for UTF-8 encoded strings and
27 Unicode general category properties. However, UTF-8 and Unicode support
28 has to be explicitly enabled; it is not the default. The Unicode tables
29 correspond to Unicode release 5.0.0.
30
31 In addition to the Perl-compatible matching function, PCRE contains an
32 alternative matching function that matches the same compiled patterns
33 in a different way. In certain circumstances, the alternative function
34 has some advantages. For a discussion of the two matching algorithms,
35 see the pcrematching page.
36
37 PCRE is written in C and released as a C library. A number of people
38 have written wrappers and interfaces of various kinds. In particular,
39 Google Inc. have provided a comprehensive C++ wrapper. This is now
40 included as part of the PCRE distribution. The pcrecpp page has details
41 of this interface. Other people's contributions can be found in the
42 Contrib directory at the primary FTP site, which is:
43
44 ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre
45
46 Details of exactly which Perl regular expression features are and are
47 not supported by PCRE are given in separate documents. See the pcrepat-
48 tern and pcrecompat pages.
49
50 Some features of PCRE can be included, excluded, or changed when the
51 library is built. The pcre_config() function makes it possible for a
52 client to discover which features are available. The features them-
53 selves are described in the pcrebuild page. Documentation about build-
54 ing PCRE for various operating systems can be found in the README file
55 in the source distribution.
56
57 The library contains a number of undocumented internal functions and
58 data tables that are used by more than one of the exported external
59 functions, but which are not intended for use by external callers.
60 Their names all begin with "_pcre_", which hopefully will not provoke
61 any name clashes. In some environments, it is possible to control which
62 external symbols are exported when a shared library is built, and in
63 these cases the undocumented symbols are not exported.
64
65
66 USER DOCUMENTATION
67
68 The user documentation for PCRE comprises a number of different sec-
69 tions. In the "man" format, each of these is a separate "man page". In
70 the HTML format, each is a separate page, linked from the index page.
71 In the plain text format, all the sections are concatenated, for ease
72 of searching. The sections are as follows:
73
74 pcre this document
75 pcreapi details of PCRE's native C API
76 pcrebuild options for building PCRE
77 pcrecallout details of the callout feature
78 pcrecompat discussion of Perl compatibility
79 pcrecpp details of the C++ wrapper
80 pcregrep description of the pcregrep command
81 pcrematching discussion of the two matching algorithms
82 pcrepartial details of the partial matching facility
83 pcrepattern syntax and semantics of supported
84 regular expressions
85 pcreperform discussion of performance issues
86 pcreposix the POSIX-compatible C API
87 pcreprecompile details of saving and re-using precompiled patterns
88 pcresample discussion of the sample program
89 pcrestack discussion of stack usage
90 pcretest description of the pcretest testing command
91
92 In addition, in the "man" and HTML formats, there is a short page for
93 each C library function, listing its arguments and results.
94
95
96 LIMITATIONS
97
98 There are some size limitations in PCRE but it is hoped that they will
99 never in practice be relevant.
100
101 The maximum length of a compiled pattern is 65539 (sic) bytes if PCRE
102 is compiled with the default internal linkage size of 2. If you want to
103 process regular expressions that are truly enormous, you can compile
104 PCRE with an internal linkage size of 3 or 4 (see the README file in
105 the source distribution and the pcrebuild documentation for details).
106 In these cases the limit is substantially larger. However, the speed
107 of execution is slower.
108
109 All values in repeating quantifiers must be less than 65536. The maxi-
110 mum compiled length of subpattern with an explicit repeat count is
111 30000 bytes. The maximum number of capturing subpatterns is 65535.
112
113 There is no limit to the number of parenthesized subpatterns, but there
114 can be no more than 65535 capturing subpatterns.
115
116 The maximum length of name for a named subpattern is 32 characters, and
117 the maximum number of named subpatterns is 10000.
118
119 The maximum length of a subject string is the largest positive number
120 that an integer variable can hold. However, when using the traditional
121 matching function, PCRE uses recursion to handle subpatterns and indef-
122 inite repetition. This means that the available stack space may limit
123 the size of a subject string that can be processed by certain patterns.
124 For a discussion of stack issues, see the pcrestack documentation.
125
126
127 UTF-8 AND UNICODE PROPERTY SUPPORT
128
129 From release 3.3, PCRE has had some support for character strings
130 encoded in the UTF-8 format. For release 4.0 this was greatly extended
131 to cover most common requirements, and in release 5.0 additional sup-
132 port for Unicode general category properties was added.
133
134 In order process UTF-8 strings, you must build PCRE to include UTF-8
135 support in the code, and, in addition, you must call pcre_compile()
136 with the PCRE_UTF8 option flag. When you do this, both the pattern and
137 any subject strings that are matched against it are treated as UTF-8
138 strings instead of just strings of bytes.
139
140 If you compile PCRE with UTF-8 support, but do not use it at run time,
141 the library will be a bit bigger, but the additional run time overhead
142 is limited to testing the PCRE_UTF8 flag occasionally, so should not be
143 very big.
144
145 If PCRE is built with Unicode character property support (which implies
146 UTF-8 support), the escape sequences \p{..}, \P{..}, and \X are sup-
147 ported. The available properties that can be tested are limited to the
148 general category properties such as Lu for an upper case letter or Nd
149 for a decimal number, the Unicode script names such as Arabic or Han,
150 and the derived properties Any and L&. A full list is given in the
151 pcrepattern documentation. Only the short names for properties are sup-
152 ported. For example, \p{L} matches a letter. Its Perl synonym, \p{Let-
153 ter}, is not supported. Furthermore, in Perl, many properties may
154 optionally be prefixed by "Is", for compatibility with Perl 5.6. PCRE
155 does not support this.
156
157 The following comments apply when PCRE is running in UTF-8 mode:
158
159 1. When you set the PCRE_UTF8 flag, the strings passed as patterns and
160 subjects are checked for validity on entry to the relevant functions.
161 If an invalid UTF-8 string is passed, an error return is given. In some
162 situations, you may already know that your strings are valid, and
163 therefore want to skip these checks in order to improve performance. If
164 you set the PCRE_NO_UTF8_CHECK flag at compile time or at run time,
165 PCRE assumes that the pattern or subject it is given (respectively)
166 contains only valid UTF-8 codes. In this case, it does not diagnose an
167 invalid UTF-8 string. If you pass an invalid UTF-8 string to PCRE when
168 PCRE_NO_UTF8_CHECK is set, the results are undefined. Your program may
169 crash.
170
171 2. An unbraced hexadecimal escape sequence (such as \xb3) matches a
172 two-byte UTF-8 character if the value is greater than 127.
173
174 3. Octal numbers up to \777 are recognized, and match two-byte UTF-8
175 characters for values greater than \177.
176
177 4. Repeat quantifiers apply to complete UTF-8 characters, not to indi-
178 vidual bytes, for example: \x{100}{3}.
179
180 5. The dot metacharacter matches one UTF-8 character instead of a sin-
181 gle byte.
182
183 6. The escape sequence \C can be used to match a single byte in UTF-8
184 mode, but its use can lead to some strange effects. This facility is
185 not available in the alternative matching function, pcre_dfa_exec().
186
187 7. The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly
188 test characters of any code value, but the characters that PCRE recog-
189 nizes as digits, spaces, or word characters remain the same set as
190 before, all with values less than 256. This remains true even when PCRE
191 includes Unicode property support, because to do otherwise would slow
192 down PCRE in many common cases. If you really want to test for a wider
193 sense of, say, "digit", you must use Unicode property tests such as
194 \p{Nd}.
195
196 8. Similarly, characters that match the POSIX named character classes
197 are all low-valued characters.
198
199 9. Case-insensitive matching applies only to characters whose values
200 are less than 128, unless PCRE is built with Unicode property support.
201 Even when Unicode property support is available, PCRE still uses its
202 own character tables when checking the case of low-valued characters,
203 so as not to degrade performance. The Unicode property information is
204 used only for characters with higher values. Even when Unicode property
205 support is available, PCRE supports case-insensitive matching only when
206 there is a one-to-one mapping between a letter's cases. There are a
207 small number of many-to-one mappings in Unicode; these are not sup-
208 ported by PCRE.
209
210
211 AUTHOR
212
213 Philip Hazel
214 University Computing Service,
215 Cambridge CB2 3QH, England.
216
217 Putting an actual email address here seems to have been a spam magnet,
218 so I've taken it away. If you want to email me, use my initial and sur-
219 name, separated by a dot, at the domain ucs.cam.ac.uk.
220
221 Last updated: 23 November 2006
222 Copyright (c) 1997-2006 University of Cambridge.
223 ------------------------------------------------------------------------------
224
225
226 PCREBUILD(3) PCREBUILD(3)
227
228
229 NAME
230 PCRE - Perl-compatible regular expressions
231
232
233 PCRE BUILD-TIME OPTIONS
234
235 This document describes the optional features of PCRE that can be
236 selected when the library is compiled. They are all selected, or dese-
237 lected, by providing options to the configure script that is run before
238 the make command. The complete list of options for configure (which
239 includes the standard ones such as the selection of the installation
240 directory) can be obtained by running
241
242 ./configure --help
243
244 The following sections describe certain options whose names begin with
245 --enable or --disable. These settings specify changes to the defaults
246 for the configure command. Because of the way that configure works,
247 --enable and --disable always come in pairs, so the complementary
248 option always exists as well, but as it specifies the default, it is
249 not described.
250
251
252 C++ SUPPORT
253
254 By default, the configure script will search for a C++ compiler and C++
255 header files. If it finds them, it automatically builds the C++ wrapper
256 library for PCRE. You can disable this by adding
257
258 --disable-cpp
259
260 to the configure command.
261
262
263 UTF-8 SUPPORT
264
265 To build PCRE with support for UTF-8 character strings, add
266
267 --enable-utf8
268
269 to the configure command. Of itself, this does not make PCRE treat
270 strings as UTF-8. As well as compiling PCRE with this option, you also
271 have have to set the PCRE_UTF8 option when you call the pcre_compile()
272 function.
273
274
275 UNICODE CHARACTER PROPERTY SUPPORT
276
277 UTF-8 support allows PCRE to process character values greater than 255
278 in the strings that it handles. On its own, however, it does not pro-
279 vide any facilities for accessing the properties of such characters. If
280 you want to be able to use the pattern escapes \P, \p, and \X, which
281 refer to Unicode character properties, you must add
282
283 --enable-unicode-properties
284
285 to the configure command. This implies UTF-8 support, even if you have
286 not explicitly requested it.
287
288 Including Unicode property support adds around 90K of tables to the
289 PCRE library, approximately doubling its size. Only the general cate-
290 gory properties such as Lu and Nd are supported. Details are given in
291 the pcrepattern documentation.
292
293
294 CODE VALUE OF NEWLINE
295
296 By default, PCRE interprets character 10 (linefeed, LF) as indicating
297 the end of a line. This is the normal newline character on Unix-like
298 systems. You can compile PCRE to use character 13 (carriage return, CR)
299 instead, by adding
300
301 --enable-newline-is-cr
302
303 to the configure command. There is also a --enable-newline-is-lf
304 option, which explicitly specifies linefeed as the newline character.
305
306 Alternatively, you can specify that line endings are to be indicated by
307 the two character sequence CRLF. If you want this, add
308
309 --enable-newline-is-crlf
310
311 to the configure command. There is a fourth option, specified by
312
313 --enable-newline-is-any
314
315 which causes PCRE to recognize any Unicode newline sequence.
316
317 Whatever line ending convention is selected when PCRE is built can be
318 overridden when the library functions are called. At build time it is
319 conventional to use the standard for your operating system.
320
321
322 BUILDING SHARED AND STATIC LIBRARIES
323
324 The PCRE building process uses libtool to build both shared and static
325 Unix libraries by default. You can suppress one of these by adding one
326 of
327
328 --disable-shared
329 --disable-static
330
331 to the configure command, as required.
332
333
334 POSIX MALLOC USAGE
335
336 When PCRE is called through the POSIX interface (see the pcreposix doc-
337 umentation), additional working storage is required for holding the
338 pointers to capturing substrings, because PCRE requires three integers
339 per substring, whereas the POSIX interface provides only two. If the
340 number of expected substrings is small, the wrapper function uses space
341 on the stack, because this is faster than using malloc() for each call.
342 The default threshold above which the stack is no longer used is 10; it
343 can be changed by adding a setting such as
344
345 --with-posix-malloc-threshold=20
346
347 to the configure command.
348
349
350 HANDLING VERY LARGE PATTERNS
351
352 Within a compiled pattern, offset values are used to point from one
353 part to another (for example, from an opening parenthesis to an alter-
354 nation metacharacter). By default, two-byte values are used for these
355 offsets, leading to a maximum size for a compiled pattern of around
356 64K. This is sufficient to handle all but the most gigantic patterns.
357 Nevertheless, some people do want to process enormous patterns, so it
358 is possible to compile PCRE to use three-byte or four-byte offsets by
359 adding a setting such as
360
361 --with-link-size=3
362
363 to the configure command. The value given must be 2, 3, or 4. Using
364 longer offsets slows down the operation of PCRE because it has to load
365 additional bytes when handling them.
366
367 If you build PCRE with an increased link size, test 2 (and test 5 if
368 you are using UTF-8) will fail. Part of the output of these tests is a
369 representation of the compiled pattern, and this changes with the link
370 size.
371
372
373 AVOIDING EXCESSIVE STACK USAGE
374
375 When matching with the pcre_exec() function, PCRE implements backtrack-
376 ing by making recursive calls to an internal function called match().
377 In environments where the size of the stack is limited, this can se-
378 verely limit PCRE's operation. (The Unix environment does not usually
379 suffer from this problem, but it may sometimes be necessary to increase
380 the maximum stack size. There is a discussion in the pcrestack docu-
381 mentation.) An alternative approach to recursion that uses memory from
382 the heap to remember data, instead of using recursive function calls,
383 has been implemented to work round the problem of limited stack size.
384 If you want to build a version of PCRE that works this way, add
385
386 --disable-stack-for-recursion
387
388 to the configure command. With this configuration, PCRE will use the
389 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
390 ment functions. Separate functions are provided because the usage is
391 very predictable: the block sizes requested are always the same, and
392 the blocks are always freed in reverse order. A calling program might
393 be able to implement optimized functions that perform better than the
394 standard malloc() and free() functions. PCRE runs noticeably more
395 slowly when built in this way. This option affects only the pcre_exec()
396 function; it is not relevant for the the pcre_dfa_exec() function.
397
398
399 LIMITING PCRE RESOURCE USAGE
400
401 Internally, PCRE has a function called match(), which it calls repeat-
402 edly (sometimes recursively) when matching a pattern with the
403 pcre_exec() function. By controlling the maximum number of times this
404 function may be called during a single matching operation, a limit can
405 be placed on the resources used by a single call to pcre_exec(). The
406 limit can be changed at run time, as described in the pcreapi documen-
407 tation. The default is 10 million, but this can be changed by adding a
408 setting such as
409
410 --with-match-limit=500000
411
412 to the configure command. This setting has no effect on the
413 pcre_dfa_exec() matching function.
414
415 In some environments it is desirable to limit the depth of recursive
416 calls of match() more strictly than the total number of calls, in order
417 to restrict the maximum amount of stack (or heap, if --disable-stack-
418 for-recursion is specified) that is used. A second limit controls this;
419 it defaults to the value that is set for --with-match-limit, which
420 imposes no additional constraints. However, you can set a lower limit
421 by adding, for example,
422
423 --with-match-limit-recursion=10000
424
425 to the configure command. This value can also be overridden at run
426 time.
427
428
429 USING EBCDIC CODE
430
431 PCRE assumes by default that it will run in an environment where the
432 character code is ASCII (or Unicode, which is a superset of ASCII).
433 PCRE can, however, be compiled to run in an EBCDIC environment by
434 adding
435
436 --enable-ebcdic
437
438 to the configure command.
439
440
441 SEE ALSO
442
443 pcreapi(3), pcre_config(3).
444
445 Last updated: 30 November 2006
446 Copyright (c) 1997-2006 University of Cambridge.
447 ------------------------------------------------------------------------------
448
449
450 PCREMATCHING(3) PCREMATCHING(3)
451
452
453 NAME
454 PCRE - Perl-compatible regular expressions
455
456
457 PCRE MATCHING ALGORITHMS
458
459 This document describes the two different algorithms that are available
460 in PCRE for matching a compiled regular expression against a given sub-
461 ject string. The "standard" algorithm is the one provided by the
462 pcre_exec() function. This works in the same was as Perl's matching
463 function, and provides a Perl-compatible matching operation.
464
465 An alternative algorithm is provided by the pcre_dfa_exec() function;
466 this operates in a different way, and is not Perl-compatible. It has
467 advantages and disadvantages compared with the standard algorithm, and
468 these are described below.
469
470 When there is only one possible way in which a given subject string can
471 match a pattern, the two algorithms give the same answer. A difference
472 arises, however, when there are multiple possibilities. For example, if
473 the pattern
474
475 ^<.*>
476
477 is matched against the string
478
479 <something> <something else> <something further>
480
481 there are three possible answers. The standard algorithm finds only one
482 of them, whereas the alternative algorithm finds all three.
483
484
485 REGULAR EXPRESSIONS AS TREES
486
487 The set of strings that are matched by a regular expression can be rep-
488 resented as a tree structure. An unlimited repetition in the pattern
489 makes the tree of infinite size, but it is still a tree. Matching the
490 pattern to a given subject string (from a given starting point) can be
491 thought of as a search of the tree. There are two ways to search a
492 tree: depth-first and breadth-first, and these correspond to the two
493 matching algorithms provided by PCRE.
494
495
496 THE STANDARD MATCHING ALGORITHM
497
498 In the terminology of Jeffrey Friedl's book Mastering Regular Expres-
499 sions, the standard algorithm is an "NFA algorithm". It conducts a
500 depth-first search of the pattern tree. That is, it proceeds along a
501 single path through the tree, checking that the subject matches what is
502 required. When there is a mismatch, the algorithm tries any alterna-
503 tives at the current point, and if they all fail, it backs up to the
504 previous branch point in the tree, and tries the next alternative
505 branch at that level. This often involves backing up (moving to the
506 left) in the subject string as well. The order in which repetition
507 branches are tried is controlled by the greedy or ungreedy nature of
508 the quantifier.
509
510 If a leaf node is reached, a matching string has been found, and at
511 that point the algorithm stops. Thus, if there is more than one possi-
512 ble match, this algorithm returns the first one that it finds. Whether
513 this is the shortest, the longest, or some intermediate length depends
514 on the way the greedy and ungreedy repetition quantifiers are specified
515 in the pattern.
516
517 Because it ends up with a single path through the tree, it is rela-
518 tively straightforward for this algorithm to keep track of the sub-
519 strings that are matched by portions of the pattern in parentheses.
520 This provides support for capturing parentheses and back references.
521
522
523 THE ALTERNATIVE MATCHING ALGORITHM
524
525 This algorithm conducts a breadth-first search of the tree. Starting
526 from the first matching point in the subject, it scans the subject
527 string from left to right, once, character by character, and as it does
528 this, it remembers all the paths through the tree that represent valid
529 matches. In Friedl's terminology, this is a kind of "DFA algorithm",
530 though it is not implemented as a traditional finite state machine (it
531 keeps multiple states active simultaneously).
532
533 The scan continues until either the end of the subject is reached, or
534 there are no more unterminated paths. At this point, terminated paths
535 represent the different matching possibilities (if there are none, the
536 match has failed). Thus, if there is more than one possible match,
537 this algorithm finds all of them, and in particular, it finds the long-
538 est. In PCRE, there is an option to stop the algorithm after the first
539 match (which is necessarily the shortest) has been found.
540
541 Note that all the matches that are found start at the same point in the
542 subject. If the pattern
543
544 cat(er(pillar)?)
545
546 is matched against the string "the caterpillar catchment", the result
547 will be the three strings "cat", "cater", and "caterpillar" that start
548 at the fourth character of the subject. The algorithm does not automat-
549 ically move on to find matches that start at later positions.
550
551 There are a number of features of PCRE regular expressions that are not
552 supported by the alternative matching algorithm. They are as follows:
553
554 1. Because the algorithm finds all possible matches, the greedy or
555 ungreedy nature of repetition quantifiers is not relevant. Greedy and
556 ungreedy quantifiers are treated in exactly the same way. However, pos-
557 sessive quantifiers can make a difference when what follows could also
558 match what is quantified, for example in a pattern like this:
559
560 ^a++\w!
561
562 This pattern matches "aaab!" but not "aaa!", which would be matched by
563 a non-possessive quantifier. Similarly, if an atomic group is present,
564 it is matched as if it were a standalone pattern at the current point,
565 and the longest match is then "locked in" for the rest of the overall
566 pattern.
567
568 2. When dealing with multiple paths through the tree simultaneously, it
569 is not straightforward to keep track of captured substrings for the
570 different matching possibilities, and PCRE's implementation of this
571 algorithm does not attempt to do this. This means that no captured sub-
572 strings are available.
573
574 3. Because no substrings are captured, back references within the pat-
575 tern are not supported, and cause errors if encountered.
576
577 4. For the same reason, conditional expressions that use a backrefer-
578 ence as the condition or test for a specific group recursion are not
579 supported.
580
581 5. Callouts are supported, but the value of the capture_top field is
582 always 1, and the value of the capture_last field is always -1.
583
584 6. The \C escape sequence, which (in the standard algorithm) matches a
585 single byte, even in UTF-8 mode, is not supported because the alterna-
586 tive algorithm moves through the subject string one character at a
587 time, for all active paths through the tree.
588
589
590 ADVANTAGES OF THE ALTERNATIVE ALGORITHM
591
592 Using the alternative matching algorithm provides the following advan-
593 tages:
594
595 1. All possible matches (at a single point in the subject) are automat-
596 ically found, and in particular, the longest match is found. To find
597 more than one match using the standard algorithm, you have to do kludgy
598 things with callouts.
599
600 2. There is much better support for partial matching. The restrictions
601 on the content of the pattern that apply when using the standard algo-
602 rithm for partial matching do not apply to the alternative algorithm.
603 For non-anchored patterns, the starting position of a partial match is
604 available.
605
606 3. Because the alternative algorithm scans the subject string just
607 once, and never needs to backtrack, it is possible to pass very long
608 subject strings to the matching function in several pieces, checking
609 for partial matching each time.
610
611
612 DISADVANTAGES OF THE ALTERNATIVE ALGORITHM
613
614 The alternative algorithm suffers from a number of disadvantages:
615
616 1. It is substantially slower than the standard algorithm. This is
617 partly because it has to search for all possible matches, but is also
618 because it is less susceptible to optimization.
619
620 2. Capturing parentheses and back references are not supported.
621
622 3. Although atomic groups are supported, their use does not provide the
623 performance advantage that it does for the standard algorithm.
624
625 Last updated: 24 November 2006
626 Copyright (c) 1997-2006 University of Cambridge.
627 ------------------------------------------------------------------------------
628
629
630 PCREAPI(3) PCREAPI(3)
631
632
633 NAME
634 PCRE - Perl-compatible regular expressions
635
636
637 PCRE NATIVE API
638
639 #include <pcre.h>
640
641 pcre *pcre_compile(const char *pattern, int options,
642 const char **errptr, int *erroffset,
643 const unsigned char *tableptr);
644
645 pcre *pcre_compile2(const char *pattern, int options,
646 int *errorcodeptr,
647 const char **errptr, int *erroffset,
648 const unsigned char *tableptr);
649
650 pcre_extra *pcre_study(const pcre *code, int options,
651 const char **errptr);
652
653 int pcre_exec(const pcre *code, const pcre_extra *extra,
654 const char *subject, int length, int startoffset,
655 int options, int *ovector, int ovecsize);
656
657 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
658 const char *subject, int length, int startoffset,
659 int options, int *ovector, int ovecsize,
660 int *workspace, int wscount);
661
662 int pcre_copy_named_substring(const pcre *code,
663 const char *subject, int *ovector,
664 int stringcount, const char *stringname,
665 char *buffer, int buffersize);
666
667 int pcre_copy_substring(const char *subject, int *ovector,
668 int stringcount, int stringnumber, char *buffer,
669 int buffersize);
670
671 int pcre_get_named_substring(const pcre *code,
672 const char *subject, int *ovector,
673 int stringcount, const char *stringname,
674 const char **stringptr);
675
676 int pcre_get_stringnumber(const pcre *code,
677 const char *name);
678
679 int pcre_get_stringtable_entries(const pcre *code,
680 const char *name, char **first, char **last);
681
682 int pcre_get_substring(const char *subject, int *ovector,
683 int stringcount, int stringnumber,
684 const char **stringptr);
685
686 int pcre_get_substring_list(const char *subject,
687 int *ovector, int stringcount, const char ***listptr);
688
689 void pcre_free_substring(const char *stringptr);
690
691 void pcre_free_substring_list(const char **stringptr);
692
693 const unsigned char *pcre_maketables(void);
694
695 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
696 int what, void *where);
697
698 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
699
700 int pcre_refcount(pcre *code, int adjust);
701
702 int pcre_config(int what, void *where);
703
704 char *pcre_version(void);
705
706 void *(*pcre_malloc)(size_t);
707
708 void (*pcre_free)(void *);
709
710 void *(*pcre_stack_malloc)(size_t);
711
712 void (*pcre_stack_free)(void *);
713
714 int (*pcre_callout)(pcre_callout_block *);
715
716
717 PCRE API OVERVIEW
718
719 PCRE has its own native API, which is described in this document. There
720 are also some wrapper functions that correspond to the POSIX regular
721 expression API. These are described in the pcreposix documentation.
722 Both of these APIs define a set of C function calls. A C++ wrapper is
723 distributed with PCRE. It is documented in the pcrecpp page.
724
725 The native API C function prototypes are defined in the header file
726 pcre.h, and on Unix systems the library itself is called libpcre. It
727 can normally be accessed by adding -lpcre to the command for linking an
728 application that uses PCRE. The header file defines the macros
729 PCRE_MAJOR and PCRE_MINOR to contain the major and minor release num-
730 bers for the library. Applications can use these to include support
731 for different releases of PCRE.
732
733 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
734 pcre_exec() are used for compiling and matching regular expressions in
735 a Perl-compatible manner. A sample program that demonstrates the sim-
736 plest way of using them is provided in the file called pcredemo.c in
737 the source distribution. The pcresample documentation describes how to
738 run it.
739
740 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
741 ble, is also provided. This uses a different algorithm for the match-
742 ing. The alternative algorithm finds all possible matches (at a given
743 point in the subject), and scans the subject just once. However, this
744 algorithm does not return captured substrings. A description of the two
745 matching algorithms and their advantages and disadvantages is given in
746 the pcrematching documentation.
747
748 In addition to the main compiling and matching functions, there are
749 convenience functions for extracting captured substrings from a subject
750 string that is matched by pcre_exec(). They are:
751
752 pcre_copy_substring()
753 pcre_copy_named_substring()
754 pcre_get_substring()
755 pcre_get_named_substring()
756 pcre_get_substring_list()
757 pcre_get_stringnumber()
758 pcre_get_stringtable_entries()
759
760 pcre_free_substring() and pcre_free_substring_list() are also provided,
761 to free the memory used for extracted strings.
762
763 The function pcre_maketables() is used to build a set of character
764 tables in the current locale for passing to pcre_compile(),
765 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
766 provided for specialist use. Most commonly, no special tables are
767 passed, in which case internal tables that are generated when PCRE is
768 built are used.
