/[pcre]/code/tags/pcre-6.3/doc/pcre.txt
ViewVC logotype

Contents of /code/tags/pcre-6.3/doc/pcre.txt

Parent Directory Parent Directory | Revision Log Revision Log


Revision 84 - (show annotations)
Sat Feb 24 21:41:08 2007 UTC (14 years, 2 months ago) by nigel
File MIME type: text/plain
File size: 224275 byte(s)
Tag code/trunk as code/tags/pcre-6.3.
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. The current implementation of PCRE (release
22 6.x) corresponds approximately with Perl 5.8, including support for
23 UTF-8 encoded strings and Unicode general category properties. However,
24 this support has to be explicitly enabled; it is not the default.
25
26 In addition to the Perl-compatible matching function, PCRE also con-
27 tains an alternative matching function that matches the same compiled
28 patterns in a different way. In certain circumstances, the alternative
29 function has some advantages. For a discussion of the two matching
30 algorithms, see the pcrematching page.
31
32 PCRE is written in C and released as a C library. A number of people
33 have written wrappers and interfaces of various kinds. In particular,
34 Google Inc. have provided a comprehensive C++ wrapper. This is now
35 included as part of the PCRE distribution. The pcrecpp page has details
36 of this interface. Other people's contributions can be found in the
37 Contrib directory at the primary FTP site, which is:
38
39 ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre
40
41 Details of exactly which Perl regular expression features are and are
42 not supported by PCRE are given in separate documents. See the pcrepat-
43 tern and pcrecompat pages.
44
45 Some features of PCRE can be included, excluded, or changed when the
46 library is built. The pcre_config() function makes it possible for a
47 client to discover which features are available. The features them-
48 selves are described in the pcrebuild page. Documentation about build-
49 ing PCRE for various operating systems can be found in the README file
50 in the source distribution.
51
52 The library contains a number of undocumented internal functions and
53 data tables that are used by more than one of the exported external
54 functions, but which are not intended for use by external callers.
55 Their names all begin with "_pcre_", which hopefully will not provoke
56 any name clashes. In some environments, it is possible to control which
57 external symbols are exported when a shared library is built, and in
58 these cases the undocumented symbols are not exported.
59
60
61 USER DOCUMENTATION
62
63 The user documentation for PCRE comprises a number of different sec-
64 tions. In the "man" format, each of these is a separate "man page". In
65 the HTML format, each is a separate page, linked from the index page.
66 In the plain text format, all the sections are concatenated, for ease
67 of searching. The sections are as follows:
68
69 pcre this document
70 pcreapi details of PCRE's native C API
71 pcrebuild options for building PCRE
72 pcrecallout details of the callout feature
73 pcrecompat discussion of Perl compatibility
74 pcrecpp details of the C++ wrapper
75 pcregrep description of the pcregrep command
76 pcrematching discussion of the two matching algorithms
77 pcrepartial details of the partial matching facility
78 pcrepattern syntax and semantics of supported
79 regular expressions
80 pcreperform discussion of performance issues
81 pcreposix the POSIX-compatible C API
82 pcreprecompile details of saving and re-using precompiled patterns
83 pcresample discussion of the sample program
84 pcretest description of the pcretest testing command
85
86 In addition, in the "man" and HTML formats, there is a short page for
87 each C library function, listing its arguments and results.
88
89
90 LIMITATIONS
91
92 There are some size limitations in PCRE but it is hoped that they will
93 never in practice be relevant.
94
95 The maximum length of a compiled pattern is 65539 (sic) bytes if PCRE
96 is compiled with the default internal linkage size of 2. If you want to
97 process regular expressions that are truly enormous, you can compile
98 PCRE with an internal linkage size of 3 or 4 (see the README file in
99 the source distribution and the pcrebuild documentation for details).
100 In these cases the limit is substantially larger. However, the speed
101 of execution will be slower.
102
103 All values in repeating quantifiers must be less than 65536. The maxi-
104 mum number of capturing subpatterns is 65535.
105
106 There is no limit to the number of non-capturing subpatterns, but the
107 maximum depth of nesting of all kinds of parenthesized subpattern,
108 including capturing subpatterns, assertions, and other types of subpat-
109 tern, is 200.
110
111 The maximum length of a subject string is the largest positive number
112 that an integer variable can hold. However, when using the traditional
113 matching function, PCRE uses recursion to handle subpatterns and indef-
114 inite repetition. This means that the available stack space may limit
115 the size of a subject string that can be processed by certain patterns.