769
770 The function pcre_fullinfo() is used to find out information about a
771 compiled pattern; pcre_info() is an obsolete version that returns only
772 some of the available information, but is retained for backwards com-
773 patibility. The function pcre_version() returns a pointer to a string
774 containing the version of PCRE and its date of release.
775
776 The function pcre_refcount() maintains a reference count in a data
777 block containing a compiled pattern. This is provided for the benefit
778 of object-oriented applications.
779
780 The global variables pcre_malloc and pcre_free initially contain the
781 entry points of the standard malloc() and free() functions, respec-
782 tively. PCRE calls the memory management functions via these variables,
783 so a calling program can replace them if it wishes to intercept the
784 calls. This should be done before calling any PCRE functions.
785
786 The global variables pcre_stack_malloc and pcre_stack_free are also
787 indirections to memory management functions. These special functions
788 are used only when PCRE is compiled to use the heap for remembering
789 data, instead of recursive function calls, when running the pcre_exec()
790 function. See the pcrebuild documentation for details of how to do
791 this. It is a non-standard way of building PCRE, for use in environ-
792 ments that have limited stacks. Because of the greater use of memory
793 management, it runs more slowly. Separate functions are provided so
794 that special-purpose external code can be used for this case. When
795 used, these functions are always called in a stack-like manner (last
796 obtained, first freed), and always for memory blocks of the same size.
797 There is a discussion about PCRE's stack usage in the pcrestack docu-
798 mentation.
799
800 The global variable pcre_callout initially contains NULL. It can be set
801 by the caller to a "callout" function, which PCRE will then call at
802 specified points during a matching operation. Details are given in the
803 pcrecallout documentation.
804
805
806 NEWLINES
807
808 PCRE supports four different conventions for indicating line breaks in
809 strings: a single CR (carriage return) character, a single LF (line-
810 feed) character, the two-character sequence CRLF, or any Unicode new-
811 line sequence. The Unicode newline sequences are the three just men-
812 tioned, plus the single characters VT (vertical tab, U+000B), FF (form-
813 feed, U+000C), NEL (next line, U+0085), LS (line separator, U+2028),
814 and PS (paragraph separator, U+2029).
815
816 Each of the first three conventions is used by at least one operating
817 system as its standard newline sequence. When PCRE is built, a default
818 can be specified. The default default is LF, which is the Unix stan-
819 dard. When PCRE is run, the default can be overridden, either when a
820 pattern is compiled, or when it is matched.
821
822 In the PCRE documentation the word "newline" is used to mean "the char-
823 acter or pair of characters that indicate a line break". The choice of
824 newline convention affects the handling of the dot, circumflex, and
825 dollar metacharacters, the handling of #-comments in /x mode, and, when
826 CRLF is a recognized line ending sequence, the match position advance-
827 ment for a non-anchored pattern. The choice of newline convention does
828 not affect the interpretation of the \n or \r escape sequences.
829
830
831 MULTITHREADING
832
833 The PCRE functions can be used in multi-threading applications, with
834 the proviso that the memory management functions pointed to by
835 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
836 callout function pointed to by pcre_callout, are shared by all threads.
837
838 The compiled form of a regular expression is not altered during match-
839 ing, so the same compiled pattern can safely be used by several threads
840 at once.
841
842
843 SAVING PRECOMPILED PATTERNS FOR LATER USE
844
845 The compiled form of a regular expression can be saved and re-used at a
846 later time, possibly by a different program, and even on a host other
847 than the one on which it was compiled. Details are given in the
848 pcreprecompile documentation.
849
850
851 CHECKING BUILD-TIME OPTIONS
852
853 int pcre_config(int what, void *where);
854
855 The function pcre_config() makes it possible for a PCRE client to dis-
856 cover which optional features have been compiled into the PCRE library.
857 The pcrebuild documentation has more details about these optional fea-
858 tures.
859
860 The first argument for pcre_config() is an integer, specifying which
861 information is required; the second argument is a pointer to a variable
862 into which the information is placed. The following information is
863 available:
864
865 PCRE_CONFIG_UTF8
866
867 The output is an integer that is set to one if UTF-8 support is avail-
868 able; otherwise it is set to zero.
869
870 PCRE_CONFIG_UNICODE_PROPERTIES
871
872 The output is an integer that is set to one if support for Unicode
873 character properties is available; otherwise it is set to zero.
874
875 PCRE_CONFIG_NEWLINE
876
877 The output is an integer whose value specifies the default character
878 sequence that is recognized as meaning "newline". The four values that
879 are supported are: 10 for LF, 13 for CR, 3338 for CRLF, and -1 for ANY.
880 The default should normally be the standard sequence for your operating
881 system.
882
883 PCRE_CONFIG_LINK_SIZE
884
885 The output is an integer that contains the number of bytes used for
886 internal linkage in compiled regular expressions. The value is 2, 3, or
887 4. Larger values allow larger regular expressions to be compiled, at
888 the expense of slower matching. The default value of 2 is sufficient
889 for all but the most massive patterns, since it allows the compiled
890 pattern to be up to 64K in size.
891
892 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
893
894 The output is an integer that contains the threshold above which the
895 POSIX interface uses malloc() for output vectors. Further details are
896 given in the pcreposix documentation.
897
898 PCRE_CONFIG_MATCH_LIMIT
899
900 The output is an integer that gives the default limit for the number of
901 internal matching function calls in a pcre_exec() execution. Further
902 details are given with pcre_exec() below.
903
904 PCRE_CONFIG_MATCH_LIMIT_RECURSION
905
906 The output is an integer that gives the default limit for the depth of
907 recursion when calling the internal matching function in a pcre_exec()
908 execution. Further details are given with pcre_exec() below.
909
910 PCRE_CONFIG_STACKRECURSE
911
912 The output is an integer that is set to one if internal recursion when
913 running pcre_exec() is implemented by recursive function calls that use
914 the stack to remember their state. This is the usual way that PCRE is
915 compiled. The output is zero if PCRE was compiled to use blocks of data
916 on the heap instead of recursive function calls. In this case,
917 pcre_stack_malloc and pcre_stack_free are called to manage memory
918 blocks on the heap, thus avoiding the use of the stack.
919
920
921 COMPILING A PATTERN
922
923 pcre *pcre_compile(const char *pattern, int options,
924 const char **errptr, int *erroffset,
925 const unsigned char *tableptr);
926
927 pcre *pcre_compile2(const char *pattern, int options,
928 int *errorcodeptr,
929 const char **errptr, int *erroffset,
930 const unsigned char *tableptr);
931
932 Either of the functions pcre_compile() or pcre_compile2() can be called
933 to compile a pattern into an internal form. The only difference between
934 the two interfaces is that pcre_compile2() has an additional argument,
935 errorcodeptr, via which a numerical error code can be returned.
936
937 The pattern is a C string terminated by a binary zero, and is passed in
938 the pattern argument. A pointer to a single block of memory that is
939 obtained via pcre_malloc is returned. This contains the compiled code
940 and related data. The pcre type is defined for the returned block; this
941 is a typedef for a structure whose contents are not externally defined.
942 It is up to the caller to free the memory (via pcre_free) when it is no
943 longer required.
944
945 Although the compiled code of a PCRE regex is relocatable, that is, it
946 does not depend on memory location, the complete pcre data block is not
947 fully relocatable, because it may contain a copy of the tableptr argu-
948 ment, which is an address (see below).
949
950 The options argument contains various bit settings that affect the com-
951 pilation. It should be zero if no options are required. The available
952 options are described below. Some of them, in particular, those that
953 are compatible with Perl, can also be set and unset from within the
954 pattern (see the detailed description in the pcrepattern documenta-
955 tion). For these options, the contents of the options argument speci-
956 fies their initial settings at the start of compilation and execution.
957 The PCRE_ANCHORED and PCRE_NEWLINE_xxx options can be set at the time
958 of matching as well as at compile time.
959
960 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
961 if compilation of a pattern fails, pcre_compile() returns NULL, and
962 sets the variable pointed to by errptr to point to a textual error mes-
963 sage. This is a static string that is part of the library. You must not
964 try to free it. The offset from the start of the pattern to the charac-
965 ter where the error was discovered is placed in the variable pointed to
966 by erroffset, which must not be NULL. If it is, an immediate error is
967 given.
968
969 If pcre_compile2() is used instead of pcre_compile(), and the error-
970 codeptr argument is not NULL, a non-zero error code number is returned
971 via this argument in the event of an error. This is in addition to the
972 textual error message. Error codes and messages are listed below.
973
974 If the final argument, tableptr, is NULL, PCRE uses a default set of
975 character tables that are built when PCRE is compiled, using the
976 default C locale. Otherwise, tableptr must be an address that is the
977 result of a call to pcre_maketables(). This value is stored with the
978 compiled pattern, and used again by pcre_exec(), unless another table
979 pointer is passed to it. For more discussion, see the section on locale
980 support below.
981
982 This code fragment shows a typical straightforward call to pcre_com-
983 pile():
984
985 pcre *re;
986 const char *error;
987 int erroffset;
988 re = pcre_compile(
989 "^A.*Z", /* the pattern */
990 0, /* default options */
991 &error, /* for error message */
992 &erroffset, /* for error offset */
993 NULL); /* use default character tables */
994
995 The following names for option bits are defined in the pcre.h header
996 file:
997
998 PCRE_ANCHORED
999
1000 If this bit is set, the pattern is forced to be "anchored", that is, it
1001 is constrained to match only at the first matching point in the string
1002 that is being searched (the "subject string"). This effect can also be
1003 achieved by appropriate constructs in the pattern itself, which is the
1004 only way to do it in Perl.
1005
1006 PCRE_AUTO_CALLOUT
1007
1008 If this bit is set, pcre_compile() automatically inserts callout items,
1009 all with number 255, before each pattern item. For discussion of the
1010 callout facility, see the pcrecallout documentation.
1011
1012 PCRE_CASELESS
1013
1014 If this bit is set, letters in the pattern match both upper and lower
1015 case letters. It is equivalent to Perl's /i option, and it can be
1016 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
1017 always understands the concept of case for characters whose values are
1018 less than 128, so caseless matching is always possible. For characters
1019 with higher values, the concept of case is supported if PCRE is com-
1020 piled with Unicode property support, but not otherwise. If you want to
1021 use caseless matching for characters 128 and above, you must ensure
1022 that PCRE is compiled with Unicode property support as well as with
1023 UTF-8 support.
1024
1025 PCRE_DOLLAR_ENDONLY
1026
1027 If this bit is set, a dollar metacharacter in the pattern matches only
1028 at the end of the subject string. Without this option, a dollar also
1029 matches immediately before a newline at the end of the string (but not
1030 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
1031 if PCRE_MULTILINE is set. There is no equivalent to this option in
1032 Perl, and no way to set it within a pattern.
1033
1034 PCRE_DOTALL
1035
1036 If this bit is set, a dot metacharater in the pattern matches all char-
1037 acters, including those that indicate newline. Without it, a dot does
1038 not match when the current position is at a newline. This option is
1039 equivalent to Perl's /s option, and it can be changed within a pattern
1040 by a (?s) option setting. A negative class such as [^a] always matches
1041 newline characters, independent of the setting of this option.
1042
1043 PCRE_DUPNAMES
1044
1045 If this bit is set, names used to identify capturing subpatterns need
1046 not be unique. This can be helpful for certain types of pattern when it
1047 is known that only one instance of the named subpattern can ever be
1048 matched. There are more details of named subpatterns below; see also
1049 the pcrepattern documentation.
1050
1051 PCRE_EXTENDED
1052
1053 If this bit is set, whitespace data characters in the pattern are
1054 totally ignored except when escaped or inside a character class. White-
1055 space does not include the VT character (code 11). In addition, charac-
1056 ters between an unescaped # outside a character class and the next new-
1057 line, inclusive, are also ignored. This is equivalent to Perl's /x
1058 option, and it can be changed within a pattern by a (?x) option set-
1059 ting.
1060
1061 This option makes it possible to include comments inside complicated
1062 patterns. Note, however, that this applies only to data characters.
1063 Whitespace characters may never appear within special character
1064 sequences in a pattern, for example within the sequence (?( which
1065 introduces a conditional subpattern.
1066
1067 PCRE_EXTRA
1068
1069 This option was invented in order to turn on additional functionality
1070 of PCRE that is incompatible with Perl, but it is currently of very
1071 little use. When set, any backslash in a pattern that is followed by a
1072 letter that has no special meaning causes an error, thus reserving
1073 these combinations for future expansion. By default, as in Perl, a
1074 backslash followed by a letter with no special meaning is treated as a
1075 literal. (Perl can, however, be persuaded to give a warning for this.)
1076 There are at present no other features controlled by this option. It
1077 can also be set by a (?X) option setting within a pattern.
1078
1079 PCRE_FIRSTLINE
1080
1081 If this option is set, an unanchored pattern is required to match
1082 before or at the first newline in the subject string, though the
1083 matched text may continue over the newline.
1084
1085 PCRE_MULTILINE
1086
1087 By default, PCRE treats the subject string as consisting of a single
1088 line of characters (even if it actually contains newlines). The "start
1089 of line" metacharacter (^) matches only at the start of the string,
1090 while the "end of line" metacharacter ($) matches only at the end of
1091 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1092 is set). This is the same as Perl.
1093
1094 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1095 constructs match immediately following or immediately before internal
1096 newlines in the subject string, respectively, as well as at the very
1097 start and end. This is equivalent to Perl's /m option, and it can be
1098 changed within a pattern by a (?m) option setting. If there are no new-
1099 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1100 setting PCRE_MULTILINE has no effect.
1101
1102 PCRE_NEWLINE_CR
1103 PCRE_NEWLINE_LF
1104 PCRE_NEWLINE_CRLF
1105 PCRE_NEWLINE_ANY
1106
1107 These options override the default newline definition that was chosen
1108 when PCRE was built. Setting the first or the second specifies that a
1109 newline is indicated by a single character (CR or LF, respectively).
1110 Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
1111 two-character CRLF sequence. Setting PCRE_NEWLINE_ANY specifies that
1112 any Unicode newline sequence should be recognized. The Unicode newline
1113 sequences are the three just mentioned, plus the single characters VT
1114 (vertical tab, U+000B), FF (formfeed, U+000C), NEL (next line, U+0085),
1115 LS (line separator, U+2028), and PS (paragraph separator, U+2029). The
1116 last two are recognized only in UTF-8 mode.
1117
1118 The newline setting in the options word uses three bits that are
1119 treated as a number, giving eight possibilities. Currently only five
1120 are used (default plus the four values above). This means that if you
1121 set more than one newline option, the combination may or may not be
1122 sensible. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equiva-
1123 lent to PCRE_NEWLINE_CRLF, but other combinations yield unused numbers
1124 and cause an error.
1125
1126 The only time that a line break is specially recognized when compiling
1127 a pattern is if PCRE_EXTENDED is set, and an unescaped # outside a
1128 character class is encountered. This indicates a comment that lasts
1129 until after the next line break sequence. In other circumstances, line
1130 break sequences are treated as literal data, except that in
1131 PCRE_EXTENDED mode, both CR and LF are treated as whitespace characters
1132 and are therefore ignored.
1133
1134 The newline option that is set at compile time becomes the default that
1135 is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1136
1137 PCRE_NO_AUTO_CAPTURE
1138
1139 If this option is set, it disables the use of numbered capturing paren-
1140 theses in the pattern. Any opening parenthesis that is not followed by
1141 ? behaves as if it were followed by ?: but named parentheses can still
1142 be used for capturing (and they acquire numbers in the usual way).
1143 There is no equivalent of this option in Perl.
1144
1145 PCRE_UNGREEDY
1146
1147 This option inverts the "greediness" of the quantifiers so that they
1148 are not greedy by default, but become greedy if followed by "?". It is
1149 not compatible with Perl. It can also be set by a (?U) option setting
1150 within the pattern.
1151
1152 PCRE_UTF8
1153
1154 This option causes PCRE to regard both the pattern and the subject as
1155 strings of UTF-8 characters instead of single-byte character strings.
1156 However, it is available only when PCRE is built to include UTF-8 sup-
1157 port. If not, the use of this option provokes an error. Details of how
1158 this option changes the behaviour of PCRE are given in the section on
1159 UTF-8 support in the main pcre page.
1160
1161 PCRE_NO_UTF8_CHECK
1162
1163 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1164 automatically checked. If an invalid UTF-8 sequence of bytes is found,
1165 pcre_compile() returns an error. If you already know that your pattern
1166 is valid, and you want to skip this check for performance reasons, you
1167 can set the PCRE_NO_UTF8_CHECK option. When it is set, the effect of
1168 passing an invalid UTF-8 string as a pattern is undefined. It may cause
1169 your program to crash. Note that this option can also be passed to
1170 pcre_exec() and pcre_dfa_exec(), to suppress the UTF-8 validity check-
1171 ing of subject strings.
1172
1173
1174 COMPILATION ERROR CODES
1175
1176 The following table lists the error codes than may be returned by
1177 pcre_compile2(), along with the error messages that may be returned by
1178 both compiling functions. As PCRE has developed, some error codes have
1179 fallen out of use. To avoid confusion, they have not been re-used.
1180
1181 0 no error
1182 1 \ at end of pattern
1183 2 \c at end of pattern
1184 3 unrecognized character follows \
1185 4 numbers out of order in {} quantifier
1186 5 number too big in {} quantifier
1187 6 missing terminating ] for character class
1188 7 invalid escape sequence in character class
1189 8 range out of order in character class
1190 9 nothing to repeat
1191 10 [this code is not in use]
1192 11 internal error: unexpected repeat
1193 12 unrecognized character after (?
1194 13 POSIX named classes are supported only within a class
1195 14 missing )
1196 15 reference to non-existent subpattern
1197 16 erroffset passed as NULL
1198 17 unknown option bit(s) set
1199 18 missing ) after comment
1200 19 [this code is not in use]
1201 20 regular expression too large
1202 21 failed to get memory
1203 22 unmatched parentheses
1204 23 internal error: code overflow
1205 24 unrecognized character after (?<
1206 25 lookbehind assertion is not fixed length
1207 26 malformed number or name after (?(
1208 27 conditional group contains more than two branches
1209 28 assertion expected after (?(
1210 29 (?R or (?digits must be followed by )
1211 30 unknown POSIX class name
1212 31 POSIX collating elements are not supported
1213 32 this version of PCRE is not compiled with PCRE_UTF8 support
1214 33 [this code is not in use]
1215 34 character value in \x{...} sequence is too large
1216 35 invalid condition (?(0)
1217 36 \C not allowed in lookbehind assertion
1218 37 PCRE does not support \L, \l, \N, \U, or \u
1219 38 number after (?C is > 255
1220 39 closing ) for (?C expected
1221 40 recursive call could loop indefinitely
1222 41 unrecognized character after (?P
1223 42 syntax error in subpattern name (missing terminator)
1224 43 two named subpatterns have the same name
1225 44 invalid UTF-8 string
1226 45 support for \P, \p, and \X has not been compiled
1227 46 malformed \P or \p sequence
1228 47 unknown property name after \P or \p
1229 48 subpattern name is too long (maximum 32 characters)
1230 49 too many named subpatterns (maximum 10,000)
1231 50 repeated subpattern is too long
1232 51 octal value is greater than \377 (not in UTF-8 mode)
1233 52 internal error: overran compiling workspace
1234 53 internal error: previously-checked referenced subpattern not
1235 found
1236 54 DEFINE group contains more than one branch
1237 55 repeating a DEFINE group is not allowed
1238 56 inconsistent NEWLINE options"
1239
1240
1241 STUDYING A PATTERN
1242
1243 pcre_extra *pcre_study(const pcre *code, int options
1244 const char **errptr);
1245
1246 If a compiled pattern is going to be used several times, it is worth
1247 spending more time analyzing it in order to speed up the time taken for
1248 matching. The function pcre_study() takes a pointer to a compiled pat-
1249 tern as its first argument. If studying the pattern produces additional
1250 information that will help speed up matching, pcre_study() returns a
1251 pointer to a pcre_extra block, in which the study_data field points to
1252 the results of the study.
1253
1254 The returned value from pcre_study() can be passed directly to
1255 pcre_exec(). However, a pcre_extra block also contains other fields
1256 that can be set by the caller before the block is passed; these are
1257 described below in the section on matching a pattern.
1258
1259 If studying the pattern does not produce any additional information
1260 pcre_study() returns NULL. In that circumstance, if the calling program
1261 wants to pass any of the other fields to pcre_exec(), it must set up
1262 its own pcre_extra block.
1263
1264 The second argument of pcre_study() contains option bits. At present,
1265 no options are defined, and this argument should always be zero.
1266
1267 The third argument for pcre_study() is a pointer for an error message.
1268 If studying succeeds (even if no data is returned), the variable it
1269 points to is set to NULL. Otherwise it is set to point to a textual
1270 error message. This is a static string that is part of the library. You
1271 must not try to free it. You should test the error pointer for NULL
1272 after calling pcre_study(), to be sure that it has run successfully.
1273
1274 This is a typical call to pcre_study():
1275
1276 pcre_extra *pe;
1277 pe = pcre_study(
1278 re, /* result of pcre_compile() */
1279 0, /* no options exist */
1280 &error); /* set to NULL or points to a message */
1281
1282 At present, studying a pattern is useful only for non-anchored patterns
1283 that do not have a single fixed starting character. A bitmap of possi-
1284 ble starting bytes is created.
1285
1286
1287 LOCALE SUPPORT
1288
1289 PCRE handles caseless matching, and determines whether characters are
1290 letters digits, or whatever, by reference to a set of tables, indexed
1291 by character value. When running in UTF-8 mode, this applies only to
1292 characters with codes less than 128. Higher-valued codes never match
1293 escapes such as \w or \d, but can be tested with \p if PCRE is built
1294 with Unicode character property support. The use of locales with Uni-
1295 code is discouraged.
1296
1297 An internal set of tables is created in the default C locale when PCRE
1298 is built. This is used when the final argument of pcre_compile() is
1299 NULL, and is sufficient for many applications. An alternative set of
1300 tables can, however, be supplied. These may be created in a different
1301 locale from the default. As more and more applications change to using
1302 Unicode, the need for this locale support is expected to die away.
1303
1304 External tables are built by calling the pcre_maketables() function,
1305 which has no arguments, in the relevant locale. The result can then be
1306 passed to pcre_compile() or pcre_exec() as often as necessary. For
1307 example, to build and use tables that are appropriate for the French
1308 locale (where accented characters with values greater than 128 are
1309 treated as letters), the following code could be used:
1310
1311 setlocale(LC_CTYPE, "fr_FR");
1312 tables = pcre_maketables();
1313 re = pcre_compile(..., tables);
1314
1315 When pcre_maketables() runs, the tables are built in memory that is
1316 obtained via pcre_malloc. It is the caller's responsibility to ensure
1317 that the memory containing the tables remains available for as long as
1318 it is needed.
1319
1320 The pointer that is passed to pcre_compile() is saved with the compiled
1321 pattern, and the same tables are used via this pointer by pcre_study()
1322 and normally also by pcre_exec(). Thus, by default, for any single pat-
1323 tern, compilation, studying and matching all happen in the same locale,
1324 but different patterns can be compiled in different locales.
1325
1326 It is possible to pass a table pointer or NULL (indicating the use of
1327 the internal tables) to pcre_exec(). Although not intended for this
1328 purpose, this facility could be used to match a pattern in a different
1329 locale from the one in which it was compiled. Passing table pointers at
1330 run time is discussed below in the section on matching a pattern.
1331
1332
1333 INFORMATION ABOUT A PATTERN
1334
1335 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1336 int what, void *where);
1337
1338 The pcre_fullinfo() function returns information about a compiled pat-
1339 tern. It replaces the obsolete pcre_info() function, which is neverthe-
1340 less retained for backwards compability (and is documented below).
1341
1342 The first argument for pcre_fullinfo() is a pointer to the compiled
1343 pattern. The second argument is the result of pcre_study(), or NULL if
1344 the pattern was not studied. The third argument specifies which piece
1345 of information is required, and the fourth argument is a pointer to a
1346 variable to receive the data. The yield of the function is zero for
1347 success, or one of the following negative numbers:
1348
1349 PCRE_ERROR_NULL the argument code was NULL
1350 the argument where was NULL
1351 PCRE_ERROR_BADMAGIC the "magic number" was not found
1352 PCRE_ERROR_BADOPTION the value of what was invalid
1353
1354 The "magic number" is placed at the start of each compiled pattern as
1355 an simple check against passing an arbitrary memory pointer. Here is a
1356 typical call of pcre_fullinfo(), to obtain the length of the compiled
1357 pattern:
1358
1359 int rc;
1360 size_t length;
1361 rc = pcre_fullinfo(
1362 re, /* result of pcre_compile() */
1363 pe, /* result of pcre_study(), or NULL */
1364 PCRE_INFO_SIZE, /* what is required */
1365 &length); /* where to put the data */
1366
1367 The possible values for the third argument are defined in pcre.h, and
1368 are as follows:
1369
1370 PCRE_INFO_BACKREFMAX
1371
1372 Return the number of the highest back reference in the pattern. The
1373 fourth argument should point to an int variable. Zero is returned if
1374 there are no back references.
1375
1376 PCRE_INFO_CAPTURECOUNT
1377
1378 Return the number of capturing subpatterns in the pattern. The fourth
1379 argument should point to an int variable.
1380
1381 PCRE_INFO_DEFAULT_TABLES
1382
1383 Return a pointer to the internal default character tables within PCRE.
1384 The fourth argument should point to an unsigned char * variable. This
1385 information call is provided for internal use by the pcre_study() func-
1386 tion. External callers can cause PCRE to use its internal tables by
1387 passing a NULL table pointer.
1388
1389 PCRE_INFO_FIRSTBYTE
1390
1391 Return information about the first byte of any matched string, for a
1392 non-anchored pattern. The fourth argument should point to an int vari-
1393 able. (This option used to be called PCRE_INFO_FIRSTCHAR; the old name
1394 is still recognized for backwards compatibility.)
1395
1396 If there is a fixed first byte, for example, from a pattern such as
1397 (cat|cow|coyote), its value is returned. Otherwise, if either
1398
1399 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
1400 branch starts with "^", or
1401
1402 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
1403 set (if it were set, the pattern would be anchored),
1404
1405 -1 is returned, indicating that the pattern matches only at the start
1406 of a subject string or after any newline within the string. Otherwise
1407 -2 is returned. For anchored patterns, -2 is returned.
1408
1409 PCRE_INFO_FIRSTTABLE
1410
1411 If the pattern was studied, and this resulted in the construction of a
1412 256-bit table indicating a fixed set of bytes for the first byte in any
1413 matching string, a pointer to the table is returned. Otherwise NULL is
1414 returned. The fourth argument should point to an unsigned char * vari-
1415 able.
1416
1417 PCRE_INFO_LASTLITERAL
1418
1419 Return the value of the rightmost literal byte that must exist in any
1420 matched string, other than at its start, if such a byte has been
1421 recorded. The fourth argument should point to an int variable. If there
1422 is no such byte, -1 is returned. For anchored patterns, a last literal
1423 byte is recorded only if it follows something of variable length. For
1424 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
1425 /^a\dz\d/ the returned value is -1.
1426
1427 PCRE_INFO_NAMECOUNT
1428 PCRE_INFO_NAMEENTRYSIZE
1429 PCRE_INFO_NAMETABLE
1430
1431 PCRE supports the use of named as well as numbered capturing parenthe-
1432 ses. The names are just an additional way of identifying the parenthe-
1433 ses, which still acquire numbers. Several convenience functions such as
1434 pcre_get_named_substring() are provided for extracting captured sub-
1435 strings by name. It is also possible to extract the data directly, by
1436 first converting the name to a number in order to access the correct
1437 pointers in the output vector (described with pcre_exec() below). To do
1438 the conversion, you need to use the name-to-number map, which is
1439 described by these three values.
1440
1441 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
1442 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
1443 of each entry; both of these return an int value. The entry size
1444 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
1445 a pointer to the first entry of the table (a pointer to char). The
1446 first two bytes of each entry are the number of the capturing parenthe-
1447 sis, most significant byte first. The rest of the entry is the corre-
1448 sponding name, zero terminated. The names are in alphabetical order.
1449 When PCRE_DUPNAMES is set, duplicate names are in order of their paren-
1450 theses numbers. For example, consider the following pattern (assume
1451 PCRE_EXTENDED is set, so white space - including newlines - is
1452 ignored):
1453
1454 (?<date> (?<year>(\d\d)?\d\d) -
1455 (?<month>\d\d) - (?<day>\d\d) )
1456
1457 There are four named subpatterns, so the table has four entries, and
1458 each entry in the table is eight bytes long. The table is as follows,
1459 with non-printing bytes shows in hexadecimal, and undefined bytes shown
1460 as ??:
1461
1462 00 01 d a t e 00 ??
1463 00 05 d a y 00 ?? ??
1464 00 04 m o n t h 00
1465 00 02 y e a r 00 ??
1466
1467 When writing code to extract data from named subpatterns using the
1468 name-to-number map, remember that the length of the entries is likely
1469 to be different for each compiled pattern.
1470
1471 PCRE_INFO_OPTIONS
1472
1473 Return a copy of the options with which the pattern was compiled. The
1474 fourth argument should point to an unsigned long int variable. These
1475 option bits are those specified in the call to pcre_compile(), modified
1476 by any top-level option settings within the pattern itself.
1477
1478 A pattern is automatically anchored by PCRE if all of its top-level
1479 alternatives begin with one of the following:
1480
1481 ^ unless PCRE_MULTILINE is set
1482 \A always
1483 \G always
1484 .* if PCRE_DOTALL is set and there are no back
1485 references to the subpattern in which .* appears
1486
1487 For such patterns, the PCRE_ANCHORED bit is set in the options returned
1488 by pcre_fullinfo().
1489
1490 PCRE_INFO_SIZE
1491
1492 Return the size of the compiled pattern, that is, the value that was
1493 passed as the argument to pcre_malloc() when PCRE was getting memory in
1494 which to place the compiled data. The fourth argument should point to a
1495 size_t variable.
1496
1497 PCRE_INFO_STUDYSIZE
1498
1499 Return the size of the data block pointed to by the study_data field in
1500 a pcre_extra block. That is, it is the value that was passed to
1501 pcre_malloc() when PCRE was getting memory into which to place the data
1502 created by pcre_study(). The fourth argument should point to a size_t
1503 variable.
1504
1505
1506 OBSOLETE INFO FUNCTION
1507
1508 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
1509
1510 The pcre_info() function is now obsolete because its interface is too
1511 restrictive to return all the available data about a compiled pattern.
1512 New programs should use pcre_fullinfo() instead. The yield of
1513 pcre_info() is the number of capturing subpatterns, or one of the fol-
1514 lowing negative numbers:
1515
1516 PCRE_ERROR_NULL the argument code was NULL
1517 PCRE_ERROR_BADMAGIC the "magic number" was not found
1518
1519 If the optptr argument is not NULL, a copy of the options with which
1520 the pattern was compiled is placed in the integer it points to (see
1521 PCRE_INFO_OPTIONS above).
1522
1523 If the pattern is not anchored and the firstcharptr argument is not
1524 NULL, it is used to pass back information about the first character of
1525 any matched string (see PCRE_INFO_FIRSTBYTE above).
1526
1527
1528 REFERENCE COUNTS
1529
1530 int pcre_refcount(pcre *code, int adjust);
1531
1532 The pcre_refcount() function is used to maintain a reference count in
1533 the data block that contains a compiled pattern. It is provided for the
1534 benefit of applications that operate in an object-oriented manner,
1535 where different parts of the application may be using the same compiled
1536 pattern, but you want to free the block when they are all done.
1537
1538 When a pattern is compiled, the reference count field is initialized to
1539 zero. It is changed only by calling this function, whose action is to
1540 add the adjust value (which may be positive or negative) to it. The
1541 yield of the function is the new value. However, the value of the count
1542 is constrained to lie between 0 and 65535, inclusive. If the new value
1543 is outside these limits, it is forced to the appropriate limit value.
1544
1545 Except when it is zero, the reference count is not correctly preserved
1546 if a pattern is compiled on one host and then transferred to a host
1547 whose byte-order is different. (This seems a highly unlikely scenario.)
1548
1549
1550 MATCHING A PATTERN: THE TRADITIONAL FUNCTION
1551
1552 int pcre_exec(const pcre *code, const pcre_extra *extra,
1553 const char *subject, int length, int startoffset,
1554 int options, int *ovector, int ovecsize);
1555
1556 The function pcre_exec() is called to match a subject string against a
1557 compiled pattern, which is passed in the code argument. If the pattern
1558 has been studied, the result of the study should be passed in the extra
1559 argument. This function is the main matching facility of the library,
1560 and it operates in a Perl-like manner. For specialist use there is also
1561 an alternative matching function, which is described below in the sec-
1562 tion about the pcre_dfa_exec() function.
1563
1564 In most applications, the pattern will have been compiled (and option-
1565 ally studied) in the same process that calls pcre_exec(). However, it
1566 is possible to save compiled patterns and study data, and then use them
1567 later in different processes, possibly even on different hosts. For a
1568 discussion about this, see the pcreprecompile documentation.
1569
1570 Here is an example of a simple call to pcre_exec():
1571
1572 int rc;
1573 int ovector[30];
1574 rc = pcre_exec(
1575 re, /* result of pcre_compile() */
1576 NULL, /* we didn't study the pattern */
1577 "some string", /* the subject string */
1578 11, /* the length of the subject string */
1579 0, /* start at offset 0 in the subject */
1580 0, /* default options */
1581 ovector, /* vector of integers for substring information */
1582 30); /* number of elements (NOT size in bytes) */
1583
1584 Extra data for pcre_exec()
1585
1586 If the extra argument is not NULL, it must point to a pcre_extra data
1587 block. The pcre_study() function returns such a block (when it doesn't
1588 return NULL), but you can also create one for yourself, and pass addi-
1589 tional information in it. The pcre_extra block contains the following
1590 fields (not necessarily in this order):
1591
1592 unsigned long int flags;
1593 void *study_data;
1594 unsigned long int match_limit;
1595 unsigned long int match_limit_recursion;
1596 void *callout_data;
1597 const unsigned char *tables;
1598
1599 The flags field is a bitmap that specifies which of the other fields
1600 are set. The flag bits are:
1601
1602 PCRE_EXTRA_STUDY_DATA
1603 PCRE_EXTRA_MATCH_LIMIT
1604 PCRE_EXTRA_MATCH_LIMIT_RECURSION
1605 PCRE_EXTRA_CALLOUT_DATA
1606 PCRE_EXTRA_TABLES
1607
1608 Other flag bits should be set to zero. The study_data field is set in
1609 the pcre_extra block that is returned by pcre_study(), together with
1610 the appropriate flag bit. You should not set this yourself, but you may
1611 add to the block by setting the other fields and their corresponding
1612 flag bits.
1613
1614 The match_limit field provides a means of preventing PCRE from using up
1615 a vast amount of resources when running patterns that are not going to
1616 match, but which have a very large number of possibilities in their
1617 search trees. The classic example is the use of nested unlimited
1618 repeats.
1619
1620 Internally, PCRE uses a function called match() which it calls repeat-
1621 edly (sometimes recursively). The limit set by match_limit is imposed
1622 on the number of times this function is called during a match, which
1623 has the effect of limiting the amount of backtracking that can take
1624 place. For patterns that are not anchored, the count restarts from zero
1625 for each position in the subject string.
1626
1627 The default value for the limit can be set when PCRE is built; the
1628 default default is 10 million, which handles all but the most extreme
1629 cases. You can override the default by suppling pcre_exec() with a
1630 pcre_extra block in which match_limit is set, and
1631 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
1632 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
1633
1634 The match_limit_recursion field is similar to match_limit, but instead
1635 of limiting the total number of times that match() is called, it limits
1636 the depth of recursion. The recursion depth is a smaller number than
1637 the total number of calls, because not all calls to match() are recur-
1638 sive. This limit is of use only if it is set smaller than match_limit.
1639
1640 Limiting the recursion depth limits the amount of stack that can be
1641 used, or, when PCRE has been compiled to use memory on the heap instead
1642 of the stack, the amount of heap memory that can be used.
1643
1644 The default value for match_limit_recursion can be set when PCRE is
1645 built; the default default is the same value as the default for
1646 match_limit. You can override the default by suppling pcre_exec() with
1647 a pcre_extra block in which match_limit_recursion is set, and
1648 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
1649 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
1650
1651 The pcre_callout field is used in conjunction with the "callout" fea-
1652 ture, which is described in the pcrecallout documentation.
1653
1654 The tables field is used to pass a character tables pointer to
1655 pcre_exec(); this overrides the value that is stored with the compiled
1656 pattern. A non-NULL value is stored with the compiled pattern only if
1657 custom tables were supplied to pcre_compile() via its tableptr argu-
1658 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
1659 PCRE's internal tables to be used. This facility is helpful when re-
1660 using patterns that have been saved after compiling with an external
1661 set of tables, because the external tables might be at a different
1662 address when pcre_exec() is called. See the pcreprecompile documenta-
1663 tion for a discussion of saving compiled patterns for later use.
1664
1665 Option bits for pcre_exec()
1666
1667 The unused bits of the options argument for pcre_exec() must be zero.
1668 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
1669 PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK and
1670 PCRE_PARTIAL.
1671
1672 PCRE_ANCHORED
1673
1674 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
1675 matching position. If a pattern was compiled with PCRE_ANCHORED, or
1676 turned out to be anchored by virtue of its contents, it cannot be made
1677 unachored at matching time.
1678
1679 PCRE_NEWLINE_CR
1680 PCRE_NEWLINE_LF
1681 PCRE_NEWLINE_CRLF
1682 PCRE_NEWLINE_ANY
1683
1684 These options override the newline definition that was chosen or
1685 defaulted when the pattern was compiled. For details, see the descrip-
1686 tion of pcre_compile() above. During matching, the newline choice
1687 affects the behaviour of the dot, circumflex, and dollar metacharac-
1688 ters. It may also alter the way the match position is advanced after a
1689 match failure for an unanchored pattern. When PCRE_NEWLINE_CRLF or
1690 PCRE_NEWLINE_ANY is set, and a match attempt fails when the current
1691 position is at a CRLF sequence, the match position is advanced by two
1692 characters instead of one, in other words, to after the CRLF.
1693
1694 PCRE_NOTBOL
1695
1696 This option specifies that first character of the subject string is not
1697 the beginning of a line, so the circumflex metacharacter should not
1698 match before it. Setting this without PCRE_MULTILINE (at compile time)
1699 causes circumflex never to match. This option affects only the behav-
1700 iour of the circumflex metacharacter. It does not affect \A.
1701
1702 PCRE_NOTEOL
1703
1704 This option specifies that the end of the subject string is not the end
1705 of a line, so the dollar metacharacter should not match it nor (except
1706 in multiline mode) a newline immediately before it. Setting this with-
1707 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
1708 option affects only the behaviour of the dollar metacharacter. It does
1709 not affect \Z or \z.
1710
1711 PCRE_NOTEMPTY
1712
1713 An empty string is not considered to be a valid match if this option is
1714 set. If there are alternatives in the pattern, they are tried. If all
1715 the alternatives match the empty string, the entire match fails. For
1716 example, if the pattern
1717
1718 a?b?
1719
1720 is applied to a string not beginning with "a" or "b", it matches the
1721 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
1722 match is not valid, so PCRE searches further into the string for occur-
1723 rences of "a" or "b".
1724
1725 Perl has no direct equivalent of PCRE_NOTEMPTY, but it does make a spe-
1726 cial case of a pattern match of the empty string within its split()
1727 function, and when using the /g modifier. It is possible to emulate
1728 Perl's behaviour after matching a null string by first trying the match
1729 again at the same offset with PCRE_NOTEMPTY and PCRE_ANCHORED, and then
1730 if that fails by advancing the starting offset (see below) and trying
1731 an ordinary match again. There is some code that demonstrates how to do
1732 this in the pcredemo.c sample program.
1733
1734 PCRE_NO_UTF8_CHECK
1735
1736 When PCRE_UTF8 is set at compile time, the validity of the subject as a
1737 UTF-8 string is automatically checked when pcre_exec() is subsequently
1738 called. The value of startoffset is also checked to ensure that it
1739 points to the start of a UTF-8 character. If an invalid UTF-8 sequence
1740 of bytes is found, pcre_exec() returns the error PCRE_ERROR_BADUTF8. If
1741 startoffset contains an invalid value, PCRE_ERROR_BADUTF8_OFFSET is
1742 returned.
1743
1744 If you already know that your subject is valid, and you want to skip
1745 these checks for performance reasons, you can set the
1746 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
1747 do this for the second and subsequent calls to pcre_exec() if you are
1748 making repeated calls to find all the matches in a single subject
1749 string. However, you should be sure that the value of startoffset
1750 points to the start of a UTF-8 character. When PCRE_NO_UTF8_CHECK is
1751 set, the effect of passing an invalid UTF-8 string as a subject, or a
1752 value of startoffset that does not point to the start of a UTF-8 char-
1753 acter, is undefined. Your program may crash.
1754
1755 PCRE_PARTIAL
1756
1757 This option turns on the partial matching feature. If the subject
1758 string fails to match the pattern, but at some point during the match-
1759 ing process the end of the subject was reached (that is, the subject
1760 partially matches the pattern and the failure to match occurred only
1761 because there were not enough subject characters), pcre_exec() returns
1762 PCRE_ERROR_PARTIAL instead of PCRE_ERROR_NOMATCH. When PCRE_PARTIAL is
1763 used, there are restrictions on what may appear in the pattern. These
1764 are discussed in the pcrepartial documentation.
1765
1766 The string to be matched by pcre_exec()
1767
1768 The subject string is passed to pcre_exec() as a pointer in subject, a
1769 length in length, and a starting byte offset in startoffset. In UTF-8
1770 mode, the byte offset must point to the start of a UTF-8 character.
1771 Unlike the pattern string, the subject may contain binary zero bytes.
1772 When the starting offset is zero, the search for a match starts at the
1773 beginning of the subject, and this is by far the most common case.
1774
1775 A non-zero starting offset is useful when searching for another match
1776 in the same subject by calling pcre_exec() again after a previous suc-
1777 cess. Setting startoffset differs from just passing over a shortened
1778 string and setting PCRE_NOTBOL in the case of a pattern that begins
1779 with any kind of lookbehind. For example, consider the pattern
1780
1781 \Biss\B
1782
1783 which finds occurrences of "iss" in the middle of words. (\B matches
1784 only if the current position in the subject is not a word boundary.)
1785 When applied to the string "Mississipi" the first call to pcre_exec()
1786 finds the first occurrence. If pcre_exec() is called again with just
1787 the remainder of the subject, namely "issipi", it does not match,
1788 because \B is always false at the start of the subject, which is deemed
1789 to be a word boundary. However, if pcre_exec() is passed the entire
1790 string again, but with startoffset set to 4, it finds the second occur-
1791 rence of "iss" because it is able to look behind the starting point to
1792 discover that it is preceded by a letter.
1793
1794 If a non-zero starting offset is passed when the pattern is anchored,
1795 one attempt to match at the given offset is made. This can only succeed
1796 if the pattern does not require the match to be at the start of the
1797 subject.
1798
1799 How pcre_exec() returns captured substrings
1800
1801 In general, a pattern matches a certain portion of the subject, and in
1802 addition, further substrings from the subject may be picked out by
1803 parts of the pattern. Following the usage in Jeffrey Friedl's book,
1804 this is called "capturing" in what follows, and the phrase "capturing
1805 subpattern" is used for a fragment of a pattern that picks out a sub-
1806 string. PCRE supports several other kinds of parenthesized subpattern
1807 that do not cause substrings to be captured.
1808
1809 Captured substrings are returned to the caller via a vector of integer
1810 offsets whose address is passed in ovector. The number of elements in
1811 the vector is passed in ovecsize, which must be a non-negative number.
1812 Note: this argument is NOT the size of ovector in bytes.
1813
1814 The first two-thirds of the vector is used to pass back captured sub-
1815 strings, each substring using a pair of integers. The remaining third
1816 of the vector is used as workspace by pcre_exec() while matching cap-
1817 turing subpatterns, and is not available for passing back information.
1818 The length passed in ovecsize should always be a multiple of three. If
1819 it is not, it is rounded down.
1820
1821 When a match is successful, information about captured substrings is
1822 returned in pairs of integers, starting at the beginning of ovector,
1823 and continuing up to two-thirds of its length at the most. The first
1824 element of a pair is set to the offset of the first character in a sub-
1825 string, and the second is set to the offset of the first character
1826 after the end of a substring. The first pair, ovector[0] and ovec-
1827 tor[1], identify the portion of the subject string matched by the
1828 entire pattern. The next pair is used for the first capturing subpat-
1829 tern, and so on. The value returned by pcre_exec() is one more than the
1830 highest numbered pair that has been set. For example, if two substrings
1831 have been captured, the returned value is 3. If there are no capturing
1832 subpatterns, the return value from a successful match is 1, indicating
1833 that just the first pair of offsets has been set.
1834
1835 If a capturing subpattern is matched repeatedly, it is the last portion
1836 of the string that it matched that is returned.
1837
1838 If the vector is too small to hold all the captured substring offsets,
1839 it is used as far as possible (up to two-thirds of its length), and the
1840 function returns a value of zero. In particular, if the substring off-
1841 sets are not of interest, pcre_exec() may be called with ovector passed
1842 as NULL and ovecsize as zero. However, if the pattern contains back
1843 references and the ovector is not big enough to remember the related
1844 substrings, PCRE has to get additional memory for use during matching.
1845 Thus it is usually advisable to supply an ovector.
1846
1847 The pcre_info() function can be used to find out how many capturing
1848 subpatterns there are in a compiled pattern. The smallest size for
1849 ovector that will allow for n captured substrings, in addition to the
1850 offsets of the substring matched by the whole pattern, is (n+1)*3.
1851
1852 It is possible for capturing subpattern number n+1 to match some part
1853 of the subject when subpattern n has not been used at all. For example,
1854 if the string "abc" is matched against the pattern (a|(z))(bc) the
1855 return from the function is 4, and subpatterns 1 and 3 are matched, but
1856 2 is not. When this happens, both values in the offset pairs corre-
1857 sponding to unused subpatterns are set to -1.
1858
1859 Offset values that correspond to unused subpatterns at the end of the
1860 expression are also set to -1. For example, if the string "abc" is
1861 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
1862 matched. The return from the function is 2, because the highest used
1863 capturing subpattern number is 1. However, you can refer to the offsets
1864 for the second and third capturing subpatterns if you wish (assuming
1865 the vector is large enough, of course).
1866
1867 Some convenience functions are provided for extracting the captured
1868 substrings as separate strings. These are described below.
1869
1870 Error return values from pcre_exec()
1871
1872 If pcre_exec() fails, it returns a negative number. The following are
1873 defined in the header file:
1874
1875 PCRE_ERROR_NOMATCH (-1)
1876
1877 The subject string did not match the pattern.
1878
1879 PCRE_ERROR_NULL (-2)
1880
1881 Either code or subject was passed as NULL, or ovector was NULL and
1882 ovecsize was not zero.
1883
1884 PCRE_ERROR_BADOPTION (-3)
1885
1886 An unrecognized bit was set in the options argument.
1887
1888 PCRE_ERROR_BADMAGIC (-4)
1889
1890 PCRE stores a 4-byte "magic number" at the start of the compiled code,
1891 to catch the case when it is passed a junk pointer and to detect when a
1892 pattern that was compiled in an environment of one endianness is run in
1893 an environment with the other endianness. This is the error that PCRE
1894 gives when the magic number is not present.
1895
1896 PCRE_ERROR_UNKNOWN_OPCODE (-5)
1897
1898 While running the pattern match, an unknown item was encountered in the
1899 compiled pattern. This error could be caused by a bug in PCRE or by
1900 overwriting of the compiled pattern.
1901
1902 PCRE_ERROR_NOMEMORY (-6)
1903
1904 If a pattern contains back references, but the ovector that is passed
1905 to pcre_exec() is not big enough to remember the referenced substrings,
1906 PCRE gets a block of memory at the start of matching to use for this
1907 purpose. If the call via pcre_malloc() fails, this error is given. The
1908 memory is automatically freed at the end of matching.
1909
1910 PCRE_ERROR_NOSUBSTRING (-7)
1911
1912 This error is used by the pcre_copy_substring(), pcre_get_substring(),
1913 and pcre_get_substring_list() functions (see below). It is never
1914 returned by pcre_exec().
1915
1916 PCRE_ERROR_MATCHLIMIT (-8)
1917
1918 The backtracking limit, as specified by the match_limit field in a
1919 pcre_extra structure (or defaulted) was reached. See the description
1920 above.
1921
1922 PCRE_ERROR_CALLOUT (-9)
1923
1924 This error is never generated by pcre_exec() itself. It is provided for
1925 use by callout functions that want to yield a distinctive error code.
1926 See the pcrecallout documentation for details.
1927
1928 PCRE_ERROR_BADUTF8 (-10)
1929
1930 A string that contains an invalid UTF-8 byte sequence was passed as a
1931 subject.
1932
1933 PCRE_ERROR_BADUTF8_OFFSET (-11)
1934
1935 The UTF-8 byte sequence that was passed as a subject was valid, but the
1936 value of startoffset did not point to the beginning of a UTF-8 charac-
1937 ter.
1938
1939 PCRE_ERROR_PARTIAL (-12)
1940
1941 The subject string did not match, but it did match partially. See the
1942 pcrepartial documentation for details of partial matching.
1943
1944 PCRE_ERROR_BADPARTIAL (-13)
1945
1946 The PCRE_PARTIAL option was used with a compiled pattern containing
1947 items that are not supported for partial matching. See the pcrepartial
1948 documentation for details of partial matching.
1949
1950 PCRE_ERROR_INTERNAL (-14)
1951
1952 An unexpected internal error has occurred. This error could be caused
1953 by a bug in PCRE or by overwriting of the compiled pattern.
1954
1955 PCRE_ERROR_BADCOUNT (-15)
1956
1957 This error is given if the value of the ovecsize argument is negative.
1958
1959 PCRE_ERROR_RECURSIONLIMIT (-21)
1960
1961 The internal recursion limit, as specified by the match_limit_recursion
1962 field in a pcre_extra structure (or defaulted) was reached. See the
1963 description above.
1964
1965 PCRE_ERROR_NULLWSLIMIT (-22)
1966
1967 When a group that can match an empty substring is repeated with an
1968 unbounded upper limit, the subject position at the start of the group
1969 must be remembered, so that a test for an empty string can be made when
1970 the end of the group is reached. Some workspace is required for this;
1971 if it runs out, this error is given.
1972
1973 PCRE_ERROR_BADNEWLINE (-23)
1974
1975 An invalid combination of PCRE_NEWLINE_xxx options was given.
1976
1977 Error numbers -16 to -20 are not used by pcre_exec().
1978
1979
1980 EXTRACTING CAPTURED SUBSTRINGS BY NUMBER
1981
1982 int pcre_copy_substring(const char *subject, int *ovector,
1983 int stringcount, int stringnumber, char *buffer,
1984 int buffersize);
1985
1986 int pcre_get_substring(const char *subject, int *ovector,
1987 int stringcount, int stringnumber,
1988 const char **stringptr);
1989
1990 int pcre_get_substring_list(const char *subject,
1991 int *ovector, int stringcount, const char ***listptr);
1992
1993 Captured substrings can be accessed directly by using the offsets
1994 returned by pcre_exec() in ovector. For convenience, the functions
1995 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
1996 string_list() are provided for extracting captured substrings as new,
1997 separate, zero-terminated strings. These functions identify substrings
1998 by number. The next section describes functions for extracting named
1999 substrings.
2000
2001 A substring that contains a binary zero is correctly extracted and has
2002 a further zero added on the end, but the result is not, of course, a C
2003 string. However, you can process such a string by referring to the
2004 length that is returned by pcre_copy_substring() and pcre_get_sub-
2005 string(). Unfortunately, the interface to pcre_get_substring_list() is
2006 not adequate for handling strings containing binary zeros, because the
2007 end of the final string is not independently indicated.
2008
2009 The first three arguments are the same for all three of these func-
2010 tions: subject is the subject string that has just been successfully
2011 matched, ovector is a pointer to the vector of integer offsets that was
2012 passed to pcre_exec(), and stringcount is the number of substrings that
2013 were captured by the match, including the substring that matched the
2014 entire regular expression. This is the value returned by pcre_exec() if
2015 it is greater than zero. If pcre_exec() returned zero, indicating that
2016 it ran out of space in ovector, the value passed as stringcount should
2017 be the number of elements in the vector divided by three.
2018
2019 The functions pcre_copy_substring() and pcre_get_substring() extract a
2020 single substring, whose number is given as stringnumber. A value of
2021 zero extracts the substring that matched the entire pattern, whereas
2022 higher values extract the captured substrings. For pcre_copy_sub-
2023 string(), the string is placed in buffer, whose length is given by
2024 buffersize, while for pcre_get_substring() a new block of memory is
2025 obtained via pcre_malloc, and its address is returned via stringptr.
2026 The yield of the function is the length of the string, not including
2027 the terminating zero, or one of these error codes:
2028
2029 PCRE_ERROR_NOMEMORY (-6)
2030
2031 The buffer was too small for pcre_copy_substring(), or the attempt to
2032 get memory failed for pcre_get_substring().
2033
2034 PCRE_ERROR_NOSUBSTRING (-7)
2035
2036 There is no substring whose number is stringnumber.
2037
2038 The pcre_get_substring_list() function extracts all available sub-
2039 strings and builds a list of pointers to them. All this is done in a
2040 single block of memory that is obtained via pcre_malloc. The address of
2041 the memory block is returned via listptr, which is also the start of
2042 the list of string pointers. The end of the list is marked by a NULL
2043 pointer. The yield of the function is zero if all went well, or the
2044 error code
2045
2046 PCRE_ERROR_NOMEMORY (-6)
2047
2048 if the attempt to get the memory block failed.
2049
2050 When any of these functions encounter a substring that is unset, which
2051 can happen when capturing subpattern number n+1 matches some part of
2052 the subject, but subpattern n has not been used at all, they return an
2053 empty string. This can be distinguished from a genuine zero-length sub-
2054 string by inspecting the appropriate offset in ovector, which is nega-
2055 tive for unset substrings.
2056
2057 The two convenience functions pcre_free_substring() and pcre_free_sub-
2058 string_list() can be used to free the memory returned by a previous
2059 call of pcre_get_substring() or pcre_get_substring_list(), respec-
2060 tively. They do nothing more than call the function pointed to by
2061 pcre_free, which of course could be called directly from a C program.
2062 However, PCRE is used in some situations where it is linked via a spe-
2063 cial interface to another programming language that cannot use
2064 pcre_free directly; it is for these cases that the functions are pro-
2065 vided.