116
117
118 UTF-8 AND UNICODE PROPERTY SUPPORT
119
120 From release 3.3, PCRE has had some support for character strings
121 encoded in the UTF-8 format. For release 4.0 this was greatly extended
122 to cover most common requirements, and in release 5.0 additional sup-
123 port for Unicode general category properties was added.
124
125 In order process UTF-8 strings, you must build PCRE to include UTF-8
126 support in the code, and, in addition, you must call pcre_compile()
127 with the PCRE_UTF8 option flag. When you do this, both the pattern and
128 any subject strings that are matched against it are treated as UTF-8
129 strings instead of just strings of bytes.
130
131 If you compile PCRE with UTF-8 support, but do not use it at run time,
132 the library will be a bit bigger, but the additional run time overhead
133 is limited to testing the PCRE_UTF8 flag in several places, so should
134 not be very large.
135
136 If PCRE is built with Unicode character property support (which implies
137 UTF-8 support), the escape sequences \p{..}, \P{..}, and \X are sup-
138 ported. The available properties that can be tested are limited to the
139 general category properties such as Lu for an upper case letter or Nd
140 for a decimal number. A full list is given in the pcrepattern documen-
141 tation. The PCRE library is increased in size by about 90K when Unicode
142 property support is included.
143
144 The following comments apply when PCRE is running in UTF-8 mode:
145
146 1. When you set the PCRE_UTF8 flag, the strings passed as patterns and
147 subjects are checked for validity on entry to the relevant functions.
148 If an invalid UTF-8 string is passed, an error return is given. In some
149 situations, you may already know that your strings are valid, and
150 therefore want to skip these checks in order to improve performance. If
151 you set the PCRE_NO_UTF8_CHECK flag at compile time or at run time,
152 PCRE assumes that the pattern or subject it is given (respectively)
153 contains only valid UTF-8 codes. In this case, it does not diagnose an
154 invalid UTF-8 string. If you pass an invalid UTF-8 string to PCRE when
155 PCRE_NO_UTF8_CHECK is set, the results are undefined. Your program may
156 crash.
157
158 2. In a pattern, the escape sequence \x{...}, where the contents of the
159 braces is a string of hexadecimal digits, is interpreted as a UTF-8
160 character whose code number is the given hexadecimal number, for exam-
161 ple: \x{1234}. If a non-hexadecimal digit appears between the braces,
162 the item is not recognized. This escape sequence can be used either as
163 a literal, or within a character class.
164
165 3. The original hexadecimal escape sequence, \xhh, matches a two-byte
166 UTF-8 character if the value is greater than 127.
167
168 4. Repeat quantifiers apply to complete UTF-8 characters, not to indi-
169 vidual bytes, for example: \x{100}{3}.
170
171 5. The dot metacharacter matches one UTF-8 character instead of a sin-
172 gle byte.
173
174 6. The escape sequence \C can be used to match a single byte in UTF-8
175 mode, but its use can lead to some strange effects. This facility is
176 not available in the alternative matching function, pcre_dfa_exec().
177
178 7. The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly
179 test characters of any code value, but the characters that PCRE recog-
180 nizes as digits, spaces, or word characters remain the same set as
181 before, all with values less than 256. This remains true even when PCRE
182 includes Unicode property support, because to do otherwise would slow
183 down PCRE in many common cases. If you really want to test for a wider
184 sense of, say, "digit", you must use Unicode property tests such as
185 \p{Nd}.
186
187 8. Similarly, characters that match the POSIX named character classes
188 are all low-valued characters.
189
190 9. Case-insensitive matching applies only to characters whose values
191 are less than 128, unless PCRE is built with Unicode property support.
192 Even when Unicode property support is available, PCRE still uses its
193 own character tables when checking the case of low-valued characters,
194 so as not to degrade performance. The Unicode property information is
195 used only for characters with higher values.
196
197
198 AUTHOR
199
200 Philip Hazel
201 University Computing Service,
202 Cambridge CB2 3QG, England.
203
204 Putting an actual email address here seems to have been a spam magnet,
205 so I've taken it away. If you want to email me, use my initial and sur-
206 name, separated by a dot, at the domain ucs.cam.ac.uk.