2066
2067
2068 EXTRACTING CAPTURED SUBSTRINGS BY NAME
2069
2070 int pcre_get_stringnumber(const pcre *code,
2071 const char *name);
2072
2073 int pcre_copy_named_substring(const pcre *code,
2074 const char *subject, int *ovector,
2075 int stringcount, const char *stringname,
2076 char *buffer, int buffersize);
2077
2078 int pcre_get_named_substring(const pcre *code,
2079 const char *subject, int *ovector,
2080 int stringcount, const char *stringname,
2081 const char **stringptr);
2082
2083 To extract a substring by name, you first have to find associated num-
2084 ber. For example, for this pattern
2085
2086 (a+)b(?<xxx>\d+)...
2087
2088 the number of the subpattern called "xxx" is 2. If the name is known to
2089 be unique (PCRE_DUPNAMES was not set), you can find the number from the
2090 name by calling pcre_get_stringnumber(). The first argument is the com-
2091 piled pattern, and the second is the name. The yield of the function is
2092 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
2093 subpattern of that name.
2094
2095 Given the number, you can extract the substring directly, or use one of
2096 the functions described in the previous section. For convenience, there
2097 are also two functions that do the whole job.
2098
2099 Most of the arguments of pcre_copy_named_substring() and
2100 pcre_get_named_substring() are the same as those for the similarly
2101 named functions that extract by number. As these are described in the
2102 previous section, they are not re-described here. There are just two
2103 differences:
2104
2105 First, instead of a substring number, a substring name is given. Sec-
2106 ond, there is an extra argument, given at the start, which is a pointer
2107 to the compiled pattern. This is needed in order to gain access to the
2108 name-to-number translation table.
2109
2110 These functions call pcre_get_stringnumber(), and if it succeeds, they
2111 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
2112 ate.
2113
2114
2115 DUPLICATE SUBPATTERN NAMES
2116
2117 int pcre_get_stringtable_entries(const pcre *code,
2118 const char *name, char **first, char **last);
2119
2120 When a pattern is compiled with the PCRE_DUPNAMES option, names for
2121 subpatterns are not required to be unique. Normally, patterns with
2122 duplicate names are such that in any one match, only one of the named
2123 subpatterns participates. An example is shown in the pcrepattern docu-
2124 mentation. When duplicates are present, pcre_copy_named_substring() and
2125 pcre_get_named_substring() return the first substring corresponding to
2126 the given name that is set. If none are set, an empty string is
2127 returned. The pcre_get_stringnumber() function returns one of the num-
2128 bers that are associated with the name, but it is not defined which it
2129 is.
2130
2131 If you want to get full details of all captured substrings for a given
2132 name, you must use the pcre_get_stringtable_entries() function. The
2133 first argument is the compiled pattern, and the second is the name. The
2134 third and fourth are pointers to variables which are updated by the
2135 function. After it has run, they point to the first and last entries in
2136 the name-to-number table for the given name. The function itself
2137 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
2138 there are none. The format of the table is described above in the sec-
2139 tion entitled Information about a pattern. Given all the relevant
2140 entries for the name, you can extract each of their numbers, and hence
2141 the captured data, if any.
2142
2143
2144 FINDING ALL POSSIBLE MATCHES
2145
2146 The traditional matching function uses a similar algorithm to Perl,
2147 which stops when it finds the first match, starting at a given point in
2148 the subject. If you want to find all possible matches, or the longest
2149 possible match, consider using the alternative matching function (see
2150 below) instead. If you cannot use the alternative function, but still
2151 need to find all possible matches, you can kludge it up by making use
2152 of the callout facility, which is described in the pcrecallout documen-
2153 tation.
2154
2155 What you have to do is to insert a callout right at the end of the pat-
2156 tern. When your callout function is called, extract and save the cur-
2157 rent matched substring. Then return 1, which forces pcre_exec() to
2158 backtrack and try other alternatives. Ultimately, when it runs out of
2159 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
2160
2161
2162 MATCHING A PATTERN: THE ALTERNATIVE FUNCTION
2163
2164 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
2165 const char *subject, int length, int startoffset,
2166 int options, int *ovector, int ovecsize,
2167 int *workspace, int wscount);
2168
2169 The function pcre_dfa_exec() is called to match a subject string
2170 against a compiled pattern, using a matching algorithm that scans the
2171 subject string just once, and does not backtrack. This has different
2172 characteristics to the normal algorithm, and is not compatible with
2173 Perl. Some of the features of PCRE patterns are not supported. Never-
2174 theless, there are times when this kind of matching can be useful. For
2175 a discussion of the two matching algorithms, see the pcrematching docu-
2176 mentation.
2177
2178 The arguments for the pcre_dfa_exec() function are the same as for
2179 pcre_exec(), plus two extras. The ovector argument is used in a differ-
2180 ent way, and this is described below. The other common arguments are
2181 used in the same way as for pcre_exec(), so their description is not
2182 repeated here.
2183
2184 The two additional arguments provide workspace for the function. The
2185 workspace vector should contain at least 20 elements. It is used for
2186 keeping track of multiple paths through the pattern tree. More
2187 workspace will be needed for patterns and subjects where there are a
2188 lot of potential matches.
2189
2190 Here is an example of a simple call to pcre_dfa_exec():
2191
2192 int rc;
2193 int ovector[10];
2194 int wspace[20];
2195 rc = pcre_dfa_exec(
2196 re, /* result of pcre_compile() */
2197 NULL, /* we didn't study the pattern */
2198 "some string", /* the subject string */
2199 11, /* the length of the subject string */
2200 0, /* start at offset 0 in the subject */
2201 0, /* default options */
2202 ovector, /* vector of integers for substring information */
2203 10, /* number of elements (NOT size in bytes) */
2204 wspace, /* working space vector */
2205 20); /* number of elements (NOT size in bytes) */
2206
2207 Option bits for pcre_dfa_exec()
2208
2209 The unused bits of the options argument for pcre_dfa_exec() must be
2210 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
2211 LINE_xxx, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK,
2212 PCRE_PARTIAL, PCRE_DFA_SHORTEST, and PCRE_DFA_RESTART. All but the last
2213 three of these are the same as for pcre_exec(), so their description is
2214 not repeated here.
2215
2216 PCRE_PARTIAL
2217
2218 This has the same general effect as it does for pcre_exec(), but the
2219 details are slightly different. When PCRE_PARTIAL is set for
2220 pcre_dfa_exec(), the return code PCRE_ERROR_NOMATCH is converted into
2221 PCRE_ERROR_PARTIAL if the end of the subject is reached, there have
2222 been no complete matches, but there is still at least one matching pos-
2223 sibility. The portion of the string that provided the partial match is
2224 set as the first matching string.
2225
2226 PCRE_DFA_SHORTEST
2227
2228 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
2229 stop as soon as it has found one match. Because of the way the alterna-
2230 tive algorithm works, this is necessarily the shortest possible match
2231 at the first possible matching point in the subject string.
2232
2233 PCRE_DFA_RESTART
2234
2235 When pcre_dfa_exec() is called with the PCRE_PARTIAL option, and
2236 returns a partial match, it is possible to call it again, with addi-
2237 tional subject characters, and have it continue with the same match.
2238 The PCRE_DFA_RESTART option requests this action; when it is set, the
2239 workspace and wscount options must reference the same vector as before
2240 because data about the match so far is left in them after a partial
2241 match. There is more discussion of this facility in the pcrepartial
2242 documentation.
2243
2244 Successful returns from pcre_dfa_exec()
2245
2246 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
2247 string in the subject. Note, however, that all the matches from one run
2248 of the function start at the same point in the subject. The shorter
2249 matches are all initial substrings of the longer matches. For example,
2250 if the pattern
2251
2252 <.*>
2253
2254 is matched against the string
2255
2256 This is <something> <something else> <something further> no more
2257
2258 the three matched strings are
2259
2260 <something>
2261 <something> <something else>
2262 <something> <something else> <something further>
2263
2264 On success, the yield of the function is a number greater than zero,
2265 which is the number of matched substrings. The substrings themselves
2266 are returned in ovector. Each string uses two elements; the first is
2267 the offset to the start, and the second is the offset to the end. In
2268 fact, all the strings have the same start offset. (Space could have
2269 been saved by giving this only once, but it was decided to retain some
2270 compatibility with the way pcre_exec() returns data, even though the
2271 meaning of the strings is different.)
2272
2273 The strings are returned in reverse order of length; that is, the long-
2274 est matching string is given first. If there were too many matches to
2275 fit into ovector, the yield of the function is zero, and the vector is
2276 filled with the longest matches.
2277
2278 Error returns from pcre_dfa_exec()
2279
2280 The pcre_dfa_exec() function returns a negative number when it fails.
2281 Many of the errors are the same as for pcre_exec(), and these are
2282 described above. There are in addition the following errors that are
2283 specific to pcre_dfa_exec():
2284
2285 PCRE_ERROR_DFA_UITEM (-16)
2286
2287 This return is given if pcre_dfa_exec() encounters an item in the pat-
2288 tern that it does not support, for instance, the use of \C or a back
2289 reference.
2290
2291 PCRE_ERROR_DFA_UCOND (-17)
2292
2293 This return is given if pcre_dfa_exec() encounters a condition item
2294 that uses a back reference for the condition, or a test for recursion
2295 in a specific group. These are not supported.
2296
2297 PCRE_ERROR_DFA_UMLIMIT (-18)
2298
2299 This return is given if pcre_dfa_exec() is called with an extra block
2300 that contains a setting of the match_limit field. This is not supported
2301 (it is meaningless).
2302
2303 PCRE_ERROR_DFA_WSSIZE (-19)
2304
2305 This return is given if pcre_dfa_exec() runs out of space in the
2306 workspace vector.
2307
2308 PCRE_ERROR_DFA_RECURSE (-20)
2309
2310 When a recursive subpattern is processed, the matching function calls
2311 itself recursively, using private vectors for ovector and workspace.
2312 This error is given if the output vector is not large enough. This
2313 should be extremely rare, as a vector of size 1000 is used.
2314
2315
2316 SEE ALSO
2317
2318 pcrebuild(3), pcrecallout(3), pcrecpp(3)(3), pcrematching(3), pcrepar-
2319 tial(3), pcreposix(3), pcreprecompile(3), pcresample(3), pcrestack(3).
2320
2321 Last updated: 30 November 2006
2322 Copyright (c) 1997-2006 University of Cambridge.
2323 ------------------------------------------------------------------------------
2324
2325
2326 PCRECALLOUT(3) PCRECALLOUT(3)
2327
2328
2329 NAME
2330 PCRE - Perl-compatible regular expressions
2331
2332
2333 PCRE CALLOUTS
2334
2335 int (*pcre_callout)(pcre_callout_block *);
2336
2337 PCRE provides a feature called "callout", which is a means of temporar-
2338 ily passing control to the caller of PCRE in the middle of pattern
2339 matching. The caller of PCRE provides an external function by putting
2340 its entry point in the global variable pcre_callout. By default, this
2341 variable contains NULL, which disables all calling out.
2342
2343 Within a regular expression, (?C) indicates the points at which the
2344 external function is to be called. Different callout points can be
2345 identified by putting a number less than 256 after the letter C. The
2346 default value is zero. For example, this pattern has two callout
2347 points:
2348
2349 (?C1)eabc(?C2)def
2350
2351 If the PCRE_AUTO_CALLOUT option bit is set when pcre_compile() is
2352 called, PCRE automatically inserts callouts, all with number 255,
2353 before each item in the pattern. For example, if PCRE_AUTO_CALLOUT is
2354 used with the pattern
2355
2356 A(\d{2}|--)
2357
2358 it is processed as if it were
2359
2360 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
2361
2362 Notice that there is a callout before and after each parenthesis and
2363 alternation bar. Automatic callouts can be used for tracking the
2364 progress of pattern matching. The pcretest command has an option that
2365 sets automatic callouts; when it is used, the output indicates how the
2366 pattern is matched. This is useful information when you are trying to
2367 optimize the performance of a particular pattern.
2368
2369
2370 MISSING CALLOUTS
2371
2372 You should be aware that, because of optimizations in the way PCRE
2373 matches patterns, callouts sometimes do not happen. For example, if the
2374 pattern is
2375
2376 ab(?C4)cd
2377
2378 PCRE knows that any matching string must contain the letter "d". If the
2379 subject string is "abyz", the lack of "d" means that matching doesn't
2380 ever start, and the callout is never reached. However, with "abyd",
2381 though the result is still no match, the callout is obeyed.
2382
2383
2384 THE CALLOUT INTERFACE
2385
2386 During matching, when PCRE reaches a callout point, the external func-
2387 tion defined by pcre_callout is called (if it is set). This applies to
2388 both the pcre_exec() and the pcre_dfa_exec() matching functions. The
2389 only argument to the callout function is a pointer to a pcre_callout
2390 block. This structure contains the following fields:
2391
2392 int version;
2393 int callout_number;
2394 int *offset_vector;
2395 const char *subject;
2396 int subject_length;
2397 int start_match;
2398 int current_position;
2399 int capture_top;
2400 int capture_last;
2401 void *callout_data;
2402 int pattern_position;
2403 int next_item_length;
2404
2405 The version field is an integer containing the version number of the
2406 block format. The initial version was 0; the current version is 1. The
2407 version number will change again in future if additional fields are
2408 added, but the intention is never to remove any of the existing fields.
2409
2410 The callout_number field contains the number of the callout, as com-
2411 piled into the pattern (that is, the number after ?C for manual call-
2412 outs, and 255 for automatically generated callouts).
2413
2414 The offset_vector field is a pointer to the vector of offsets that was
2415 passed by the caller to pcre_exec() or pcre_dfa_exec(). When
2416 pcre_exec() is used, the contents can be inspected in order to extract
2417 substrings that have been matched so far, in the same way as for
2418 extracting substrings after a match has completed. For pcre_dfa_exec()
2419 this field is not useful.
2420
2421 The subject and subject_length fields contain copies of the values that
2422 were passed to pcre_exec().
2423
2424 The start_match field contains the offset within the subject at which
2425 the current match attempt started. If the pattern is not anchored, the
2426 callout function may be called several times from the same point in the
2427 pattern for different starting points in the subject.
2428
2429 The current_position field contains the offset within the subject of
2430 the current match pointer.
2431
2432 When the pcre_exec() function is used, the capture_top field contains
2433 one more than the number of the highest numbered captured substring so
2434 far. If no substrings have been captured, the value of capture_top is
2435 one. This is always the case when pcre_dfa_exec() is used, because it
2436 does not support captured substrings.
2437
2438 The capture_last field contains the number of the most recently cap-
2439 tured substring. If no substrings have been captured, its value is -1.
2440 This is always the case when pcre_dfa_exec() is used.
2441
2442 The callout_data field contains a value that is passed to pcre_exec()
2443 or pcre_dfa_exec() specifically so that it can be passed back in call-
2444 outs. It is passed in the pcre_callout field of the pcre_extra data
2445 structure. If no such data was passed, the value of callout_data in a
2446 pcre_callout block is NULL. There is a description of the pcre_extra
2447 structure in the pcreapi documentation.
2448
2449 The pattern_position field is present from version 1 of the pcre_call-
2450 out structure. It contains the offset to the next item to be matched in
2451 the pattern string.
2452
2453 The next_item_length field is present from version 1 of the pcre_call-
2454 out structure. It contains the length of the next item to be matched in
2455 the pattern string. When the callout immediately precedes an alterna-
2456 tion bar, a closing parenthesis, or the end of the pattern, the length
2457 is zero. When the callout precedes an opening parenthesis, the length
2458 is that of the entire subpattern.
2459
2460 The pattern_position and next_item_length fields are intended to help
2461 in distinguishing between different automatic callouts, which all have
2462 the same callout number. However, they are set for all callouts.
2463
2464
2465 RETURN VALUES
2466
2467 The external callout function returns an integer to PCRE. If the value
2468 is zero, matching proceeds as normal. If the value is greater than
2469 zero, matching fails at the current point, but the testing of other
2470 matching possibilities goes ahead, just as if a lookahead assertion had
2471 failed. If the value is less than zero, the match is abandoned, and
2472 pcre_exec() (or pcre_dfa_exec()) returns the negative value.
2473
2474 Negative values should normally be chosen from the set of
2475 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
2476 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
2477 reserved for use by callout functions; it will never be used by PCRE
2478 itself.
2479
2480 Last updated: 28 February 2005
2481 Copyright (c) 1997-2005 University of Cambridge.
2482 ------------------------------------------------------------------------------
2483
2484
2485 PCRECOMPAT(3) PCRECOMPAT(3)
2486
2487
2488 NAME
2489 PCRE - Perl-compatible regular expressions
2490
2491
2492 DIFFERENCES BETWEEN PCRE AND PERL
2493
2494 This document describes the differences in the ways that PCRE and Perl
2495 handle regular expressions. The differences described here are mainly
2496 with respect to Perl 5.8, though PCRE version 7.0 contains some fea-
2497 tures that are expected to be in the forthcoming Perl 5.10.
2498
2499 1. PCRE has only a subset of Perl's UTF-8 and Unicode support. Details
2500 of what it does have are given in the section on UTF-8 support in the
2501 main pcre page.
2502
2503 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl
2504 permits them, but they do not mean what you might think. For example,
2505 (?!a){3} does not assert that the next three characters are not "a". It
2506 just asserts that the next character is not "a" three times.
2507
2508 3. Capturing subpatterns that occur inside negative lookahead asser-
2509 tions are counted, but their entries in the offsets vector are never
2510 set. Perl sets its numerical variables from any such patterns that are
2511 matched before the assertion fails to match something (thereby succeed-
2512 ing), but only if the negative lookahead assertion contains just one
2513 branch.
2514
2515 4. Though binary zero characters are supported in the subject string,
2516 they are not allowed in a pattern string because it is passed as a nor-
2517 mal C string, terminated by zero. The escape sequence \0 can be used in
2518 the pattern to represent a binary zero.
2519
2520 5. The following Perl escape sequences are not supported: \l, \u, \L,
2521 \U, and \N. In fact these are implemented by Perl's general string-han-
2522 dling and are not part of its pattern matching engine. If any of these
2523 are encountered by PCRE, an error is generated.
2524
2525 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
2526 is built with Unicode character property support. The properties that
2527 can be tested with \p and \P are limited to the general category prop-
2528 erties such as Lu and Nd, script names such as Greek or Han, and the
2529 derived properties Any and L&.
2530
2531 7. PCRE does support the \Q...\E escape for quoting substrings. Charac-
2532 ters in between are treated as literals. This is slightly different
2533 from Perl in that $ and @ are also handled as literals inside the
2534 quotes. In Perl, they cause variable interpolation (but of course PCRE
2535 does not have variables). Note the following examples:
2536
2537 Pattern PCRE matches Perl matches
2538
2539 \Qabc$xyz\E abc$xyz abc followed by the
2540 contents of $xyz
2541 \Qabc\$xyz\E abc\$xyz abc\$xyz
2542 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2543
2544 The \Q...\E sequence is recognized both inside and outside character
2545 classes.
2546
2547 8. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
2548 constructions. However, there is support for recursive patterns. This
2549 is not available in Perl 5.8, but will be in Perl 5.10. Also, the PCRE
2550 "callout" feature allows an external function to be called during pat-
2551 tern matching. See the pcrecallout documentation for details.
2552
2553 9. Subpatterns that are called recursively or as "subroutines" are
2554 always treated as atomic groups in PCRE. This is like Python, but
2555 unlike Perl.
2556
2557 10. There are some differences that are concerned with the settings of
2558 captured strings when part of a pattern is repeated. For example,
2559 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
2560 unset, but in PCRE it is set to "b".
2561
2562 11. PCRE provides some extensions to the Perl regular expression facil-
2563 ities. Perl 5.10 will include new features that are not in earlier
2564 versions, some of which (such as named parentheses) have been in PCRE
2565 for some time. This list is with respect to Perl 5.10:
2566
2567 (a) Although lookbehind assertions must match fixed length strings,
2568 each alternative branch of a lookbehind assertion can match a different
2569 length of string. Perl requires them all to have the same length.
2570
2571 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
2572 meta-character matches only at the very end of the string.
2573
2574 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
2575 cial meaning is faulted. Otherwise, like Perl, the backslash is
2576 ignored. (Perl can be made to issue a warning.)
2577
2578 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
2579 fiers is inverted, that is, by default they are not greedy, but if fol-
2580 lowed by a question mark they are.
2581
2582 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
2583 tried only at the first matching position in the subject string.
2584
2585 (f) The PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, and PCRE_NO_AUTO_CAP-
2586 TURE options for pcre_exec() have no Perl equivalents.
2587
2588 (g) The callout facility is PCRE-specific.
2589
2590 (h) The partial matching facility is PCRE-specific.
2591
2592 (i) Patterns compiled by PCRE can be saved and re-used at a later time,
2593 even on different hosts that have the other endianness.
2594
2595 (j) The alternative matching function (pcre_dfa_exec()) matches in a
2596 different way and is not Perl-compatible.
2597
2598 Last updated: 28 November 2006
2599 Copyright (c) 1997-2006 University of Cambridge.
2600 ------------------------------------------------------------------------------
2601
2602
2603 PCREPATTERN(3) PCREPATTERN(3)
2604
2605
2606 NAME
2607 PCRE - Perl-compatible regular expressions
2608
2609
2610 PCRE REGULAR EXPRESSION DETAILS
2611
2612 The syntax and semantics of the regular expressions supported by PCRE
2613 are described below. Regular expressions are also described in the Perl
2614 documentation and in a number of books, some of which have copious
2615 examples. Jeffrey Friedl's "Mastering Regular Expressions", published
2616 by O'Reilly, covers regular expressions in great detail. This descrip-
2617 tion of PCRE's regular expressions is intended as reference material.
2618
2619 The original operation of PCRE was on strings of one-byte characters.
2620 However, there is now also support for UTF-8 character strings. To use
2621 this, you must build PCRE to include UTF-8 support, and then call
2622 pcre_compile() with the PCRE_UTF8 option. How this affects pattern
2623 matching is mentioned in several places below. There is also a summary
2624 of UTF-8 features in the section on UTF-8 support in the main pcre
2625 page.
2626
2627 The remainder of this document discusses the patterns that are sup-
2628 ported by PCRE when its main matching function, pcre_exec(), is used.
2629 From release 6.0, PCRE offers a second matching function,
2630 pcre_dfa_exec(), which matches using a different algorithm that is not
2631 Perl-compatible. The advantages and disadvantages of the alternative
2632 function, and how it differs from the normal function, are discussed in
2633 the pcrematching page.
2634
2635
2636 CHARACTERS AND METACHARACTERS
2637
2638 A regular expression is a pattern that is matched against a subject
2639 string from left to right. Most characters stand for themselves in a
2640 pattern, and match the corresponding characters in the subject. As a
2641 trivial example, the pattern
2642
2643 The quick brown fox
2644
2645 matches a portion of a subject string that is identical to itself. When
2646 caseless matching is specified (the PCRE_CASELESS option), letters are
2647 matched independently of case. In UTF-8 mode, PCRE always understands
2648 the concept of case for characters whose values are less than 128, so
2649 caseless matching is always possible. For characters with higher val-
2650 ues, the concept of case is supported if PCRE is compiled with Unicode
2651 property support, but not otherwise. If you want to use caseless
2652 matching for characters 128 and above, you must ensure that PCRE is
2653 compiled with Unicode property support as well as with UTF-8 support.
2654
2655 The power of regular expressions comes from the ability to include
2656 alternatives and repetitions in the pattern. These are encoded in the
2657 pattern by the use of metacharacters, which do not stand for themselves
2658 but instead are interpreted in some special way.
2659
2660 There are two different sets of metacharacters: those that are recog-
2661 nized anywhere in the pattern except within square brackets, and those
2662 that are recognized within square brackets. Outside square brackets,
2663 the metacharacters are as follows:
2664
2665 \ general escape character with several uses
2666 ^ assert start of string (or line, in multiline mode)
2667 $ assert end of string (or line, in multiline mode)
2668 . match any character except newline (by default)
2669 [ start character class definition
2670 | start of alternative branch
2671 ( start subpattern
2672 ) end subpattern
2673 ? extends the meaning of (
2674 also 0 or 1 quantifier
2675 also quantifier minimizer
2676 * 0 or more quantifier
2677 + 1 or more quantifier
2678 also "possessive quantifier"
2679 { start min/max quantifier
2680
2681 Part of a pattern that is in square brackets is called a "character
2682 class". In a character class the only metacharacters are:
2683
2684 \ general escape character
2685 ^ negate the class, but only if the first character
2686 - indicates character range
2687 [ POSIX character class (only if followed by POSIX
2688 syntax)
2689 ] terminates the character class
2690
2691 The following sections describe the use of each of the metacharacters.
2692
2693
2694 BACKSLASH
2695
2696 The backslash character has several uses. Firstly, if it is followed by
2697 a non-alphanumeric character, it takes away any special meaning that
2698 character may have. This use of backslash as an escape character
2699 applies both inside and outside character classes.
2700
2701 For example, if you want to match a * character, you write \* in the
2702 pattern. This escaping action applies whether or not the following
2703 character would otherwise be interpreted as a metacharacter, so it is
2704 always safe to precede a non-alphanumeric with backslash to specify
2705 that it stands for itself. In particular, if you want to match a back-
2706 slash, you write \\.
2707
2708 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
2709 the pattern (other than in a character class) and characters between a
2710 # outside a character class and the next newline are ignored. An escap-
2711 ing backslash can be used to include a whitespace or # character as
2712 part of the pattern.
2713
2714 If you want to remove the special meaning from a sequence of charac-
2715 ters, you can do so by putting them between \Q and \E. This is differ-
2716 ent from Perl in that $ and @ are handled as literals in \Q...\E
2717 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
2718 tion. Note the following examples:
2719
2720 Pattern PCRE matches Perl matches
2721
2722 \Qabc$xyz\E abc$xyz abc followed by the
2723 contents of $xyz
2724 \Qabc\$xyz\E abc\$xyz abc\$xyz
2725 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2726
2727 The \Q...\E sequence is recognized both inside and outside character
2728 classes.
2729
2730 Non-printing characters
2731
2732 A second use of backslash provides a way of encoding non-printing char-
2733 acters in patterns in a visible manner. There is no restriction on the
2734 appearance of non-printing characters, apart from the binary zero that
2735 terminates a pattern, but when a pattern is being prepared by text
2736 editing, it is usually easier to use one of the following escape
2737 sequences than the binary character it represents:
2738
2739 \a alarm, that is, the BEL character (hex 07)
2740 \cx "control-x", where x is any character
2741 \e escape (hex 1B)
2742 \f formfeed (hex 0C)
2743 \n newline (hex 0A)
2744 \r carriage return (hex 0D)
2745 \t tab (hex 09)
2746 \ddd character with octal code ddd, or backreference
2747 \xhh character with hex code hh
2748 \x{hhh..} character with hex code hhh..
2749
2750 The precise effect of \cx is as follows: if x is a lower case letter,
2751 it is converted to upper case. Then bit 6 of the character (hex 40) is
2752 inverted. Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c;
2753 becomes hex 7B.
2754
2755 After \x, from zero to two hexadecimal digits are read (letters can be
2756 in upper or lower case). Any number of hexadecimal digits may appear
2757 between \x{ and }, but the value of the character code must be less
2758 than 256 in non-UTF-8 mode, and less than 2**31 in UTF-8 mode (that is,
2759 the maximum hexadecimal value is 7FFFFFFF). If characters other than
2760 hexadecimal digits appear between \x{ and }, or if there is no termi-
2761 nating }, this form of escape is not recognized. Instead, the initial
2762 \x will be interpreted as a basic hexadecimal escape, with no following
2763 digits, giving a character whose value is zero.
2764
2765 Characters whose value is less than 256 can be defined by either of the
2766 two syntaxes for \x. There is no difference in the way they are han-
2767 dled. For example, \xdc is exactly the same as \x{dc}.
2768
2769 After \0 up to two further octal digits are read. If there are fewer
2770 than two digits, just those that are present are used. Thus the
2771 sequence \0\x\07 specifies two binary zeros followed by a BEL character
2772 (code value 7). Make sure you supply two digits after the initial zero
2773 if the pattern character that follows is itself an octal digit.
2774
2775 The handling of a backslash followed by a digit other than 0 is compli-
2776 cated. Outside a character class, PCRE reads it and any following dig-
2777 its as a decimal number. If the number is less than 10, or if there
2778 have been at least that many previous capturing left parentheses in the
2779 expression, the entire sequence is taken as a back reference. A
2780 description of how this works is given later, following the discussion
2781 of parenthesized subpatterns.
2782
2783 Inside a character class, or if the decimal number is greater than 9
2784 and there have not been that many capturing subpatterns, PCRE re-reads
2785 up to three octal digits following the backslash, and uses them to gen-
2786 erate a data character. Any subsequent digits stand for themselves. In
2787 non-UTF-8 mode, the value of a character specified in octal must be
2788 less than \400. In UTF-8 mode, values up to \777 are permitted. For
2789 example:
2790
2791 \040 is another way of writing a space
2792 \40 is the same, provided there are fewer than 40
2793 previous capturing subpatterns
2794 \7 is always a back reference
2795 \11 might be a back reference, or another way of
2796 writing a tab
2797 \011 is always a tab
2798 \0113 is a tab followed by the character "3"
2799 \113 might be a back reference, otherwise the
2800 character with octal code 113
2801 \377 might be a back reference, otherwise
2802 the byte consisting entirely of 1 bits
2803 \81 is either a back reference, or a binary zero
2804 followed by the two characters "8" and "1"
2805
2806 Note that octal values of 100 or greater must not be introduced by a
2807 leading zero, because no more than three octal digits are ever read.
2808
2809 All the sequences that define a single character value can be used both
2810 inside and outside character classes. In addition, inside a character
2811 class, the sequence \b is interpreted as the backspace character (hex
2812 08), and the sequences \R and \X are interpreted as the characters "R"
2813 and "X", respectively. Outside a character class, these sequences have
2814 different meanings (see below).
2815
2816 Absolute and relative back references
2817
2818 The sequence \g followed by a positive or negative number, optionally
2819 enclosed in braces, is an absolute or relative back reference. Back
2820 references are discussed later, following the discussion of parenthe-
2821 sized subpatterns.
2822
2823 Generic character types
2824
2825 Another use of backslash is for specifying generic character types. The
2826 following are always recognized:
2827
2828 \d any decimal digit
2829 \D any character that is not a decimal digit
2830 \s any whitespace character
2831 \S any character that is not a whitespace character
2832 \w any "word" character
2833 \W any "non-word" character
2834
2835 Each pair of escape sequences partitions the complete set of characters
2836 into two disjoint sets. Any given character matches one, and only one,
2837 of each pair.
2838
2839 These character type sequences can appear both inside and outside char-
2840 acter classes. They each match one character of the appropriate type.
2841 If the current matching point is at the end of the subject string, all
2842 of them fail, since there is no character to match.
2843
2844 For compatibility with Perl, \s does not match the VT character (code
2845 11). This makes it different from the the POSIX "space" class. The \s
2846 characters are HT (9), LF (10), FF (12), CR (13), and space (32). (If
2847 "use locale;" is included in a Perl script, \s may match the VT charac-
2848 ter. In PCRE, it never does.)
2849
2850 A "word" character is an underscore or any character less than 256 that
2851 is a letter or digit. The definition of letters and digits is con-
2852 trolled by PCRE's low-valued character tables, and may vary if locale-
2853 specific matching is taking place (see "Locale support" in the pcreapi
2854 page). For example, in the "fr_FR" (French) locale, some character
2855 codes greater than 128 are used for accented letters, and these are
2856 matched by \w.
2857
2858 In UTF-8 mode, characters with values greater than 128 never match \d,
2859 \s, or \w, and always match \D, \S, and \W. This is true even when Uni-
2860 code character property support is available. The use of locales with
2861 Unicode is discouraged.
2862
2863 Newline sequences
2864
2865 Outside a character class, the escape sequence \R matches any Unicode
2866 newline sequence. This is an extension to Perl. In non-UTF-8 mode \R is
2867 equivalent to the following:
2868
2869 (?>\r\n|\n|\x0b|\f|\r|\x85)
2870
2871 This is an example of an "atomic group", details of which are given
2872 below. This particular group matches either the two-character sequence
2873 CR followed by LF, or one of the single characters LF (linefeed,
2874 U+000A), VT (vertical tab, U+000B), FF (formfeed, U+000C), CR (carriage
2875 return, U+000D), or NEL (next line, U+0085). The two-character sequence
2876 is treated as a single unit that cannot be split.
2877
2878 In UTF-8 mode, two additional characters whose codepoints are greater
2879 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
2880 rator, U+2029). Unicode character property support is not needed for
2881 these characters to be recognized.
2882
2883 Inside a character class, \R matches the letter "R".
2884
2885 Unicode character properties
2886
2887 When PCRE is built with Unicode character property support, three addi-
2888 tional escape sequences to match character properties are available
2889 when UTF-8 mode is selected. They are:
2890
2891 \p{xx} a character with the xx property
2892 \P{xx} a character without the xx property
2893 \X an extended Unicode sequence
2894
2895 The property names represented by xx above are limited to the Unicode
2896 script names, the general category properties, and "Any", which matches
2897 any character (including newline). Other properties such as "InMusical-
2898 Symbols" are not currently supported by PCRE. Note that \P{Any} does
2899 not match any characters, so always causes a match failure.
2900
2901 Sets of Unicode characters are defined as belonging to certain scripts.
2902 A character from one of these sets can be matched using a script name.
2903 For example:
2904
2905 \p{Greek}
2906 \P{Han}
2907
2908 Those that are not part of an identified script are lumped together as
2909 "Common". The current list of scripts is:
2910
2911 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
2912 Buhid, Canadian_Aboriginal, Cherokee, Common, Coptic, Cuneiform,
2913 Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian, Glagolitic,
2914 Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
2915 gana, Inherited, Kannada, Katakana, Kharoshthi, Khmer, Lao, Latin,
2916 Limbu, Linear_B, Malayalam, Mongolian, Myanmar, New_Tai_Lue, Nko,
2917 Ogham, Old_Italic, Old_Persian, Oriya, Osmanya, Phags_Pa, Phoenician,
2918 Runic, Shavian, Sinhala, Syloti_Nagri, Syriac, Tagalog, Tagbanwa,
2919 Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Yi.
2920
2921 Each character has exactly one general category property, specified by
2922 a two-letter abbreviation. For compatibility with Perl, negation can be
2923 specified by including a circumflex between the opening brace and the
2924 property name. For example, \p{^Lu} is the same as \P{Lu}.
2925
2926 If only one letter is specified with \p or \P, it includes all the gen-
2927 eral category properties that start with that letter. In this case, in
2928 the absence of negation, the curly brackets in the escape sequence are
2929 optional; these two examples have the same effect:
2930
2931 \p{L}
2932 \pL
2933
2934 The following general category property codes are supported:
2935
2936 C Other
2937 Cc Control
2938 Cf Format
2939 Cn Unassigned
2940 Co Private use
2941 Cs Surrogate
2942
2943 L Letter
2944 Ll Lower case letter
2945 Lm Modifier letter
2946 Lo Other letter
2947 Lt Title case letter
2948 Lu Upper case letter
2949
2950 M Mark
2951 Mc Spacing mark
2952 Me Enclosing mark
2953 Mn Non-spacing mark
2954
2955 N Number
2956 Nd Decimal number
2957 Nl Letter number
2958 No Other number
2959
2960 P Punctuation
2961 Pc Connector punctuation
2962 Pd Dash punctuation
2963 Pe Close punctuation
2964 Pf Final punctuation
2965 Pi Initial punctuation
2966 Po Other punctuation
2967 Ps Open punctuation
2968
2969 S Symbol
2970 Sc Currency symbol
2971 Sk Modifier symbol
2972 Sm Mathematical symbol
2973 So Other symbol
2974
2975 Z Separator
2976 Zl Line separator
2977 Zp Paragraph separator
2978 Zs Space separator
2979
2980 The special property L& is also supported: it matches a character that
2981 has the Lu, Ll, or Lt property, in other words, a letter that is not
2982 classified as a modifier or "other".
2983
2984 The long synonyms for these properties that Perl supports (such as
2985 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
2986 any of these properties with "Is".
2987
2988 No character that is in the Unicode table has the Cn (unassigned) prop-
2989 erty. Instead, this property is assumed for any code point that is not
2990 in the Unicode table.
2991
2992 Specifying caseless matching does not affect these escape sequences.
2993 For example, \p{Lu} always matches only upper case letters.
2994
2995 The \X escape matches any number of Unicode characters that form an
2996 extended Unicode sequence. \X is equivalent to
2997
2998 (?>\PM\pM*)
2999
3000 That is, it matches a character without the "mark" property, followed
3001 by zero or more characters with the "mark" property, and treats the
3002 sequence as an atomic group (see below). Characters with the "mark"
3003 property are typically accents that affect the preceding character.
3004
3005 Matching characters by Unicode property is not fast, because PCRE has
3006 to search a structure that contains data for over fifteen thousand
3007 characters. That is why the traditional escape sequences such as \d and
3008 \w do not use Unicode properties in PCRE.
3009
3010 Simple assertions
3011
3012 The final use of backslash is for certain simple assertions. An asser-
3013 tion specifies a condition that has to be met at a particular point in
3014 a match, without consuming any characters from the subject string. The
3015 use of subpatterns for more complicated assertions is described below.
3016 The backslashed assertions are:
3017
3018 \b matches at a word boundary
3019 \B matches when not at a word boundary
3020 \A matches at the start of the subject
3021 \Z matches at the end of the subject
3022 also matches before a newline at the end of the subject
3023 \z matches only at the end of the subject
3024 \G matches at the first matching position in the subject
3025
3026 These assertions may not appear in character classes (but note that \b
3027 has a different meaning, namely the backspace character, inside a char-
3028 acter class).
3029
3030 A word boundary is a position in the subject string where the current
3031 character and the previous character do not both match \w or \W (i.e.
3032 one matches \w and the other matches \W), or the start or end of the
3033 string if the first or last character matches \w, respectively.
3034
3035 The \A, \Z, and \z assertions differ from the traditional circumflex
3036 and dollar (described in the next section) in that they only ever match
3037 at the very start and end of the subject string, whatever options are
3038 set. Thus, they are independent of multiline mode. These three asser-
3039 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
3040 affect only the behaviour of the circumflex and dollar metacharacters.
3041 However, if the startoffset argument of pcre_exec() is non-zero, indi-
3042 cating that matching is to start at a point other than the beginning of
3043 the subject, \A can never match. The difference between \Z and \z is
3044 that \Z matches before a newline at the end of the string as well as at
3045 the very end, whereas \z matches only at the end.
3046
3047 The \G assertion is true only when the current matching position is at
3048 the start point of the match, as specified by the startoffset argument
3049 of pcre_exec(). It differs from \A when the value of startoffset is
3050 non-zero. By calling pcre_exec() multiple times with appropriate argu-
3051 ments, you can mimic Perl's /g option, and it is in this kind of imple-
3052 mentation where \G can be useful.
3053
3054 Note, however, that PCRE's interpretation of \G, as the start of the
3055 current match, is subtly different from Perl's, which defines it as the
3056 end of the previous match. In Perl, these can be different when the
3057 previously matched string was empty. Because PCRE does just one match
3058 at a time, it cannot reproduce this behaviour.
3059
3060 If all the alternatives of a pattern begin with \G, the expression is
3061 anchored to the starting match position, and the "anchored" flag is set
3062 in the compiled regular expression.
3063
3064
3065 CIRCUMFLEX AND DOLLAR
3066
3067 Outside a character class, in the default matching mode, the circumflex
3068 character is an assertion that is true only if the current matching
3069 point is at the start of the subject string. If the startoffset argu-
3070 ment of pcre_exec() is non-zero, circumflex can never match if the
3071 PCRE_MULTILINE option is unset. Inside a character class, circumflex
3072 has an entirely different meaning (see below).
3073
3074 Circumflex need not be the first character of the pattern if a number
3075 of alternatives are involved, but it should be the first thing in each
3076 alternative in which it appears if the pattern is ever to match that
3077 branch. If all possible alternatives start with a circumflex, that is,
3078 if the pattern is constrained to match only at the start of the sub-
3079 ject, it is said to be an "anchored" pattern. (There are also other
3080 constructs that can cause a pattern to be anchored.)
3081
3082 A dollar character is an assertion that is true only if the current
3083 matching point is at the end of the subject string, or immediately
3084 before a newline at the end of the string (by default). Dollar need not
3085 be the last character of the pattern if a number of alternatives are
3086 involved, but it should be the last item in any branch in which it
3087 appears. Dollar has no special meaning in a character class.
3088
3089 The meaning of dollar can be changed so that it matches only at the
3090 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
3091 compile time. This does not affect the \Z assertion.
3092
3093 The meanings of the circumflex and dollar characters are changed if the
3094 PCRE_MULTILINE option is set. When this is the case, a circumflex
3095 matches immediately after internal newlines as well as at the start of
3096 the subject string. It does not match after a newline that ends the
3097 string. A dollar matches before any newlines in the string, as well as
3098 at the very end, when PCRE_MULTILINE is set. When newline is specified
3099 as the two-character sequence CRLF, isolated CR and LF characters do
3100 not indicate newlines.
3101
3102 For example, the pattern /^abc$/ matches the subject string "def\nabc"
3103 (where \n represents a newline) in multiline mode, but not otherwise.
3104 Consequently, patterns that are anchored in single line mode because
3105 all branches start with ^ are not anchored in multiline mode, and a
3106 match for circumflex is possible when the startoffset argument of
3107 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
3108 PCRE_MULTILINE is set.
3109
3110 Note that the sequences \A, \Z, and \z can be used to match the start
3111 and end of the subject in both modes, and if all branches of a pattern
3112 start with \A it is always anchored, whether or not PCRE_MULTILINE is
3113 set.
3114
3115
3116 FULL STOP (PERIOD, DOT)
3117
3118 Outside a character class, a dot in the pattern matches any one charac-
3119 ter in the subject string except (by default) a character that signi-
3120 fies the end of a line. In UTF-8 mode, the matched character may be
3121 more than one byte long.
3122
3123 When a line ending is defined as a single character, dot never matches
3124 that character; when the two-character sequence CRLF is used, dot does
3125 not match CR if it is immediately followed by LF, but otherwise it
3126 matches all characters (including isolated CRs and LFs). When any Uni-
3127 code line endings are being recognized, dot does not match CR or LF or
3128 any of the other line ending characters.
3129
3130 The behaviour of dot with regard to newlines can be changed. If the
3131 PCRE_DOTALL option is set, a dot matches any one character, without
3132 exception. If the two-character sequence CRLF is present in the subject
3133 string, it takes two dots to match it.
3134
3135 The handling of dot is entirely independent of the handling of circum-
3136 flex and dollar, the only relationship being that they both involve
3137 newlines. Dot has no special meaning in a character class.
3138
3139
3140 MATCHING A SINGLE BYTE
3141
3142 Outside a character class, the escape sequence \C matches any one byte,
3143 both in and out of UTF-8 mode. Unlike a dot, it always matches any
3144 line-ending characters. The feature is provided in Perl in order to
3145 match individual bytes in UTF-8 mode. Because it breaks up UTF-8 char-
3146 acters into individual bytes, what remains in the string may be a mal-
3147 formed UTF-8 string. For this reason, the \C escape sequence is best
3148 avoided.
3149
3150 PCRE does not allow \C to appear in lookbehind assertions (described
3151 below), because in UTF-8 mode this would make it impossible to calcu-
3152 late the length of the lookbehind.
3153
3154
3155 SQUARE BRACKETS AND CHARACTER CLASSES
3156
3157 An opening square bracket introduces a character class, terminated by a
3158 closing square bracket. A closing square bracket on its own is not spe-
3159 cial. If a closing square bracket is required as a member of the class,
3160 it should be the first data character in the class (after an initial
3161 circumflex, if present) or escaped with a backslash.
3162
3163 A character class matches a single character in the subject. In UTF-8
3164 mode, the character may occupy more than one byte. A matched character
3165 must be in the set of characters defined by the class, unless the first
3166 character in the class definition is a circumflex, in which case the
3167 subject character must not be in the set defined by the class. If a
3168 circumflex is actually required as a member of the class, ensure it is
3169 not the first character, or escape it with a backslash.
3170
3171 For example, the character class [aeiou] matches any lower case vowel,
3172 while [^aeiou] matches any character that is not a lower case vowel.
3173 Note that a circumflex is just a convenient notation for specifying the
3174 characters that are in the class by enumerating those that are not. A
3175 class that starts with a circumflex is not an assertion: it still con-
3176 sumes a character from the subject string, and therefore it fails if
3177 the current pointer is at the end of the string.
3178
3179 In UTF-8 mode, characters with values greater than 255 can be included
3180 in a class as a literal string of bytes, or by using the \x{ escaping
3181 mechanism.
3182
3183 When caseless matching is set, any letters in a class represent both
3184 their upper case and lower case versions, so for example, a caseless
3185 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
3186 match "A", whereas a caseful version would. In UTF-8 mode, PCRE always
3187 understands the concept of case for characters whose values are less
3188 than 128, so caseless matching is always possible. For characters with
3189 higher values, the concept of case is supported if PCRE is compiled
3190 with Unicode property support, but not otherwise. If you want to use
3191 caseless matching for characters 128 and above, you must ensure that
3192 PCRE is compiled with Unicode property support as well as with UTF-8
3193 support.
3194
3195 Characters that might indicate line breaks are never treated in any
3196 special way when matching character classes, whatever line-ending
3197 sequence is in use, and whatever setting of the PCRE_DOTALL and
3198 PCRE_MULTILINE options is used. A class such as [^a] always matches one
3199 of these characters.
3200
3201 The minus (hyphen) character can be used to specify a range of charac-
3202 ters in a character class. For example, [d-m] matches any letter
3203 between d and m, inclusive. If a minus character is required in a
3204 class, it must be escaped with a backslash or appear in a position
3205 where it cannot be interpreted as indicating a range, typically as the
3206 first or last character in the class.
3207
3208 It is not possible to have the literal character "]" as the end charac-
3209 ter of a range. A pattern such as [W-]46] is interpreted as a class of
3210 two characters ("W" and "-") followed by a literal string "46]", so it
3211 would match "W46]" or "-46]". However, if the "]" is escaped with a
3212 backslash it is interpreted as the end of range, so [W-\]46] is inter-
3213 preted as a class containing a range followed by two other characters.
3214 The octal or hexadecimal representation of "]" can also be used to end
3215 a range.
3216
3217 Ranges operate in the collating sequence of character values. They can
3218 also be used for characters specified numerically, for example
3219 [\000-\037]. In UTF-8 mode, ranges can include characters whose values
3220 are greater than 255, for example [\x{100}-\x{2ff}].
3221
3222 If a range that includes letters is used when caseless matching is set,
3223 it matches the letters in either case. For example, [W-c] is equivalent
3224 to [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if
3225 character tables for the "fr_FR" locale are in use, [\xc8-\xcb] matches
3226 accented E characters in both cases. In UTF-8 mode, PCRE supports the
3227 concept of case for characters with values greater than 128 only when
3228 it is compiled with Unicode property support.
3229
3230 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
3231 in a character class, and add the characters that they match to the
3232 class. For example, [\dABCDEF] matches any hexadecimal digit. A circum-
3233 flex can conveniently be used with the upper case character types to
3234 specify a more restricted set of characters than the matching lower
3235 case type. For example, the class [^\W_] matches any letter or digit,
3236 but not underscore.
3237
3238 The only metacharacters that are recognized in character classes are
3239 backslash, hyphen (only where it can be interpreted as specifying a
3240 range), circumflex (only at the start), opening square bracket (only
3241 when it can be interpreted as introducing a POSIX class name - see the
3242 next section), and the terminating closing square bracket. However,
3243 escaping other non-alphanumeric characters does no harm.
3244
3245
3246 POSIX CHARACTER CLASSES
3247
3248 Perl supports the POSIX notation for character classes. This uses names
3249 enclosed by [: and :] within the enclosing square brackets. PCRE also
3250 supports this notation. For example,
3251
3252 [01[:alpha:]%]
3253
3254 matches "0", "1", any alphabetic character, or "%". The supported class
3255 names are
3256
3257 alnum letters and digits
3258 alpha letters
3259 ascii character codes 0 - 127
3260 blank space or tab only
3261 cntrl control characters
3262 digit decimal digits (same as \d)
3263 graph printing characters, excluding space
3264 lower lower case letters
3265 print printing characters, including space
3266 punct printing characters, excluding letters and digits
3267 space white space (not quite the same as \s)
3268 upper upper case letters
3269 word "word" characters (same as \w)
3270 xdigit hexadecimal digits
3271
3272 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
3273 and space (32). Notice that this list includes the VT character (code
3274 11). This makes "space" different to \s, which does not include VT (for
3275 Perl compatibility).
3276
3277 The name "word" is a Perl extension, and "blank" is a GNU extension
3278 from Perl 5.8. Another Perl extension is negation, which is indicated
3279 by a ^ character after the colon. For example,
3280
3281 [12[:^digit:]]
3282
3283 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
3284 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
3285 these are not supported, and an error is given if they are encountered.
3286
3287 In UTF-8 mode, characters with values greater than 128 do not match any
3288 of the POSIX character classes.
3289
3290
3291 VERTICAL BAR
3292
3293 Vertical bar characters are used to separate alternative patterns. For
3294 example, the pattern
3295
3296 gilbert|sullivan
3297
3298 matches either "gilbert" or "sullivan". Any number of alternatives may
3299 appear, and an empty alternative is permitted (matching the empty
3300 string). The matching process tries each alternative in turn, from left
3301 to right, and the first one that succeeds is used. If the alternatives
3302 are within a subpattern (defined below), "succeeds" means matching the
3303 rest of the main pattern as well as the alternative in the subpattern.
3304
3305
3306 INTERNAL OPTION SETTING
3307
3308 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
3309 PCRE_EXTENDED options can be changed from within the pattern by a
3310 sequence of Perl option letters enclosed between "(?" and ")". The
3311 option letters are
3312
3313 i for PCRE_CASELESS
3314 m for PCRE_MULTILINE
3315 s for PCRE_DOTALL
3316 x for PCRE_EXTENDED
3317
3318 For example, (?im) sets caseless, multiline matching. It is also possi-
3319 ble to unset these options by preceding the letter with a hyphen, and a
3320 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
3321 LESS and PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED,
3322 is also permitted. If a letter appears both before and after the
3323 hyphen, the option is unset.
3324
3325 When an option change occurs at top level (that is, not inside subpat-
3326 tern parentheses), the change applies to the remainder of the pattern
3327 that follows. If the change is placed right at the start of a pattern,
3328 PCRE extracts it into the global options (and it will therefore show up
3329 in data extracted by the pcre_fullinfo() function).
3330
3331 An option change within a subpattern (see below for a description of
3332 subpatterns) affects only that part of the current pattern that follows
3333 it, so
3334
3335 (a(?i)b)c
3336
3337 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
3338 used). By this means, options can be made to have different settings
3339 in different parts of the pattern. Any changes made in one alternative
3340 do carry on into subsequent branches within the same subpattern. For
3341 example,
3342
3343 (a(?i)b|c)
3344
3345 matches "ab", "aB", "c", and "C", even though when matching "C" the
3346 first branch is abandoned before the option setting. This is because
3347 the effects of option settings happen at compile time. There would be
3348 some very weird behaviour otherwise.
3349
3350 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA
3351 can be changed in the same way as the Perl-compatible options by using
3352 the characters J, U and X respectively.
3353
3354
3355 SUBPATTERNS
3356
3357 Subpatterns are delimited by parentheses (round brackets), which can be
3358 nested. Turning part of a pattern into a subpattern does two things:
3359
3360 1. It localizes a set of alternatives. For example, the pattern
3361
3362 cat(aract|erpillar|)
3363
3364 matches one of the words "cat", "cataract", or "caterpillar". Without
3365 the parentheses, it would match "cataract", "erpillar" or an empty
3366 string.
3367
3368 2. It sets up the subpattern as a capturing subpattern. This means
3369 that, when the whole pattern matches, that portion of the subject
3370 string that matched the subpattern is passed back to the caller via the
3371 ovector argument of pcre_exec(). Opening parentheses are counted from
3372 left to right (starting from 1) to obtain numbers for the capturing
3373 subpatterns.
3374
3375 For example, if the string "the red king" is matched against the pat-
3376 tern
3377
3378 the ((red|white) (king|queen))
3379
3380 the captured substrings are "red king", "red", and "king", and are num-
3381 bered 1, 2, and 3, respectively.
3382
3383 The fact that plain parentheses fulfil two functions is not always
3384 helpful. There are often times when a grouping subpattern is required
3385 without a capturing requirement. If an opening parenthesis is followed
3386 by a question mark and a colon, the subpattern does not do any captur-
3387 ing, and is not counted when computing the number of any subsequent
3388 capturing subpatterns. For example, if the string "the white queen" is
3389 matched against the pattern
3390
3391 the ((?:red|white) (king|queen))
3392
3393 the captured substrings are "white queen" and "queen", and are numbered
3394 1 and 2. The maximum number of capturing subpatterns is 65535.
3395
3396 As a convenient shorthand, if any option settings are required at the
3397 start of a non-capturing subpattern, the option letters may appear
3398 between the "?" and the ":". Thus the two patterns
3399
3400 (?i:saturday|sunday)
3401 (?:(?i)saturday|sunday)
3402
3403 match exactly the same set of strings. Because alternative branches are
3404 tried from left to right, and options are not reset until the end of
3405 the subpattern is reached, an option setting in one branch does affect
3406 subsequent branches, so the above patterns match "SUNDAY" as well as
3407 "Saturday".
3408
3409
3410 NAMED SUBPATTERNS
3411
3412 Identifying capturing parentheses by number is simple, but it can be
3413 very hard to keep track of the numbers in complicated regular expres-
3414 sions. Furthermore, if an expression is modified, the numbers may
3415 change. To help with this difficulty, PCRE supports the naming of sub-
3416 patterns. This feature was not added to Perl until release 5.10. Python
3417 had the feature earlier, and PCRE introduced it at release 4.0, using
3418 the Python syntax. PCRE now supports both the Perl and the Python syn-
3419 tax.
3420
3421 In PCRE, a subpattern can be named in one of three ways: (?<name>...)
3422 or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
3423 to capturing parentheses from other parts of the pattern, such as back-
3424 references, recursion, and conditions, can be made by name as well as
3425 by number.
3426
3427 Names consist of up to 32 alphanumeric characters and underscores.
3428 Named capturing parentheses are still allocated numbers as well as
3429 names, exactly as if the names were not present. The PCRE API provides
3430 function calls for extracting the name-to-number translation table from
3431 a compiled pattern. There is also a convenience function for extracting
3432 a captured substring by name.