207
208 Last updated: 07 March 2005
209 Copyright (c) 1997-2005 University of Cambridge.
210 ------------------------------------------------------------------------------
211
212
213 PCREBUILD(3) PCREBUILD(3)
214
215
216 NAME
217 PCRE - Perl-compatible regular expressions
218
219
220 PCRE BUILD-TIME OPTIONS
221
222 This document describes the optional features of PCRE that can be
223 selected when the library is compiled. They are all selected, or dese-
224 lected, by providing options to the configure script that is run before
225 the make command. The complete list of options for configure (which
226 includes the standard ones such as the selection of the installation
227 directory) can be obtained by running
228
229 ./configure --help
230
231 The following sections describe certain options whose names begin with
232 --enable or --disable. These settings specify changes to the defaults
233 for the configure command. Because of the way that configure works,
234 --enable and --disable always come in pairs, so the complementary
235 option always exists as well, but as it specifies the default, it is
236 not described.
237
238
239 C++ SUPPORT
240
241 By default, the configure script will search for a C++ compiler and C++
242 header files. If it finds them, it automatically builds the C++ wrapper
243 library for PCRE. You can disable this by adding
244
245 --disable-cpp
246
247 to the configure command.
248
249
250 UTF-8 SUPPORT
251
252 To build PCRE with support for UTF-8 character strings, add
253
254 --enable-utf8
255
256 to the configure command. Of itself, this does not make PCRE treat
257 strings as UTF-8. As well as compiling PCRE with this option, you also
258 have have to set the PCRE_UTF8 option when you call the pcre_compile()
259 function.
260
261
262 UNICODE CHARACTER PROPERTY SUPPORT
263
264 UTF-8 support allows PCRE to process character values greater than 255
265 in the strings that it handles. On its own, however, it does not pro-
266 vide any facilities for accessing the properties of such characters. If
267 you want to be able to use the pattern escapes \P, \p, and \X, which
268 refer to Unicode character properties, you must add
269
270 --enable-unicode-properties
271
272 to the configure command. This implies UTF-8 support, even if you have
273 not explicitly requested it.
274
275 Including Unicode property support adds around 90K of tables to the
276 PCRE library, approximately doubling its size. Only the general cate-
277 gory properties such as Lu and Nd are supported. Details are given in
278 the pcrepattern documentation.
279
280
281 CODE VALUE OF NEWLINE
282
283 By default, PCRE treats character 10 (linefeed) as the newline charac-
284 ter. This is the normal newline character on Unix-like systems. You can
285 compile PCRE to use character 13 (carriage return) instead by adding
286
287 --enable-newline-is-cr
288
289 to the configure command. For completeness there is also a --enable-
290 newline-is-lf option, which explicitly specifies linefeed as the new-
291 line character.
292
293
294 BUILDING SHARED AND STATIC LIBRARIES
295
296 The PCRE building process uses libtool to build both shared and static
297 Unix libraries by default. You can suppress one of these by adding one
298 of
299
300 --disable-shared
301 --disable-static
302
303 to the configure command, as required.
304
305
306 POSIX MALLOC USAGE
307
308 When PCRE is called through the POSIX interface (see the pcreposix doc-
309 umentation), additional working storage is required for holding the
310 pointers to capturing substrings, because PCRE requires three integers
311 per substring, whereas the POSIX interface provides only two. If the
312 number of expected substrings is small, the wrapper function uses space
313 on the stack, because this is faster than using malloc() for each call.
314 The default threshold above which the stack is no longer used is 10; it
315 can be changed by adding a setting such as
316
317 --with-posix-malloc-threshold=20
318
319 to the configure command.
320
321
322 LIMITING PCRE RESOURCE USAGE
323
324 Internally, PCRE has a function called match(), which it calls repeat-
325 edly (possibly recursively) when matching a pattern with the
326 pcre_exec() function. By controlling the maximum number of times this
327 function may be called during a single matching operation, a limit can
328 be placed on the resources used by a single call to pcre_exec(). The
329 limit can be changed at run time, as described in the pcreapi documen-
330 tation. The default is 10 million, but this can be changed by adding a
331 setting such as
332
333 --with-match-limit=500000
334
335 to the configure command. This setting has no effect on the
336 pcre_dfa_exec() matching function.
337
338
339 HANDLING VERY LARGE PATTERNS
340
341 Within a compiled pattern, offset values are used to point from one
342 part to another (for example, from an opening parenthesis to an alter-
343 nation metacharacter). By default, two-byte values are used for these
344 offsets, leading to a maximum size for a compiled pattern of around
345 64K. This is sufficient to handle all but the most gigantic patterns.
346 Nevertheless, some people do want to process enormous patterns, so it
347 is possible to compile PCRE to use three-byte or four-byte offsets by
348 adding a setting such as
349
350 --with-link-size=3
351
352 to the configure command. The value given must be 2, 3, or 4. Using
353 longer offsets slows down the operation of PCRE because it has to load
354 additional bytes when handling them.