3433
3434 By default, a name must be unique within a pattern, but it is possible
3435 to relax this constraint by setting the PCRE_DUPNAMES option at compile
3436 time. This can be useful for patterns where only one instance of the
3437 named parentheses can match. Suppose you want to match the name of a
3438 weekday, either as a 3-letter abbreviation or as the full name, and in
3439 both cases you want to extract the abbreviation. This pattern (ignoring
3440 the line breaks) does the job:
3441
3442 (?<DN>Mon|Fri|Sun)(?:day)?|
3443 (?<DN>Tue)(?:sday)?|
3444 (?<DN>Wed)(?:nesday)?|
3445 (?<DN>Thu)(?:rsday)?|
3446 (?<DN>Sat)(?:urday)?
3447
3448 There are five capturing substrings, but only one is ever set after a
3449 match. The convenience function for extracting the data by name
3450 returns the substring for the first (and in this example, the only)
3451 subpattern of that name that matched. This saves searching to find
3452 which numbered subpattern it was. If you make a reference to a non-
3453 unique named subpattern from elsewhere in the pattern, the one that
3454 corresponds to the lowest number is used. For further details of the
3455 interfaces for handling named subpatterns, see the pcreapi documenta-
3456 tion.
3457
3458
3459 REPETITION
3460
3461 Repetition is specified by quantifiers, which can follow any of the
3462 following items:
3463
3464 a literal data character
3465 the dot metacharacter
3466 the \C escape sequence
3467 the \X escape sequence (in UTF-8 mode with Unicode properties)
3468 the \R escape sequence
3469 an escape such as \d that matches a single character
3470 a character class
3471 a back reference (see next section)
3472 a parenthesized subpattern (unless it is an assertion)
3473
3474 The general repetition quantifier specifies a minimum and maximum num-
3475 ber of permitted matches, by giving the two numbers in curly brackets
3476 (braces), separated by a comma. The numbers must be less than 65536,
3477 and the first must be less than or equal to the second. For example:
3478
3479 z{2,4}
3480
3481 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
3482 special character. If the second number is omitted, but the comma is
3483 present, there is no upper limit; if the second number and the comma
3484 are both omitted, the quantifier specifies an exact number of required
3485 matches. Thus
3486
3487 [aeiou]{3,}
3488
3489 matches at least 3 successive vowels, but may match many more, while
3490
3491 \d{8}
3492
3493 matches exactly 8 digits. An opening curly bracket that appears in a
3494 position where a quantifier is not allowed, or one that does not match
3495 the syntax of a quantifier, is taken as a literal character. For exam-
3496 ple, {,6} is not a quantifier, but a literal string of four characters.
3497
3498 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to
3499 individual bytes. Thus, for example, \x{100}{2} matches two UTF-8 char-
3500 acters, each of which is represented by a two-byte sequence. Similarly,
3501 when Unicode property support is available, \X{3} matches three Unicode
3502 extended sequences, each of which may be several bytes long (and they
3503 may be of different lengths).
3504
3505 The quantifier {0} is permitted, causing the expression to behave as if
3506 the previous item and the quantifier were not present.
3507
3508 For convenience, the three most common quantifiers have single-charac-
3509 ter abbreviations:
3510
3511 * is equivalent to {0,}
3512 + is equivalent to {1,}
3513 ? is equivalent to {0,1}
3514
3515 It is possible to construct infinite loops by following a subpattern
3516 that can match no characters with a quantifier that has no upper limit,
3517 for example:
3518
3519 (a?)*
3520
3521 Earlier versions of Perl and PCRE used to give an error at compile time
3522 for such patterns. However, because there are cases where this can be
3523 useful, such patterns are now accepted, but if any repetition of the
3524 subpattern does in fact match no characters, the loop is forcibly bro-
3525 ken.
3526
3527 By default, the quantifiers are "greedy", that is, they match as much
3528 as possible (up to the maximum number of permitted times), without
3529 causing the rest of the pattern to fail. The classic example of where
3530 this gives problems is in trying to match comments in C programs. These
3531 appear between /* and */ and within the comment, individual * and /
3532 characters may appear. An attempt to match C comments by applying the
3533 pattern
3534
3535 /\*.*\*/
3536
3537 to the string
3538
3539 /* first comment */ not comment /* second comment */
3540
3541 fails, because it matches the entire string owing to the greediness of
3542 the .* item.
3543
3544 However, if a quantifier is followed by a question mark, it ceases to
3545 be greedy, and instead matches the minimum number of times possible, so
3546 the pattern
3547
3548 /\*.*?\*/
3549
3550 does the right thing with the C comments. The meaning of the various
3551 quantifiers is not otherwise changed, just the preferred number of
3552 matches. Do not confuse this use of question mark with its use as a
3553 quantifier in its own right. Because it has two uses, it can sometimes
3554 appear doubled, as in
3555
3556 \d??\d
3557
3558 which matches one digit by preference, but can match two if that is the
3559 only way the rest of the pattern matches.
3560
3561 If the PCRE_UNGREEDY option is set (an option that is not available in
3562 Perl), the quantifiers are not greedy by default, but individual ones
3563 can be made greedy by following them with a question mark. In other
3564 words, it inverts the default behaviour.
3565
3566 When a parenthesized subpattern is quantified with a minimum repeat
3567 count that is greater than 1 or with a limited maximum, more memory is
3568 required for the compiled pattern, in proportion to the size of the
3569 minimum or maximum.
3570
3571 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
3572 alent to Perl's /s) is set, thus allowing the dot to match newlines,
3573 the pattern is implicitly anchored, because whatever follows will be
3574 tried against every character position in the subject string, so there
3575 is no point in retrying the overall match at any position after the
3576 first. PCRE normally treats such a pattern as though it were preceded
3577 by \A.
3578
3579 In cases where it is known that the subject string contains no new-
3580 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
3581 mization, or alternatively using ^ to indicate anchoring explicitly.
3582
3583 However, there is one situation where the optimization cannot be used.
3584 When .* is inside capturing parentheses that are the subject of a
3585 backreference elsewhere in the pattern, a match at the start may fail
3586 where a later one succeeds. Consider, for example:
3587
3588 (.*)abc\1
3589
3590 If the subject is "xyz123abc123" the match point is the fourth charac-
3591 ter. For this reason, such a pattern is not implicitly anchored.
3592
3593 When a capturing subpattern is repeated, the value captured is the sub-
3594 string that matched the final iteration. For example, after
3595
3596 (tweedle[dume]{3}\s*)+
3597
3598 has matched "tweedledum tweedledee" the value of the captured substring
3599 is "tweedledee". However, if there are nested capturing subpatterns,
3600 the corresponding captured values may have been set in previous itera-
3601 tions. For example, after
3602
3603 /(a|(b))+/
3604
3605 matches "aba" the value of the second captured substring is "b".
3606
3607
3608 ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS
3609
3610 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
3611 repetition, failure of what follows normally causes the repeated item
3612 to be re-evaluated to see if a different number of repeats allows the
3613 rest of the pattern to match. Sometimes it is useful to prevent this,
3614 either to change the nature of the match, or to cause it fail earlier
3615 than it otherwise might, when the author of the pattern knows there is
3616 no point in carrying on.
3617
3618 Consider, for example, the pattern \d+foo when applied to the subject
3619 line
3620
3621 123456bar
3622
3623 After matching all 6 digits and then failing to match "foo", the normal
3624 action of the matcher is to try again with only 5 digits matching the
3625 \d+ item, and then with 4, and so on, before ultimately failing.
3626 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
3627 the means for specifying that once a subpattern has matched, it is not
3628 to be re-evaluated in this way.
3629
3630 If we use atomic grouping for the previous example, the matcher gives
3631 up immediately on failing to match "foo" the first time. The notation
3632 is a kind of special parenthesis, starting with (?> as in this example:
3633
3634 (?>\d+)foo
3635
3636 This kind of parenthesis "locks up" the part of the pattern it con-
3637 tains once it has matched, and a failure further into the pattern is
3638 prevented from backtracking into it. Backtracking past it to previous
3639 items, however, works as normal.
3640
3641 An alternative description is that a subpattern of this type matches
3642 the string of characters that an identical standalone pattern would
3643 match, if anchored at the current point in the subject string.
3644
3645 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
3646 such as the above example can be thought of as a maximizing repeat that
3647 must swallow everything it can. So, while both \d+ and \d+? are pre-
3648 pared to adjust the number of digits they match in order to make the
3649 rest of the pattern match, (?>\d+) can only match an entire sequence of
3650 digits.
3651
3652 Atomic groups in general can of course contain arbitrarily complicated
3653 subpatterns, and can be nested. However, when the subpattern for an
3654 atomic group is just a single repeated item, as in the example above, a
3655 simpler notation, called a "possessive quantifier" can be used. This
3656 consists of an additional + character following a quantifier. Using
3657 this notation, the previous example can be rewritten as
3658
3659 \d++foo
3660
3661 Possessive quantifiers are always greedy; the setting of the
3662 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
3663 simpler forms of atomic group. However, there is no difference in the
3664 meaning of a possessive quantifier and the equivalent atomic group,
3665 though there may be a performance difference; possessive quantifiers
3666 should be slightly faster.
3667
3668 The possessive quantifier syntax is an extension to the Perl 5.8 syn-
3669 tax. Jeffrey Friedl originated the idea (and the name) in the first
3670 edition of his book. Mike McCloskey liked it, so implemented it when he
3671 built Sun's Java package, and PCRE copied it from there. It ultimately
3672 found its way into Perl at release 5.10.
3673
3674 PCRE has an optimization that automatically "possessifies" certain sim-
3675 ple pattern constructs. For example, the sequence A+B is treated as
3676 A++B because there is no point in backtracking into a sequence of A's
3677 when B must follow.
3678
3679 When a pattern contains an unlimited repeat inside a subpattern that
3680 can itself be repeated an unlimited number of times, the use of an
3681 atomic group is the only way to avoid some failing matches taking a
3682 very long time indeed. The pattern
3683
3684 (\D+|<\d+>)*[!?]
3685
3686 matches an unlimited number of substrings that either consist of non-
3687 digits, or digits enclosed in <>, followed by either ! or ?. When it
3688 matches, it runs quickly. However, if it is applied to
3689
3690 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
3691
3692 it takes a long time before reporting failure. This is because the
3693 string can be divided between the internal \D+ repeat and the external
3694 * repeat in a large number of ways, and all have to be tried. (The
3695 example uses [!?] rather than a single character at the end, because
3696 both PCRE and Perl have an optimization that allows for fast failure
3697 when a single character is used. They remember the last single charac-
3698 ter that is required for a match, and fail early if it is not present
3699 in the string.) If the pattern is changed so that it uses an atomic
3700 group, like this:
3701
3702 ((?>\D+)|<\d+>)*[!?]
3703
3704 sequences of non-digits cannot be broken, and failure happens quickly.
3705
3706
3707 BACK REFERENCES
3708
3709 Outside a character class, a backslash followed by a digit greater than
3710 0 (and possibly further digits) is a back reference to a capturing sub-
3711 pattern earlier (that is, to its left) in the pattern, provided there
3712 have been that many previous capturing left parentheses.
3713
3714 However, if the decimal number following the backslash is less than 10,
3715 it is always taken as a back reference, and causes an error only if
3716 there are not that many capturing left parentheses in the entire pat-
3717 tern. In other words, the parentheses that are referenced need not be
3718 to the left of the reference for numbers less than 10. A "forward back
3719 reference" of this type can make sense when a repetition is involved
3720 and the subpattern to the right has participated in an earlier itera-
3721 tion.
3722
3723 It is not possible to have a numerical "forward back reference" to a
3724 subpattern whose number is 10 or more using this syntax because a
3725 sequence such as \50 is interpreted as a character defined in octal.
3726 See the subsection entitled "Non-printing characters" above for further
3727 details of the handling of digits following a backslash. There is no
3728 such problem when named parentheses are used. A back reference to any
3729 subpattern is possible using named parentheses (see below).
3730
3731 Another way of avoiding the ambiguity inherent in the use of digits
3732 following a backslash is to use the \g escape sequence, which is a fea-
3733 ture introduced in Perl 5.10. This escape must be followed by a posi-
3734 tive or a negative number, optionally enclosed in braces. These exam-
3735 ples are all identical:
3736
3737 (ring), \1
3738 (ring), \g1
3739 (ring), \g{1}
3740
3741 A positive number specifies an absolute reference without the ambiguity
3742 that is present in the older syntax. It is also useful when literal
3743 digits follow the reference. A negative number is a relative reference.
3744 Consider this example:
3745
3746 (abc(def)ghi)\g{-1}
3747
3748 The sequence \g{-1} is a reference to the most recently started captur-
3749 ing subpattern before \g, that is, is it equivalent to \2. Similarly,
3750 \g{-2} would be equivalent to \1. The use of relative references can be
3751 helpful in long patterns, and also in patterns that are created by
3752 joining together fragments that contain references within themselves.
3753
3754 A back reference matches whatever actually matched the capturing sub-
3755 pattern in the current subject string, rather than anything matching
3756 the subpattern itself (see "Subpatterns as subroutines" below for a way
3757 of doing that). So the pattern
3758
3759 (sens|respons)e and \1ibility
3760
3761 matches "sense and sensibility" and "response and responsibility", but
3762 not "sense and responsibility". If caseful matching is in force at the
3763 time of the back reference, the case of letters is relevant. For exam-
3764 ple,
3765
3766 ((?i)rah)\s+\1
3767
3768 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
3769 original capturing subpattern is matched caselessly.
3770
3771 Back references to named subpatterns use the Perl syntax \k<name> or
3772 \k'name' or the Python syntax (?P=name). We could rewrite the above
3773 example in either of the following ways:
3774
3775 (?<p1>(?i)rah)\s+\k<p1>
3776 (?P<p1>(?i)rah)\s+(?P=p1)
3777
3778 A subpattern that is referenced by name may appear in the pattern
3779 before or after the reference.
3780
3781 There may be more than one back reference to the same subpattern. If a
3782 subpattern has not actually been used in a particular match, any back
3783 references to it always fail. For example, the pattern
3784
3785 (a|(bc))\2
3786
3787 always fails if it starts to match "a" rather than "bc". Because there
3788 may be many capturing parentheses in a pattern, all digits following
3789 the backslash are taken as part of a potential back reference number.
3790 If the pattern continues with a digit character, some delimiter must be
3791 used to terminate the back reference. If the PCRE_EXTENDED option is
3792 set, this can be whitespace. Otherwise an empty comment (see "Com-
3793 ments" below) can be used.
3794
3795 A back reference that occurs inside the parentheses to which it refers
3796 fails when the subpattern is first used, so, for example, (a\1) never
3797 matches. However, such references can be useful inside repeated sub-
3798 patterns. For example, the pattern
3799
3800 (a|b\1)+
3801
3802 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
3803 ation of the subpattern, the back reference matches the character
3804 string corresponding to the previous iteration. In order for this to
3805 work, the pattern must be such that the first iteration does not need
3806 to match the back reference. This can be done using alternation, as in
3807 the example above, or by a quantifier with a minimum of zero.
3808
3809
3810 ASSERTIONS
3811
3812 An assertion is a test on the characters following or preceding the
3813 current matching point that does not actually consume any characters.
3814 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
3815 described above.
3816
3817 More complicated assertions are coded as subpatterns. There are two
3818 kinds: those that look ahead of the current position in the subject
3819 string, and those that look behind it. An assertion subpattern is
3820 matched in the normal way, except that it does not cause the current
3821 matching position to be changed.
3822
3823 Assertion subpatterns are not capturing subpatterns, and may not be
3824 repeated, because it makes no sense to assert the same thing several
3825 times. If any kind of assertion contains capturing subpatterns within
3826 it, these are counted for the purposes of numbering the capturing sub-
3827 patterns in the whole pattern. However, substring capturing is carried
3828 out only for positive assertions, because it does not make sense for
3829 negative assertions.
3830
3831 Lookahead assertions
3832
3833 Lookahead assertions start with (?= for positive assertions and (?! for
3834 negative assertions. For example,
3835
3836 \w+(?=;)
3837
3838 matches a word followed by a semicolon, but does not include the semi-
3839 colon in the match, and
3840
3841 foo(?!bar)
3842
3843 matches any occurrence of "foo" that is not followed by "bar". Note
3844 that the apparently similar pattern
3845
3846 (?!foo)bar
3847
3848 does not find an occurrence of "bar" that is preceded by something
3849 other than "foo"; it finds any occurrence of "bar" whatsoever, because
3850 the assertion (?!foo) is always true when the next three characters are
3851 "bar". A lookbehind assertion is needed to achieve the other effect.
3852
3853 If you want to force a matching failure at some point in a pattern, the
3854 most convenient way to do it is with (?!) because an empty string
3855 always matches, so an assertion that requires there not to be an empty
3856 string must always fail.
3857
3858 Lookbehind assertions
3859
3860 Lookbehind assertions start with (?<= for positive assertions and (?<!
3861 for negative assertions. For example,
3862
3863 (?<!foo)bar
3864
3865 does find an occurrence of "bar" that is not preceded by "foo". The
3866 contents of a lookbehind assertion are restricted such that all the
3867 strings it matches must have a fixed length. However, if there are sev-
3868 eral top-level alternatives, they do not all have to have the same
3869 fixed length. Thus
3870
3871 (?<=bullock|donkey)
3872
3873 is permitted, but
3874
3875 (?<!dogs?|cats?)
3876
3877 causes an error at compile time. Branches that match different length
3878 strings are permitted only at the top level of a lookbehind assertion.
3879 This is an extension compared with Perl (at least for 5.8), which
3880 requires all branches to match the same length of string. An assertion
3881 such as
3882
3883 (?<=ab(c|de))
3884
3885 is not permitted, because its single top-level branch can match two
3886 different lengths, but it is acceptable if rewritten to use two top-
3887 level branches:
3888
3889 (?<=abc|abde)
3890
3891 The implementation of lookbehind assertions is, for each alternative,
3892 to temporarily move the current position back by the fixed length and
3893 then try to match. If there are insufficient characters before the cur-
3894 rent position, the assertion fails.
3895
3896 PCRE does not allow the \C escape (which matches a single byte in UTF-8
3897 mode) to appear in lookbehind assertions, because it makes it impossi-
3898 ble to calculate the length of the lookbehind. The \X and \R escapes,
3899 which can match different numbers of bytes, are also not permitted.
3900
3901 Possessive quantifiers can be used in conjunction with lookbehind
3902 assertions to specify efficient matching at the end of the subject
3903 string. Consider a simple pattern such as
3904
3905 abcd$
3906
3907 when applied to a long string that does not match. Because matching
3908 proceeds from left to right, PCRE will look for each "a" in the subject
3909 and then see if what follows matches the rest of the pattern. If the
3910 pattern is specified as
3911
3912 ^.*abcd$
3913
3914 the initial .* matches the entire string at first, but when this fails
3915 (because there is no following "a"), it backtracks to match all but the
3916 last character, then all but the last two characters, and so on. Once
3917 again the search for "a" covers the entire string, from right to left,
3918 so we are no better off. However, if the pattern is written as
3919
3920 ^.*+(?<=abcd)
3921
3922 there can be no backtracking for the .*+ item; it can match only the
3923 entire string. The subsequent lookbehind assertion does a single test
3924 on the last four characters. If it fails, the match fails immediately.
3925 For long strings, this approach makes a significant difference to the
3926 processing time.
3927
3928 Using multiple assertions
3929
3930 Several assertions (of any sort) may occur in succession. For example,
3931
3932 (?<=\d{3})(?<!999)foo
3933
3934 matches "foo" preceded by three digits that are not "999". Notice that
3935 each of the assertions is applied independently at the same point in
3936 the subject string. First there is a check that the previous three
3937 characters are all digits, and then there is a check that the same
3938 three characters are not "999". This pattern does not match "foo" pre-
3939 ceded by six characters, the first of which are digits and the last
3940 three of which are not "999". For example, it doesn't match "123abc-
3941 foo". A pattern to do that is
3942
3943 (?<=\d{3}...)(?<!999)foo
3944
3945 This time the first assertion looks at the preceding six characters,
3946 checking that the first three are digits, and then the second assertion
3947 checks that the preceding three characters are not "999".
3948
3949 Assertions can be nested in any combination. For example,
3950
3951 (?<=(?<!foo)bar)baz
3952
3953 matches an occurrence of "baz" that is preceded by "bar" which in turn
3954 is not preceded by "foo", while
3955
3956 (?<=\d{3}(?!999)...)foo
3957
3958 is another pattern that matches "foo" preceded by three digits and any
3959 three characters that are not "999".
3960
3961
3962 CONDITIONAL SUBPATTERNS
3963
3964 It is possible to cause the matching process to obey a subpattern con-
3965 ditionally or to choose between two alternative subpatterns, depending
3966 on the result of an assertion, or whether a previous capturing subpat-
3967 tern matched or not. The two possible forms of conditional subpattern
3968 are
3969
3970 (?(condition)yes-pattern)
3971 (?(condition)yes-pattern|no-pattern)
3972
3973 If the condition is satisfied, the yes-pattern is used; otherwise the
3974 no-pattern (if present) is used. If there are more than two alterna-
3975 tives in the subpattern, a compile-time error occurs.
3976
3977 There are four kinds of condition: references to subpatterns, refer-
3978 ences to recursion, a pseudo-condition called DEFINE, and assertions.
3979
3980 Checking for a used subpattern by number
3981
3982 If the text between the parentheses consists of a sequence of digits,
3983 the condition is true if the capturing subpattern of that number has
3984 previously matched.
3985
3986 Consider the following pattern, which contains non-significant white
3987 space to make it more readable (assume the PCRE_EXTENDED option) and to
3988 divide it into three parts for ease of discussion:
3989
3990 ( \( )? [^()]+ (?(1) \) )
3991
3992 The first part matches an optional opening parenthesis, and if that
3993 character is present, sets it as the first captured substring. The sec-
3994 ond part matches one or more characters that are not parentheses. The
3995 third part is a conditional subpattern that tests whether the first set
3996 of parentheses matched or not. If they did, that is, if subject started
3997 with an opening parenthesis, the condition is true, and so the yes-pat-
3998 tern is executed and a closing parenthesis is required. Otherwise,
3999 since no-pattern is not present, the subpattern matches nothing. In
4000 other words, this pattern matches a sequence of non-parentheses,
4001 optionally enclosed in parentheses.
4002
4003 Checking for a used subpattern by name
4004
4005 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
4006 used subpattern by name. For compatibility with earlier versions of
4007 PCRE, which had this facility before Perl, the syntax (?(name)...) is
4008 also recognized. However, there is a possible ambiguity with this syn-
4009 tax, because subpattern names may consist entirely of digits. PCRE
4010 looks first for a named subpattern; if it cannot find one and the name
4011 consists entirely of digits, PCRE looks for a subpattern of that num-
4012 ber, which must be greater than zero. Using subpattern names that con-
4013 sist entirely of digits is not recommended.
4014
4015 Rewriting the above example to use a named subpattern gives this:
4016
4017 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
4018
4019
4020 Checking for pattern recursion
4021
4022 If the condition is the string (R), and there is no subpattern with the
4023 name R, the condition is true if a recursive call to the whole pattern
4024 or any subpattern has been made. If digits or a name preceded by amper-
4025 sand follow the letter R, for example:
4026
4027 (?(R3)...) or (?(R&name)...)
4028
4029 the condition is true if the most recent recursion is into the subpat-
4030 tern whose number or name is given. This condition does not check the
4031 entire recursion stack.
4032
4033 At "top level", all these recursion test conditions are false. Recur-
4034 sive patterns are described below.
4035
4036 Defining subpatterns for use by reference only
4037
4038 If the condition is the string (DEFINE), and there is no subpattern
4039 with the name DEFINE, the condition is always false. In this case,
4040 there may be only one alternative in the subpattern. It is always
4041 skipped if control reaches this point in the pattern; the idea of
4042 DEFINE is that it can be used to define "subroutines" that can be ref-
4043 erenced from elsewhere. (The use of "subroutines" is described below.)
4044 For example, a pattern to match an IPv4 address could be written like
4045 this (ignore whitespace and line breaks):
4046
4047 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
4048 \b (?&byte) (\.(?&byte)){3} \b
4049
4050 The first part of the pattern is a DEFINE group inside which a another
4051 group named "byte" is defined. This matches an individual component of
4052 an IPv4 address (a number less than 256). When matching takes place,
4053 this part of the pattern is skipped because DEFINE acts like a false
4054 condition.
4055
4056 The rest of the pattern uses references to the named group to match the
4057 four dot-separated components of an IPv4 address, insisting on a word
4058 boundary at each end.
4059
4060 Assertion conditions
4061
4062 If the condition is not in any of the above formats, it must be an
4063 assertion. This may be a positive or negative lookahead or lookbehind
4064 assertion. Consider this pattern, again containing non-significant
4065 white space, and with the two alternatives on the second line:
4066
4067 (?(?=[^a-z]*[a-z])
4068 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
4069
4070 The condition is a positive lookahead assertion that matches an
4071 optional sequence of non-letters followed by a letter. In other words,
4072 it tests for the presence of at least one letter in the subject. If a
4073 letter is found, the subject is matched against the first alternative;
4074 otherwise it is matched against the second. This pattern matches
4075 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
4076 letters and dd are digits.
4077
4078
4079 COMMENTS
4080
4081 The sequence (?# marks the start of a comment that continues up to the
4082 next closing parenthesis. Nested parentheses are not permitted. The
4083 characters that make up a comment play no part in the pattern matching
4084 at all.
4085
4086 If the PCRE_EXTENDED option is set, an unescaped # character outside a
4087 character class introduces a comment that continues to immediately
4088 after the next newline in the pattern.
4089
4090
4091 RECURSIVE PATTERNS
4092
4093 Consider the problem of matching a string in parentheses, allowing for
4094 unlimited nested parentheses. Without the use of recursion, the best
4095 that can be done is to use a pattern that matches up to some fixed
4096 depth of nesting. It is not possible to handle an arbitrary nesting
4097 depth.
4098
4099 For some time, Perl has provided a facility that allows regular expres-
4100 sions to recurse (amongst other things). It does this by interpolating
4101 Perl code in the expression at run time, and the code can refer to the
4102 expression itself. A Perl pattern using code interpolation to solve the
4103 parentheses problem can be created like this:
4104
4105 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
4106
4107 The (?p{...}) item interpolates Perl code at run time, and in this case
4108 refers recursively to the pattern in which it appears.
4109
4110 Obviously, PCRE cannot support the interpolation of Perl code. Instead,
4111 it supports special syntax for recursion of the entire pattern, and
4112 also for individual subpattern recursion. After its introduction in
4113 PCRE and Python, this kind of recursion was introduced into Perl at
4114 release 5.10.
4115
4116 A special item that consists of (? followed by a number greater than
4117 zero and a closing parenthesis is a recursive call of the subpattern of
4118 the given number, provided that it occurs inside that subpattern. (If
4119 not, it is a "subroutine" call, which is described in the next sec-
4120 tion.) The special item (?R) or (?0) is a recursive call of the entire
4121 regular expression.
4122
4123 In PCRE (like Python, but unlike Perl), a recursive subpattern call is
4124 always treated as an atomic group. That is, once it has matched some of
4125 the subject string, it is never re-entered, even if it contains untried
4126 alternatives and there is a subsequent matching failure.
4127
4128 This PCRE pattern solves the nested parentheses problem (assume the
4129 PCRE_EXTENDED option is set so that white space is ignored):
4130
4131 \( ( (?>[^()]+) | (?R) )* \)
4132
4133 First it matches an opening parenthesis. Then it matches any number of
4134 substrings which can either be a sequence of non-parentheses, or a
4135 recursive match of the pattern itself (that is, a correctly parenthe-
4136 sized substring). Finally there is a closing parenthesis.
4137
4138 If this were part of a larger pattern, you would not want to recurse
4139 the entire pattern, so instead you could use this:
4140
4141 ( \( ( (?>[^()]+) | (?1) )* \) )
4142
4143 We have put the pattern into parentheses, and caused the recursion to
4144 refer to them instead of the whole pattern. In a larger pattern, keep-
4145 ing track of parenthesis numbers can be tricky. It may be more conve-
4146 nient to use named parentheses instead. The Perl syntax for this is
4147 (?&name); PCRE's earlier syntax (?P>name) is also supported. We could
4148 rewrite the above example as follows:
4149
4150 (?<pn> \( ( (?>[^()]+) | (?&pn) )* \) )
4151
4152 If there is more than one subpattern with the same name, the earliest
4153 one is used. This particular example pattern contains nested unlimited
4154 repeats, and so the use of atomic grouping for matching strings of non-
4155 parentheses is important when applying the pattern to strings that do
4156 not match. For example, when this pattern is applied to
4157
4158 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
4159
4160 it yields "no match" quickly. However, if atomic grouping is not used,
4161 the match runs for a very long time indeed because there are so many
4162 different ways the + and * repeats can carve up the subject, and all
4163 have to be tested before failure can be reported.
4164
4165 At the end of a match, the values set for any capturing subpatterns are
4166 those from the outermost level of the recursion at which the subpattern
4167 value is set. If you want to obtain intermediate values, a callout
4168 function can be used (see below and the pcrecallout documentation). If
4169 the pattern above is matched against
4170
4171 (ab(cd)ef)
4172
4173 the value for the capturing parentheses is "ef", which is the last
4174 value taken on at the top level. If additional parentheses are added,
4175 giving
4176
4177 \( ( ( (?>[^()]+) | (?R) )* ) \)
4178 ^ ^
4179 ^ ^
4180
4181 the string they capture is "ab(cd)ef", the contents of the top level
4182 parentheses. If there are more than 15 capturing parentheses in a pat-
4183 tern, PCRE has to obtain extra memory to store data during a recursion,
4184 which it does by using pcre_malloc, freeing it via pcre_free after-
4185 wards. If no memory can be obtained, the match fails with the
4186 PCRE_ERROR_NOMEMORY error.
4187
4188 Do not confuse the (?R) item with the condition (R), which tests for
4189 recursion. Consider this pattern, which matches text in angle brack-
4190 ets, allowing for arbitrary nesting. Only digits are allowed in nested
4191 brackets (that is, when recursing), whereas any characters are permit-
4192 ted at the outer level.
4193
4194 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
4195
4196 In this pattern, (?(R) is the start of a conditional subpattern, with
4197 two different alternatives for the recursive and non-recursive cases.
4198 The (?R) item is the actual recursive call.
4199
4200
4201 SUBPATTERNS AS SUBROUTINES
4202
4203 If the syntax for a recursive subpattern reference (either by number or
4204 by name) is used outside the parentheses to which it refers, it oper-
4205 ates like a subroutine in a programming language. The "called" subpat-
4206 tern may be defined before or after the reference. An earlier example
4207 pointed out that the pattern
4208
4209 (sens|respons)e and \1ibility
4210
4211 matches "sense and sensibility" and "response and responsibility", but
4212 not "sense and responsibility". If instead the pattern
4213
4214 (sens|respons)e and (?1)ibility
4215
4216 is used, it does match "sense and responsibility" as well as the other
4217 two strings. Another example is given in the discussion of DEFINE
4218 above.
4219
4220 Like recursive subpatterns, a "subroutine" call is always treated as an
4221 atomic group. That is, once it has matched some of the subject string,
4222 it is never re-entered, even if it contains untried alternatives and
4223 there is a subsequent matching failure.
4224
4225 When a subpattern is used as a subroutine, processing options such as
4226 case-independence are fixed when the subpattern is defined. They cannot
4227 be changed for different calls. For example, consider this pattern:
4228
4229 (abc)(?i:(?1))
4230
4231 It matches "abcabc". It does not match "abcABC" because the change of
4232 processing option does not affect the called subpattern.
4233
4234
4235 CALLOUTS
4236
4237 Perl has a feature whereby using the sequence (?{...}) causes arbitrary
4238 Perl code to be obeyed in the middle of matching a regular expression.
4239 This makes it possible, amongst other things, to extract different sub-
4240 strings that match the same pair of parentheses when there is a repeti-
4241 tion.
4242
4243 PCRE provides a similar feature, but of course it cannot obey arbitrary
4244 Perl code. The feature is called "callout". The caller of PCRE provides
4245 an external function by putting its entry point in the global variable
4246 pcre_callout. By default, this variable contains NULL, which disables
4247 all calling out.
4248
4249 Within a regular expression, (?C) indicates the points at which the
4250 external function is to be called. If you want to identify different
4251 callout points, you can put a number less than 256 after the letter C.
4252 The default value is zero. For example, this pattern has two callout
4253 points:
4254
4255 (?C1)abc(?C2)def
4256
4257 If the PCRE_AUTO_CALLOUT flag is passed to pcre_compile(), callouts are
4258 automatically installed before each item in the pattern. They are all
4259 numbered 255.
4260
4261 During matching, when PCRE reaches a callout point (and pcre_callout is
4262 set), the external function is called. It is provided with the number
4263 of the callout, the position in the pattern, and, optionally, one item
4264 of data originally supplied by the caller of pcre_exec(). The callout
4265 function may cause matching to proceed, to backtrack, or to fail alto-
4266 gether. A complete description of the interface to the callout function
4267 is given in the pcrecallout documentation.
4268
4269
4270 SEE ALSO
4271
4272 pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
4273
4274 Last updated: 06 December 2006
4275 Copyright (c) 1997-2006 University of Cambridge.
4276 ------------------------------------------------------------------------------
4277
4278
4279 PCREPARTIAL(3) PCREPARTIAL(3)
4280
4281
4282 NAME
4283 PCRE - Perl-compatible regular expressions
4284
4285
4286 PARTIAL MATCHING IN PCRE
4287
4288 In normal use of PCRE, if the subject string that is passed to
4289 pcre_exec() or pcre_dfa_exec() matches as far as it goes, but is too
4290 short to match the entire pattern, PCRE_ERROR_NOMATCH is returned.
4291 There are circumstances where it might be helpful to distinguish this
4292 case from other cases in which there is no match.
4293
4294 Consider, for example, an application where a human is required to type
4295 in data for a field with specific formatting requirements. An example
4296 might be a date in the form ddmmmyy, defined by this pattern:
4297
4298 ^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$
4299
4300 If the application sees the user's keystrokes one by one, and can check
4301 that what has been typed so far is potentially valid, it is able to
4302 raise an error as soon as a mistake is made, possibly beeping and not
4303 reflecting the character that has been typed. This immediate feedback
4304 is likely to be a better user interface than a check that is delayed
4305 until the entire string has been entered.
4306
4307 PCRE supports the concept of partial matching by means of the PCRE_PAR-
4308 TIAL option, which can be set when calling pcre_exec() or
4309 pcre_dfa_exec(). When this flag is set for pcre_exec(), the return code
4310 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if at any time
4311 during the matching process the last part of the subject string matched
4312 part of the pattern. Unfortunately, for non-anchored matching, it is
4313 not possible to obtain the position of the start of the partial match.
4314 No captured data is set when PCRE_ERROR_PARTIAL is returned.
4315
4316 When PCRE_PARTIAL is set for pcre_dfa_exec(), the return code
4317 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end of
4318 the subject is reached, there have been no complete matches, but there
4319 is still at least one matching possibility. The portion of the string
4320 that provided the partial match is set as the first matching string.
4321
4322 Using PCRE_PARTIAL disables one of PCRE's optimizations. PCRE remembers
4323 the last literal byte in a pattern, and abandons matching immediately
4324 if such a byte is not present in the subject string. This optimization
4325 cannot be used for a subject string that might match only partially.
4326
4327
4328 RESTRICTED PATTERNS FOR PCRE_PARTIAL
4329
4330 Because of the way certain internal optimizations are implemented in
4331 the pcre_exec() function, the PCRE_PARTIAL option cannot be used with
4332 all patterns. These restrictions do not apply when pcre_dfa_exec() is
4333 used. For pcre_exec(), repeated single characters such as
4334
4335 a{2,4}
4336
4337 and repeated single metasequences such as
4338
4339 \d+
4340
4341 are not permitted if the maximum number of occurrences is greater than
4342 one. Optional items such as \d? (where the maximum is one) are permit-
4343 ted. Quantifiers with any values are permitted after parentheses, so
4344 the invalid examples above can be coded thus:
4345
4346 (a){2,4}
4347 (\d)+
4348
4349 These constructions run more slowly, but for the kinds of application
4350 that are envisaged for this facility, this is not felt to be a major
4351 restriction.
4352
4353 If PCRE_PARTIAL is set for a pattern that does not conform to the
4354 restrictions, pcre_exec() returns the error code PCRE_ERROR_BADPARTIAL
4355 (-13).
4356
4357
4358 EXAMPLE OF PARTIAL MATCHING USING PCRETEST
4359
4360 If the escape sequence \P is present in a pcretest data line, the
4361 PCRE_PARTIAL flag is used for the match. Here is a run of pcretest that
4362 uses the date example quoted above:
4363
4364 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
4365 data> 25jun04\P
4366 0: 25jun04
4367 1: jun
4368 data> 25dec3\P
4369 Partial match
4370 data> 3ju\P
4371 Partial match
4372 data> 3juj\P
4373 No match
4374 data> j\P
4375 No match
4376
4377 The first data string is matched completely, so pcretest shows the
4378 matched substrings. The remaining four strings do not match the com-
4379 plete pattern, but the first two are partial matches. The same test,
4380 using pcre_dfa_exec() matching (by means of the \D escape sequence),
4381 produces the following output:
4382
4383 re> /^?(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)$/
4384 data> 25jun04\P\D
4385 0: 25jun04
4386 data> 23dec3\P\D
4387 Partial match: 23dec3
4388 data> 3ju\P\D
4389 Partial match: 3ju
4390 data> 3juj\P\D
4391 No match
4392 data> j\P\D
4393 No match
4394
4395 Notice that in this case the portion of the string that was matched is
4396 made available.
4397
4398
4399 MULTI-SEGMENT MATCHING WITH pcre_dfa_exec()
4400
4401 When a partial match has been found using pcre_dfa_exec(), it is possi-
4402 ble to continue the match by providing additional subject data and
4403 calling pcre_dfa_exec() again with the same compiled regular expres-
4404 sion, this time setting the PCRE_DFA_RESTART option. You must also pass
4405 the same working space as before, because this is where details of the
4406 previous partial match are stored. Here is an example using pcretest,
4407 using the \R escape sequence to set the PCRE_DFA_RESTART option (\P and
4408 \D are as above):
4409
4410 re> /^?(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)$/
4411 data> 23ja\P\D
4412 Partial match: 23ja
4413 data> n05\R\D
4414 0: n05
4415
4416 The first call has "23ja" as the subject, and requests partial match-
4417 ing; the second call has "n05" as the subject for the continued
4418 (restarted) match. Notice that when the match is complete, only the
4419 last part is shown; PCRE does not retain the previously partially-
4420 matched string. It is up to the calling program to do that if it needs
4421 to.
4422
4423 You can set PCRE_PARTIAL with PCRE_DFA_RESTART to continue partial
4424 matching over multiple segments. This facility can be used to pass very
4425 long subject strings to pcre_dfa_exec(). However, some care is needed
4426 for certain types of pattern.
4427
4428 1. If the pattern contains tests for the beginning or end of a line,
4429 you need to pass the PCRE_NOTBOL or PCRE_NOTEOL options, as appropri-
4430 ate, when the subject string for any call does not contain the begin-
4431 ning or end of a line.
4432
4433 2. If the pattern contains backward assertions (including \b or \B),
4434 you need to arrange for some overlap in the subject strings to allow
4435 for this. For example, you could pass the subject in chunks that are
4436 500 bytes long, but in a buffer of 700 bytes, with the starting offset
4437 set to 200 and the previous 200 bytes at the start of the buffer.
4438
4439 3. Matching a subject string that is split into multiple segments does
4440 not always produce exactly the same result as matching over one single
4441 long string. The difference arises when there are multiple matching
4442 possibilities, because a partial match result is given only when there
4443 are no completed matches in a call to fBpcre_dfa_exec(). This means
4444 that as soon as the shortest match has been found, continuation to a
4445 new subject segment is no longer possible. Consider this pcretest
4446 example:
4447
4448 re> /dog(sbody)?/
4449 data> do\P\D
4450 Partial match: do
4451 data> gsb\R\P\D
4452 0: g
4453 data> dogsbody\D
4454 0: dogsbody
4455 1: dog
4456
4457 The pattern matches the words "dog" or "dogsbody". When the subject is
4458 presented in several parts ("do" and "gsb" being the first two) the
4459 match stops when "dog" has been found, and it is not possible to con-
4460 tinue. On the other hand, if "dogsbody" is presented as a single
4461 string, both matches are found.
4462
4463 Because of this phenomenon, it does not usually make sense to end a
4464 pattern that is going to be matched in this way with a variable repeat.
4465
4466 4. Patterns that contain alternatives at the top level which do not all
4467 start with the same pattern item may not work as expected. For example,
4468 consider this pattern:
4469
4470 1234|3789
4471
4472 If the first part of the subject is "ABC123", a partial match of the
4473 first alternative is found at offset 3. There is no partial match for
4474 the second alternative, because such a match does not start at the same
4475 point in the subject string. Attempting to continue with the string
4476 "789" does not yield a match because only those alternatives that match
4477 at one point in the subject are remembered. The problem arises because
4478 the start of the second alternative matches within the first alterna-
4479 tive. There is no problem with anchored patterns or patterns such as:
4480
4481 1234|ABCD
4482
4483 where no string can be a partial match for both alternatives.
4484
4485 Last updated: 30 November 2006
4486 Copyright (c) 1997-2006 University of Cambridge.
4487 ------------------------------------------------------------------------------
4488
4489
4490 PCREPRECOMPILE(3) PCREPRECOMPILE(3)
4491
4492
4493 NAME
4494 PCRE - Perl-compatible regular expressions
4495
4496
4497 SAVING AND RE-USING PRECOMPILED PCRE PATTERNS
4498
4499 If you are running an application that uses a large number of regular
4500 expression patterns, it may be useful to store them in a precompiled
4501 form instead of having to compile them every time the application is
4502 run. If you are not using any private character tables (see the
4503 pcre_maketables() documentation), this is relatively straightforward.
4504 If you are using private tables, it is a little bit more complicated.
4505
4506 If you save compiled patterns to a file, you can copy them to a differ-
4507 ent host and run them there. This works even if the new host has the
4508 opposite endianness to the one on which the patterns were compiled.
4509 There may be a small performance penalty, but it should be insignifi-
4510 cant.
4511
4512
4513 SAVING A COMPILED PATTERN
4514 The value returned by pcre_compile() points to a single block of memory
4515 that holds the compiled pattern and associated data. You can find the
4516 length of this block in bytes by calling pcre_fullinfo() with an argu-
4517 ment of PCRE_INFO_SIZE. You can then save the data in any appropriate
4518 manner. Here is sample code that compiles a pattern and writes it to a
4519 file. It assumes that the variable fd refers to a file that is open for
4520 output:
4521
4522 int erroroffset, rc, size;
4523 char *error;
4524 pcre *re;
4525
4526 re = pcre_compile("my pattern", 0, &error, &erroroffset, NULL);
4527 if (re == NULL) { ... handle errors ... }
4528 rc = pcre_fullinfo(re, NULL, PCRE_INFO_SIZE, &size);
4529 if (rc < 0) { ... handle errors ... }
4530 rc = fwrite(re, 1, size, fd);
4531 if (rc != size) { ... handle errors ... }
4532
4533 In this example, the bytes that comprise the compiled pattern are
4534 copied exactly. Note that this is binary data that may contain any of
4535 the 256 possible byte values. On systems that make a distinction
4536 between binary and non-binary data, be sure that the file is opened for
4537 binary output.
4538
4539 If you want to write more than one pattern to a file, you will have to
4540 devise a way of separating them. For binary data, preceding each pat-
4541 tern with its length is probably the most straightforward approach.
4542 Another possibility is to write out the data in hexadecimal instead of
4543 binary, one pattern to a line.
4544
4545 Saving compiled patterns in a file is only one possible way of storing
4546 them for later use. They could equally well be saved in a database, or
4547 in the memory of some daemon process that passes them via sockets to
4548 the processes that want them.
4549
4550 If the pattern has been studied, it is also possible to save the study
4551 data in a similar way to the compiled pattern itself. When studying
4552 generates additional information, pcre_study() returns a pointer to a
4553 pcre_extra data block. Its format is defined in the section on matching
4554 a pattern in the pcreapi documentation. The study_data field points to
4555 the binary study data, and this is what you must save (not the
4556 pcre_extra block itself). The length of the study data can be obtained
4557 by calling pcre_fullinfo() with an argument of PCRE_INFO_STUDYSIZE.
4558 Remember to check that pcre_study() did return a non-NULL value before
4559 trying to save the study data.
4560
4561
4562 RE-USING A PRECOMPILED PATTERN
4563
4564 Re-using a precompiled pattern is straightforward. Having reloaded it
4565 into main memory, you pass its pointer to pcre_exec() or
4566 pcre_dfa_exec() in the usual way. This should work even on another
4567 host, and even if that host has the opposite endianness to the one
4568 where the pattern was compiled.
4569
4570 However, if you passed a pointer to custom character tables when the
4571 pattern was compiled (the tableptr argument of pcre_compile()), you
4572 must now pass a similar pointer to pcre_exec() or pcre_dfa_exec(),
4573 because the value saved with the compiled pattern will obviously be
4574 nonsense. A field in a pcre_extra() block is used to pass this data, as
4575 described in the section on matching a pattern in the pcreapi documen-
4576 tation.
4577
4578 If you did not provide custom character tables when the pattern was
4579 compiled, the pointer in the compiled pattern is NULL, which causes
4580 pcre_exec() to use PCRE's internal tables. Thus, you do not need to
4581 take any special action at run time in this case.
4582
4583 If you saved study data with the compiled pattern, you need to create
4584 your own pcre_extra data block and set the study_data field to point to
4585 the reloaded study data. You must also set the PCRE_EXTRA_STUDY_DATA
4586 bit in the flags field to indicate that study data is present. Then
4587 pass the pcre_extra block to pcre_exec() or pcre_dfa_exec() in the
4588 usual way.
4589
4590
4591 COMPATIBILITY WITH DIFFERENT PCRE RELEASES
4592
4593 The layout of the control block that is at the start of the data that
4594 makes up a compiled pattern was changed for release 5.0. If you have
4595 any saved patterns that were compiled with previous releases (not a
4596 facility that was previously advertised), you will have to recompile
4597 them for release 5.0 and above.
4598
4599 If you have any saved patterns in UTF-8 mode that use \p or \P that
4600 were compiled with any release up to and including 6.4, you will have
4601 to recompile them for release 6.5 and above.
4602
4603 All saved patterns from earlier releases must be recompiled for release
4604 7.0 or higher, because there was an internal reorganization at that
4605 release.
4606
4607 Last updated: 28 November 2006
4608 Copyright (c) 1997-2006 University of Cambridge.
4609 ------------------------------------------------------------------------------
4610
4611
4612 PCREPERFORM(3) PCREPERFORM(3)
4613
4614
4615 NAME
4616 PCRE - Perl-compatible regular expressions
4617
4618
4619 PCRE PERFORMANCE
4620
4621 Two aspects of performance are discussed below: memory usage and pro-
4622 cessing time. The way you express your pattern as a regular expression
4623 can affect both of them.
4624
4625
4626 MEMORY USAGE
4627
4628 Patterns are compiled by PCRE into a reasonably efficient byte code, so
4629 that most simple patterns do not use much memory. However, there is one
4630 case where memory usage can be unexpectedly large. When a parenthesized
4631 subpattern has a quantifier with a minimum greater than 1 and/or a lim-
4632 ited maximum, the whole subpattern is repeated in the compiled code.
4633 For example, the pattern
4634
4635 (abc|def){2,4}
4636
4637 is compiled as if it were
4638
4639 (abc|def)(abc|def)((abc|def)(abc|def)?)?
4640
4641 (Technical aside: It is done this way so that backtrack points within
4642 each of the repetitions can be independently maintained.)
4643
4644 For regular expressions whose quantifiers use only small numbers, this
4645 is not usually a problem. However, if the numbers are large, and par-
4646 ticularly if such repetitions are nested, the memory usage can become
4647 an embarrassment. For example, the very simple pattern
4648
4649 ((ab){1,1000}c){1,3}
4650
4651 uses 51K bytes when compiled. When PCRE is compiled with its default
4652 internal pointer size of two bytes, the size limit on a compiled pat-
4653 tern is 64K, and this is reached with the above pattern if the outer
4654 repetition is increased from 3 to 4. PCRE can be compiled to use larger
4655 internal pointers and thus handle larger compiled patterns, but it is
4656 better to try to rewrite your pattern to use less memory if you can.
4657
4658 One way of reducing the memory usage for such patterns is to make use
4659 of PCRE's "subroutine" facility. Re-writing the above pattern as
4660
4661 ((ab)(?2){0,999}c)(?1){0,2}
4662
4663 reduces the memory requirements to 18K, and indeed it remains under 20K
4664 even with the outer repetition increased to 100. However, this pattern
4665 is not exactly equivalent, because the "subroutine" calls are treated
4666 as atomic groups into which there can be no backtracking if there is a
4667 subsequent matching failure. Therefore, PCRE cannot do this kind of
4668 rewriting automatically. Furthermore, there is a noticeable loss of
4669 speed when executing the modified pattern. Nevertheless, if the atomic
4670 grouping is not a problem and the loss of speed is acceptable, this
4671 kind of rewriting will allow you to process patterns that PCRE cannot
4672 otherwise handle.
4673
4674
4675 PROCESSING TIME
4676
4677 Certain items in regular expression patterns are processed more effi-
4678 ciently than others. It is more efficient to use a character class like
4679 [aeiou] than a set of single-character alternatives such as
4680 (a|e|i|o|u). In general, the simplest construction that provides the
4681 required behaviour is usually the most efficient. Jeffrey Friedl's book
4682 contains a lot of useful general discussion about optimizing regular
4683 expressions for efficient performance. This document contains a few
4684 observations about PCRE.
4685
4686 Using Unicode character properties (the \p, \P, and \X escapes) is
4687 slow, because PCRE has to scan a structure that contains data for over
4688 fifteen thousand characters whenever it needs a character's property.
4689 If you can find an alternative pattern that does not use character
4690 properties, it will probably be faster.
4691
4692 When a pattern begins with .* not in parentheses, or in parentheses
4693 that are not the subject of a backreference, and the PCRE_DOTALL option
4694 is set, the pattern is implicitly anchored by PCRE, since it can match
4695 only at the start of a subject string. However, if PCRE_DOTALL is not
4696 set, PCRE cannot make this optimization, because the . metacharacter
4697 does not then match a newline, and if the subject string contains new-
4698 lines, the pattern may match from the character immediately following
4699 one of them instead of from the very start. For example, the pattern
4700
4701 .*second
4702
4703 matches the subject "first\nand second" (where \n stands for a newline
4704 character), with the match starting at the seventh character. In order
4705 to do this, PCRE has to retry the match starting after every newline in
4706 the subject.
4707
4708 If you are using such a pattern with subject strings that do not con-
4709 tain newlines, the best performance is obtained by setting PCRE_DOTALL,
4710 or starting the pattern with ^.* or ^.*? to indicate explicit anchor-
4711 ing. That saves PCRE from having to scan along the subject looking for
4712 a newline to restart at.
4713
4714 Beware of patterns that contain nested indefinite repeats. These can
4715 take a long time to run when applied to a string that does not match.
4716 Consider the pattern fragment
4717
4718 ^(a+)*
4719
4720 This can match "aaaa" in 16 different ways, and this number increases
4721 very rapidly as the string gets longer. (The * repeat can match 0, 1,
4722 2, 3, or 4 times, and for each of those cases other than 0 or 4, the +
4723 repeats can match different numbers of times.) When the remainder of
4724 the pattern is such that the entire match is going to fail, PCRE has in
4725 principle to try every possible variation, and this can take an
4726 extremely long time, even for relatively short strings.
4727
4728 An optimization catches some of the more simple cases such as
4729
4730 (a+)*b
4731
4732 where a literal character follows. Before embarking on the standard
4733 matching procedure, PCRE checks that there is a "b" later in the sub-
4734 ject string, and if there is not, it fails the match immediately. How-
4735 ever, when there is no following literal this optimization cannot be
4736 used. You can see the difference by comparing the behaviour of
4737
4738 (a+)*\d
4739
4740 with the pattern above. The former gives a failure almost instantly
4741 when applied to a whole line of "a" characters, whereas the latter
4742 takes an appreciable time with strings longer than about 20 characters.
4743
4744 In many cases, the solution to this kind of performance issue is to use
4745 an atomic group or a possessive quantifier.
4746
4747 Last updated: 20 September 2006
4748 Copyright (c) 1997-2006 University of Cambridge.
4749 ------------------------------------------------------------------------------
4750
4751
4752 PCREPOSIX(3) PCREPOSIX(3)
4753
4754
4755 NAME
4756 PCRE - Perl-compatible regular expressions.
4757
4758
4759 SYNOPSIS OF POSIX API
4760
4761 #include <pcreposix.h>
4762
4763 int regcomp(regex_t *preg, const char *pattern,
4764 int cflags);
4765
4766 int regexec(regex_t *preg, const char *string,
4767 size_t nmatch, regmatch_t pmatch[], int eflags);
4768
4769 size_t regerror(int errcode, const regex_t *preg,
4770 char *errbuf, size_t errbuf_size);
4771
4772 void regfree(regex_t *preg);
4773
4774
4775 DESCRIPTION
4776
4777 This set of functions provides a POSIX-style API to the PCRE regular
4778 expression package. See the pcreapi documentation for a description of
4779 PCRE's native API, which contains much additional functionality.
4780
4781 The functions described here are just wrapper functions that ultimately
4782 call the PCRE native API. Their prototypes are defined in the
4783 pcreposix.h header file, and on Unix systems the library itself is
4784 called pcreposix.a, so can be accessed by adding -lpcreposix to the
4785 command for linking an application that uses them. Because the POSIX
4786 functions call the native ones, it is also necessary to add -lpcre.
4787
4788 I have implemented only those option bits that can be reasonably mapped
4789 to PCRE native options. In addition, the option REG_EXTENDED is defined
4790 with the value zero. This has no effect, but since programs that are
4791 written to the POSIX interface often use it, this makes it easier to
4792 slot in PCRE as a replacement library. Other POSIX options are not even
4793 defined.
4794
4795 When PCRE is called via these functions, it is only the API that is
4796 POSIX-like in style. The syntax and semantics of the regular expres-
4797 sions themselves are still those of Perl, subject to the setting of
4798 various PCRE options, as described below. "POSIX-like in style" means
4799 that the API approximates to the POSIX definition; it is not fully
4800 POSIX-compatible, and in multi-byte encoding domains it is probably
4801 even less compatible.
4802
4803 The header for these functions is supplied as pcreposix.h to avoid any
4804 potential clash with other POSIX libraries. It can, of course, be
4805 renamed or aliased as regex.h, which is the "correct" name. It provides
4806 two structure types, regex_t for compiled internal forms, and reg-
4807 match_t for returning captured substrings. It also defines some con-
4808 stants whose names start with "REG_"; these are used for setting
4809 options and identifying error codes.
4810
4811
4812 COMPILING A PATTERN
4813
4814 The function regcomp() is called to compile a pattern into an internal
4815 form. The pattern is a C string terminated by a binary zero, and is
4816 passed in the argument pattern. The preg argument is a pointer to a
4817 regex_t structure that is used as a base for storing information about
4818 the compiled regular expression.
4819
4820 The argument cflags is either zero, or contains one or more of the bits
4821 defined by the following macros:
4822
4823 REG_DOTALL
4824
4825 The PCRE_DOTALL option is set when the regular expression is passed for
4826 compilation to the native function. Note that REG_DOTALL is not part of
4827 the POSIX standard.
4828
4829 REG_ICASE
4830
4831 The PCRE_CASELESS option is set when the regular expression is passed
4832 for compilation to the native function.
4833
4834 REG_NEWLINE
4835
4836 The PCRE_MULTILINE option is set when the regular expression is passed
4837 for compilation to the native function. Note that this does not mimic
4838 the defined POSIX behaviour for REG_NEWLINE (see the following sec-
4839 tion).
4840
4841 REG_NOSUB
4842
4843 The PCRE_NO_AUTO_CAPTURE option is set when the regular expression is
4844 passed for compilation to the native function. In addition, when a pat-
4845 tern that is compiled with this flag is passed to regexec() for match-
4846 ing, the nmatch and pmatch arguments are ignored, and no captured
4847 strings are returned.
4848
4849 REG_UTF8
4850
4851 The PCRE_UTF8 option is set when the regular expression is passed for
4852 compilation to the native function. This causes the pattern itself and
4853 all data strings used for matching it to be treated as UTF-8 strings.
4854 Note that REG_UTF8 is not part of the POSIX standard.
4855
4856 In the absence of these flags, no options are passed to the native
4857 function. This means the the regex is compiled with PCRE default
4858 semantics. In particular, the way it handles newline characters in the
4859 subject string is the Perl way, not the POSIX way. Note that setting
4860 PCRE_MULTILINE has only some of the effects specified for REG_NEWLINE.
4861 It does not affect the way newlines are matched by . (they aren't) or
4862 by a negative class such as [^a] (they are).
4863
4864 The yield of regcomp() is zero on success, and non-zero otherwise. The
4865 preg structure is filled in on success, and one member of the structure
4866 is public: re_nsub contains the number of capturing subpatterns in the
4867 regular expression. Various error codes are defined in the header file.
4868
4869
4870 MATCHING NEWLINE CHARACTERS
4871
4872 This area is not simple, because POSIX and Perl take different views of
4873 things. It is not possible to get PCRE to obey POSIX semantics, but
4874 then PCRE was never intended to be a POSIX engine. The following table
4875 lists the different possibilities for matching newline characters in
4876 PCRE:
4877
4878 Default Change with
4879
4880 . matches newline no PCRE_DOTALL
4881 newline matches [^a] yes not changeable
4882 $ matches \n at end yes PCRE_DOLLARENDONLY
4883 $ matches \n in middle no PCRE_MULTILINE
4884 ^ matches \n in middle no PCRE_MULTILINE
4885
4886 This is the equivalent table for POSIX:
4887
4888 Default Change with
4889
4890 . matches newline yes REG_NEWLINE
4891 newline matches [^a] yes REG_NEWLINE
4892 $ matches \n at end no REG_NEWLINE
4893 $ matches \n in middle no REG_NEWLINE
4894 ^ matches \n in middle no REG_NEWLINE
4895
4896 PCRE's behaviour is the same as Perl's, except that there is no equiva-
4897 lent for PCRE_DOLLAR_ENDONLY in Perl. In both PCRE and Perl, there is
4898 no way to stop newline from matching [^a].
4899
4900 The default POSIX newline handling can be obtained by setting
4901 PCRE_DOTALL and PCRE_DOLLAR_ENDONLY, but there is no way to make PCRE
4902 behave exactly as for the REG_NEWLINE action.
4903
4904
4905 MATCHING A PATTERN
4906
4907 The function regexec() is called to match a compiled pattern preg
4908 against a given string, which is terminated by a zero byte, subject to
4909 the options in eflags. These can be:
4910
4911 REG_NOTBOL
4912
4913 The PCRE_NOTBOL option is set when calling the underlying PCRE matching
4914 function.
4915
4916 REG_NOTEOL
4917
4918 The PCRE_NOTEOL option is set when calling the underlying PCRE matching
4919 function.
4920
4921 If the pattern was compiled with the REG_NOSUB flag, no data about any
4922 matched strings is returned. The nmatch and pmatch arguments of
4923 regexec() are ignored.
4924
4925 Otherwise,the portion of the string that was matched, and also any cap-
4926 tured substrings, are returned via the pmatch argument, which points to
4927 an array of nmatch structures of type regmatch_t, containing the mem-
4928 bers rm_so and rm_eo. These contain the offset to the first character
4929 of each substring and the offset to the first character after the end
4930 of each substring, respectively. The 0th element of the vector relates
4931 to the entire portion of string that was matched; subsequent elements
4932 relate to the capturing subpatterns of the regular expression. Unused
4933 entries in the array have both structure members set to -1.
4934
4935 A successful match yields a zero return; various error codes are
4936 defined in the header file, of which REG_NOMATCH is the "expected"
4937 failure code.
4938
4939
4940 ERROR MESSAGES
4941
4942 The regerror() function maps a non-zero errorcode from either regcomp()
4943 or regexec() to a printable message. If preg is not NULL, the error
4944 should have arisen from the use of that structure. A message terminated
4945 by a binary zero is placed in errbuf. The length of the message,
4946 including the zero, is limited to errbuf_size. The yield of the func-
4947 tion is the size of buffer needed to hold the whole message.
4948
4949
4950 MEMORY USAGE
4951
4952 Compiling a regular expression causes memory to be allocated and asso-
4953 ciated with the preg structure. The function regfree() frees all such
4954 memory, after which preg may no longer be used as a compiled expres-
4955 sion.
4956
4957
4958 AUTHOR
4959
4960 Philip Hazel
4961 University Computing Service,
4962 Cambridge CB2 3QH, England.
4963
4964 Last updated: 16 January 2006
4965 Copyright (c) 1997-2006 University of Cambridge.
4966 ------------------------------------------------------------------------------
4967
4968
4969 PCRECPP(3) PCRECPP(3)
4970
4971
4972 NAME
4973 PCRE - Perl-compatible regular expressions.
4974
4975
4976 SYNOPSIS OF C++ WRAPPER
4977
4978 #include <pcrecpp.h>
4979
4980
4981 DESCRIPTION
4982
4983 The C++ wrapper for PCRE was provided by Google Inc. Some additional
4984 functionality was added by Giuseppe Maxia. This brief man page was con-
4985 structed from the notes in the pcrecpp.h file, which should be con-
4986 sulted for further details.
4987
4988
4989 MATCHING INTERFACE
4990
4991 The "FullMatch" operation checks that supplied text matches a supplied
4992 pattern exactly. If pointer arguments are supplied, it copies matched
4993 sub-strings that match sub-patterns into them.
4994
4995 Example: successful match
4996 pcrecpp::RE re("h.*o");
4997 re.FullMatch("hello");
4998
4999 Example: unsuccessful match (requires full match):
5000 pcrecpp::RE re("e");
5001 !re.FullMatch("hello");
5002
5003 Example: creating a temporary RE object:
5004 pcrecpp::RE("h.*o").FullMatch("hello");
5005
5006 You can pass in a "const char*" or a "string" for "text". The examples
5007 below tend to use a const char*. You can, as in the different examples
5008 above, store the RE object explicitly in a variable or use a temporary
5009 RE object. The examples below use one mode or the other arbitrarily.
5010 Either could correctly be used for any of these examples.
5011
5012 You must supply extra pointer arguments to extract matched subpieces.
5013
5014 Example: extracts "ruby" into "s" and 1234 into "i"
5015 int i;
5016 string s;
5017 pcrecpp::RE re("(\\w+):(\\d+)");
5018 re.FullMatch("ruby:1234", &s, &i);
5019
5020 Example: does not try to extract any extra sub-patterns
5021 re.FullMatch("ruby:1234", &s);
5022
5023 Example: does not try to extract into NULL
5024 re.FullMatch("ruby:1234", NULL, &i);
5025
5026 Example: integer overflow causes failure
5027 !re.FullMatch("ruby:1234567891234", NULL, &i);
5028
5029 Example: fails because there aren't enough sub-patterns:
5030 !pcrecpp::RE("\\w+:\\d+").FullMatch("ruby:1234", &s);
5031
5032 Example: fails because string cannot be stored in integer
5033 !pcrecpp::RE("(.*)").FullMatch("ruby", &i);
5034
5035 The provided pointer arguments can be pointers to any scalar numeric
5036 type, or one of:
5037
5038 string (matched piece is copied to string)
5039 StringPiece (StringPiece is mutated to point to matched piece)
5040 T (where "bool T::ParseFrom(const char*, int)" exists)
5041 NULL (the corresponding matched sub-pattern is not copied)
5042
5043 The function returns true iff all of the following conditions are sat-
5044 isfied:
5045
5046 a. "text" matches "pattern" exactly;
5047
5048 b. The number of matched sub-patterns is >= number of supplied
5049 pointers;
5050
5051 c. The "i"th argument has a suitable type for holding the
5052 string captured as the "i"th sub-pattern. If you pass in
5053 NULL for the "i"th argument, or pass fewer arguments than
5054 number of sub-patterns, "i"th captured sub-pattern is
5055 ignored.
5056
5057 CAVEAT: An optional sub-pattern that does not exist in the matched
5058 string is assigned the empty string. Therefore, the following will
5059 return false (because the empty string is not a valid number):
5060
5061 int number;
5062 pcrecpp::RE::FullMatch("abc", "[a-z]+(\d+)?", &number);
5063
5064 The matching interface supports at most 16 arguments per call. If you
5065 need more, consider using the more general interface
5066 pcrecpp::RE::DoMatch. See pcrecpp.h for the signature for DoMatch.
5067
5068
5069 QUOTING METACHARACTERS
5070
5071 You can use the "QuoteMeta" operation to insert backslashes before all
5072 potentially meaningful characters in a string. The returned string,
5073 used as a regular expression, will exactly match the original string.
5074
5075 Example:
5076 string quoted = RE::QuoteMeta(unquoted);
5077
5078 Note that it's legal to escape a character even if it has no special
5079 meaning in a regular expression -- so this function does that. (This
5080 also makes it identical to the perl function of the same name; see
5081 "perldoc -f quotemeta".) For example, "1.5-2.0?" becomes
5082 "1\.5\-2\.0\?".
5083
5084
5085 PARTIAL MATCHES
5086
5087 You can use the "PartialMatch" operation when you want the pattern to
5088 match any substring of the text.
5089
5090 Example: simple search for a string:
5091 pcrecpp::RE("ell").PartialMatch("hello");
5092
5093 Example: find first number in a string:
5094 int number;
5095 pcrecpp::RE re("(\\d+)");
5096 re.PartialMatch("x*100 + 20", &number);
5097 assert(number == 100);
5098
5099
5100 UTF-8 AND THE MATCHING INTERFACE
5101
5102 By default, pattern and text are plain text, one byte per character.
5103 The UTF8 flag, passed to the constructor, causes both pattern and
5104 string to be treated as UTF-8 text, still a byte stream but potentially
5105 multiple bytes per character. In practice, the text is likelier to be
5106 UTF-8 than the pattern, but the match returned may depend on the UTF8
5107 flag, so always use it when matching UTF8 text. For example, "." will
5108 match one byte normally but with UTF8 set may match up to three bytes
5109 of a multi-byte character.
5110
5111 Example:
5112 pcrecpp::RE_Options options;
5113 options.set_utf8();
5114 pcrecpp::RE re(utf8_pattern, options);
5115 re.FullMatch(utf8_string);
5116
5117 Example: using the convenience function UTF8():
5118 pcrecpp::RE re(utf8_pattern, pcrecpp::UTF8());
5119 re.FullMatch(utf8_string);
5120
5121 NOTE: The UTF8 flag is ignored if pcre was not configured with the
5122 --enable-utf8 flag.
5123
5124
5125 PASSING MODIFIERS TO THE REGULAR EXPRESSION ENGINE
5126
5127 PCRE defines some modifiers to change the behavior of the regular
5128 expression engine. The C++ wrapper defines an auxiliary class,
5129 RE_Options, as a vehicle to pass such modifiers to a RE class. Cur-
5130 rently, the following modifiers are supported:
5131
5132 modifier description Perl corresponding
5133
5134 PCRE_CASELESS case insensitive match /i
5135 PCRE_MULTILINE multiple lines match /m
5136 PCRE_DOTALL dot matches newlines /s
5137 PCRE_DOLLAR_ENDONLY $ matches only at end N/A
5138 PCRE_EXTRA strict escape parsing N/A
5139 PCRE_EXTENDED ignore whitespaces /x
5140 PCRE_UTF8 handles UTF8 chars built-in
5141 PCRE_UNGREEDY reverses * and *? N/A
5142 PCRE_NO_AUTO_CAPTURE disables capturing parens N/A (*)
5143
5144 (*) Both Perl and PCRE allow non capturing parentheses by means of the
5145 "?:" modifier within the pattern itself. e.g. (?:ab|cd) does not cap-
5146 ture, while (ab|cd) does.
5147
5148 For a full account on how each modifier works, please check the PCRE
5149 API reference page.
5150
5151 For each modifier, there are two member functions whose name is made
5152 out of the modifier in lowercase, without the "PCRE_" prefix. For
5153 instance, PCRE_CASELESS is handled by
5154
5155 bool caseless()
5156
5157 which returns true if the modifier is set, and
5158
5159 RE_Options & set_caseless(bool)
5160
5161 which sets or unsets the modifier. Moreover, PCRE_EXTRA_MATCH_LIMIT can
5162 be accessed through the set_match_limit() and match_limit() member
5163 functions. Setting match_limit to a non-zero value will limit the exe-
5164 cution of pcre to keep it from doing bad things like blowing the stack
5165 or taking an eternity to return a result. A value of 5000 is good
5166 enough to stop stack blowup in a 2MB thread stack. Setting match_limit
5167 to zero disables match limiting. Alternatively, you can call
5168 match_limit_recursion() which uses PCRE_EXTRA_MATCH_LIMIT_RECURSION to
5169 limit how much PCRE recurses. match_limit() limits the number of
5170 matches PCRE does; match_limit_recursion() limits the depth of internal
5171 recursion, and therefore the amount of stack that is used.
5172
5173 Normally, to pass one or more modifiers to a RE class, you declare a
5174 RE_Options object, set the appropriate options, and pass this object to
5175 a RE constructor. Example:
5176
5177 RE_options opt;
5178 opt.set_caseless(true);
5179 if (RE("HELLO", opt).PartialMatch("hello world")) ...
5180
5181 RE_options has two constructors. The default constructor takes no argu-
5182 ments and creates a set of flags that are off by default. The optional
5183 parameter option_flags is to facilitate transfer of legacy code from C
5184 programs. This lets you do
5185
5186 RE(pattern,
5187 RE_Options(PCRE_CASELESS|PCRE_MULTILINE)).PartialMatch(str);
5188
5189 However, new code is better off doing
5190
5191 RE(pattern,
5192 RE_Options().set_caseless(true).set_multiline(true))
5193 .PartialMatch(str);
5194
5195 If you are going to pass one of the most used modifiers, there are some
5196 convenience functions that return a RE_Options class with the appropri-
5197 ate modifier already set: CASELESS(), UTF8(), MULTILINE(), DOTALL(),
5198 and EXTENDED().
5199
5200 If you need to set several options at once, and you don't want to go
5201 through the pains of declaring a RE_Options object and setting several
5202 options, there is a parallel method that give you such ability on the
5203 fly. You can concatenate several set_xxxxx() member functions, since
5204 each of them returns a reference to its class object. For example, to
5205 pass PCRE_CASELESS, PCRE_EXTENDED, and PCRE_MULTILINE to a RE with one
5206 statement, you may write:
5207
5208 RE(" ^ xyz \\s+ .* blah$",
5209 RE_Options()
5210 .set_caseless(true)
5211 .set_extended(true)
5212 .set_multiline(true)).PartialMatch(sometext);
5213
5214
5215 SCANNING TEXT INCREMENTALLY
5216
5217 The "Consume" operation may be useful if you want to repeatedly match
5218 regular expressions at the front of a string and skip over them as they
5219 match. This requires use of the "StringPiece" type, which represents a
5220 sub-range of a real string. Like RE, StringPiece is defined in the
5221 pcrecpp namespace.
5222
5223 Example: read lines of the form "var = value" from a string.
5224 string contents = ...; // Fill string somehow
5225 pcrecpp::StringPiece input(contents); // Wrap in a StringPiece
5226
5227 string var;
5228 int value;
5229 pcrecpp::RE re("(\\w+) = (\\d+)\n");
5230 while (re.Consume(&input, &var, &value)) {
5231 ...;
5232 }
5233
5234 Each successful call to "Consume" will set "var/value", and also
5235 advance "input" so it points past the matched text.
5236
5237 The "FindAndConsume" operation is similar to "Consume" but does not
5238 anchor your match at the beginning of the string. For example, you
5239 could extract all words from a string by repeatedly calling
5240
5241 pcrecpp::RE("(\\w+)").FindAndConsume(&input, &word)
5242
5243
5244 PARSING HEX/OCTAL/C-RADIX NUMBERS
5245
5246 By default, if you pass a pointer to a numeric value, the corresponding
5247 text is interpreted as a base-10 number. You can instead wrap the
5248 pointer with a call to one of the operators Hex(), Octal(), or CRadix()
5249 to interpret the text in another base. The CRadix operator interprets
5250 C-style "0" (base-8) and "0x" (base-16) prefixes, but defaults to
5251 base-10.
5252
5253 Example:
5254 int a, b, c, d;
5255 pcrecpp::RE re("(.*) (.*) (.*) (.*)");
5256 re.FullMatch("100 40 0100 0x40",
5257 pcrecpp::Octal(&a), pcrecpp::Hex(&b),
5258 pcrecpp::CRadix(&c), pcrecpp::CRadix(&d));
5259
5260 will leave 64 in a, b, c, and d.
5261
5262
5263 REPLACING PARTS OF STRINGS
5264
5265 You can replace the first match of "pattern" in "str" with "rewrite".
5266 Within "rewrite", backslash-escaped digits (\1 to \9) can be used to
5267 insert text matching corresponding parenthesized group from the pat-
5268 tern. \0 in "rewrite" refers to the entire matching text. For example:
5269
5270 string s = "yabba dabba doo";
5271 pcrecpp::RE("b+").Replace("d", &s);
5272
5273 will leave "s" containing "yada dabba doo". The result is true if the
5274 pattern matches and a replacement occurs, false otherwise.
5275
5276 GlobalReplace is like Replace except that it replaces all occurrences
5277 of the pattern in the string with the rewrite. Replacements are not
5278 subject to re-matching. For example:
5279
5280 string s = "yabba dabba doo";
5281 pcrecpp::RE("b+").GlobalReplace("d", &s);
5282
5283 will leave "s" containing "yada dada doo". It returns the number of
5284 replacements made.
5285
5286 Extract is like Replace, except that if the pattern matches, "rewrite"
5287 is copied into "out" (an additional argument) with substitutions. The
5288 non-matching portions of "text" are ignored. Returns true iff a match
5289 occurred and the extraction happened successfully; if no match occurs,
5290 the string is left unaffected.
5291
5292
5293 AUTHOR
5294
5295 The C++ wrapper was contributed by Google Inc.
5296 Copyright (c) 2006 Google Inc.
5297 ------------------------------------------------------------------------------
5298
5299
5300 PCRESAMPLE(3) PCRESAMPLE(3)
5301
5302
5303 NAME
5304 PCRE - Perl-compatible regular expressions
5305
5306
5307 PCRE SAMPLE PROGRAM
5308
5309 A simple, complete demonstration program, to get you started with using
5310 PCRE, is supplied in the file pcredemo.c in the PCRE distribution.
5311
5312 The program compiles the regular expression that is its first argument,
5313 and matches it against the subject string in its second argument. No
5314 PCRE options are set, and default character tables are used. If match-
5315 ing succeeds, the program outputs the portion of the subject that
5316 matched, together with the contents of any captured substrings.
5317
5318 If the -g option is given on the command line, the program then goes on
5319 to check for further matches of the same regular expression in the same
5320 subject string. The logic is a little bit tricky because of the possi-
5321 bility of matching an empty string. Comments in the code explain what
5322 is going on.
5323
5324 If PCRE is installed in the standard include and library directories
5325 for your system, you should be able to compile the demonstration pro-
5326 gram using this command:
5327
5328 gcc -o pcredemo pcredemo.c -lpcre
5329
5330 If PCRE is installed elsewhere, you may need to add additional options
5331 to the command line. For example, on a Unix-like system that has PCRE
5332 installed in /usr/local, you can compile the demonstration program
5333 using a command like this:
5334
5335 gcc -o pcredemo -I/usr/local/include pcredemo.c \
5336 -L/usr/local/lib -lpcre
5337
5338 Once you have compiled the demonstration program, you can run simple
5339 tests like this:
5340
5341 ./pcredemo 'cat|dog' 'the cat sat on the mat'
5342 ./pcredemo -g 'cat|dog' 'the dog sat on the cat'
5343
5344 Note that there is a much more comprehensive test program, called
5345 pcretest, which supports many more facilities for testing regular
5346 expressions and the PCRE library. The pcredemo program is provided as a
5347 simple coding example.
5348
5349 On some operating systems (e.g. Solaris), when PCRE is not installed in
5350 the standard library directory, you may get an error like this when you
5351 try to run pcredemo:
5352
5353 ld.so.1: a.out: fatal: libpcre.so.0: open failed: No such file or
5354 directory
5355
5356 This is caused by the way shared library support works on those sys-
5357 tems. You need to add
5358
5359 -R/usr/local/lib
5360
5361 (for example) to the compile command to get round this problem.
5362
5363 Last updated: 09 September 2004
5364 Copyright (c) 1997-2004 University of Cambridge.
5365 ------------------------------------------------------------------------------
5366 PCRESTACK(3) PCRESTACK(3)
5367
5368
5369 NAME
5370 PCRE - Perl-compatible regular expressions
5371
5372
5373 PCRE DISCUSSION OF STACK USAGE
5374
5375 When you call pcre_exec(), it makes use of an internal function called
5376 match(). This calls itself recursively at branch points in the pattern,
5377 in order to remember the state of the match so that it can back up and
5378 try a different alternative if the first one fails. As matching pro-
5379 ceeds deeper and deeper into the tree of possibilities, the recursion
5380 depth increases.
5381
5382 Not all calls of match() increase the recursion depth; for an item such
5383 as a* it may be called several times at the same level, after matching
5384 different numbers of a's. Furthermore, in a number of cases where the
5385 result of the recursive call would immediately be passed back as the
5386 result of the current call (a "tail recursion"), the function is just
5387 restarted instead.
5388
5389 The pcre_dfa_exec() function operates in an entirely different way, and
5390 hardly uses recursion at all. The limit on its complexity is the amount
5391 of workspace it is given. The comments that follow do NOT apply to
5392 pcre_dfa_exec(); they are relevant only for pcre_exec().
5393
5394 You can set limits on the number of times that match() is called, both
5395 in total and recursively. If the limit is exceeded, an error occurs.
5396 For details, see the section on extra data for pcre_exec() in the
5397 pcreapi documentation.
5398
5399 Each time that match() is actually called recursively, it uses memory
5400 from the process stack. For certain kinds of pattern and data, very
5401 large amounts of stack may be needed, despite the recognition of "tail
5402 recursion". You can often reduce the amount of recursion, and there-
5403 fore the amount of stack used, by modifying the pattern that is being
5404 matched. Consider, for example, this pattern:
5405
5406 ([^<]|<(?!inet))+
5407
5408 It matches from wherever it starts until it encounters "<inet" or the
5409 end of the data, and is the kind of pattern that might be used when
5410 processing an XML file. Each iteration of the outer parentheses matches
5411 either one character that is not "<" or a "<" that is not followed by
5412 "inet". However, each time a parenthesis is processed, a recursion
5413 occurs, so this formulation uses a stack frame for each matched charac-
5414 ter. For a long string, a lot of stack is required. Consider now this
5415 rewritten pattern, which matches exactly the same strings:
5416
5417 ([^<]++|<(?!inet))
5418
5419 This uses very much less stack, because runs of characters that do not
5420 contain "<" are "swallowed" in one item inside the parentheses. Recur-
5421 sion happens only when a "<" character that is not followed by "inet"
5422 is encountered (and we assume this is relatively rare). A possessive
5423 quantifier is used to stop any backtracking into the runs of non-"<"
5424 characters, but that is not related to stack usage.
5425
5426 This example shows that one way of avoiding stack problems when match-
5427 ing long subject strings is to write repeated parenthesized subpatterns
5428 to match more than one character whenever possible.
5429
5430 In environments where stack memory is constrained, you might want to
5431 compile PCRE to use heap memory instead of stack for remembering back-
5432 up points. This makes it run a lot more slowly, however. Details of how
5433 to do this are given in the pcrebuild documentation.
5434
5435 In Unix-like environments, there is not often a problem with the stack
5436 unless very long strings are involved, though the default limit on
5437 stack size varies from system to system. Values from 8Mb to 64Mb are
5438 common. You can find your default limit by running the command:
5439
5440 ulimit -s
5441
5442 Unfortunately, the effect of running out of stack is often SIGSEGV,
5443 though sometimes a more explicit error message is given. You can nor-
5444 mally increase the limit on stack size by code such as this:
5445
5446 struct rlimit rlim;
5447 getrlimit(RLIMIT_STACK, &rlim);
5448 rlim.rlim_cur = 100*1024*1024;
5449 setrlimit(RLIMIT_STACK, &rlim);
5450
5451 This reads the current limits (soft and hard) using getrlimit(), then
5452 attempts to increase the soft limit to 100Mb using setrlimit(). You
5453 must do this before calling pcre_exec().
5454
5455 PCRE has an internal counter that can be used to limit the depth of
5456 recursion, and thus cause pcre_exec() to give an error code before it
5457 runs out of stack. By default, the limit is very large, and unlikely
5458 ever to operate. It can be changed when PCRE is built, and it can also
5459 be set when pcre_exec() is called. For details of these interfaces, see
5460 the pcrebuild and pcreapi documentation.
5461
5462 As a very rough rule of thumb, you should reckon on about 500 bytes per
5463 recursion. Thus, if you want to limit your stack usage to 8Mb, you
5464 should set the limit at 16000 recursions. A 64Mb stack, on the other
5465 hand, can support around 128000 recursions. The pcretest test program
5466 has a command line option (-S) that can be used to increase the size of
5467 its stack.
5468
5469 Last updated: 14 September 2006
5470 Copyright (c) 1997-2006 University of Cambridge.
5471 ------------------------------------------------------------------------------
5472
5473

Properties

Name Value
svn:eol-style native
svn:keywords "Author Date Id Revision Url"

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12