355
356 If you build PCRE with an increased link size, test 2 (and test 5 if
357 you are using UTF-8) will fail. Part of the output of these tests is a
358 representation of the compiled pattern, and this changes with the link
359 size.
360
361
362 AVOIDING EXCESSIVE STACK USAGE
363
364 When matching with the pcre_exec() function, PCRE implements backtrack-
365 ing by making recursive calls to an internal function called match().
366 In environments where the size of the stack is limited, this can se-
367 verely limit PCRE's operation. (The Unix environment does not usually
368 suffer from this problem.) An alternative approach that uses memory
369 from the heap to remember data, instead of using recursive function
370 calls, has been implemented to work round this problem. If you want to
371 build a version of PCRE that works this way, add
372
373 --disable-stack-for-recursion
374
375 to the configure command. With this configuration, PCRE will use the
376 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
377 ment functions. Separate functions are provided because the usage is
378 very predictable: the block sizes requested are always the same, and
379 the blocks are always freed in reverse order. A calling program might
380 be able to implement optimized functions that perform better than the
381 standard malloc() and free() functions. PCRE runs noticeably more
382 slowly when built in this way. This option affects only the pcre_exec()
383 function; it is not relevant for the the pcre_dfa_exec() function.
384
385
386 USING EBCDIC CODE
387
388 PCRE assumes by default that it will run in an environment where the
389 character code is ASCII (or Unicode, which is a superset of ASCII).
390 PCRE can, however, be compiled to run in an EBCDIC environment by
391 adding
392
393 --enable-ebcdic
394
395 to the configure command.
396
397 Last updated: 15 August 2005
398 Copyright (c) 1997-2005 University of Cambridge.
399 ------------------------------------------------------------------------------
400
401
402 PCREMATCHING(3) PCREMATCHING(3)
403
404
405 NAME
406 PCRE - Perl-compatible regular expressions
407
408
409 PCRE MATCHING ALGORITHMS
410
411 This document describes the two different algorithms that are available
412 in PCRE for matching a compiled regular expression against a given sub-
413 ject string. The "standard" algorithm is the one provided by the
414 pcre_exec() function. This works in the same was as Perl's matching
415 function, and provides a Perl-compatible matching operation.
416
417 An alternative algorithm is provided by the pcre_dfa_exec() function;
418 this operates in a different way, and is not Perl-compatible. It has
419 advantages and disadvantages compared with the standard algorithm, and
420 these are described below.
421
422 When there is only one possible way in which a given subject string can
423 match a pattern, the two algorithms give the same answer. A difference
424 arises, however, when there are multiple possibilities. For example, if
425 the pattern
426
427 ^<.*>
428
429 is matched against the string
430
431 <something> <something else> <something further>
432
433 there are three possible answers. The standard algorithm finds only one
434 of them, whereas the DFA algorithm finds all three.
435
436
437 REGULAR EXPRESSIONS AS TREES
438
439 The set of strings that are matched by a regular expression can be rep-
440 resented as a tree structure. An unlimited repetition in the pattern
441 makes the tree of infinite size, but it is still a tree. Matching the
442 pattern to a given subject string (from a given starting point) can be
443 thought of as a search of the tree. There are two standard ways to
444 search a tree: depth-first and breadth-first, and these correspond to
445 the two matching algorithms provided by PCRE.
446
447
448 THE STANDARD MATCHING ALGORITHM
449
450 In the terminology of Jeffrey Friedl's book Mastering Regular Expres-
451 sions, the standard algorithm is an "NFA algorithm". It conducts a
452 depth-first search of the pattern tree. That is, it proceeds along a
453 single path through the tree, checking that the subject matches what is
454 required. When there is a mismatch, the algorithm tries any alterna-
455 tives at the current point, and if they all fail, it backs up to the
456 previous branch point in the tree, and tries the next alternative
457 branch at that level. This often involves backing up (moving to the
458 left) in the subject string as well. The order in which repetition
459 branches are tried is controlled by the greedy or ungreedy nature of
460 the quantifier.
461
462 If a leaf node is reached, a matching string has been found, and at
463 that point the algorithm stops. Thus, if there is more than one possi-
464 ble match, this algorithm returns the first one that it finds. Whether
465 this is the shortest, the longest, or some intermediate length depends
466 on the way the greedy and ungreedy repetition quantifiers are specified
467 in the pattern.
468
469 Because it ends up with a single path through the tree, it is rela-
470 tively straightforward for this algorithm to keep track of the sub-
471 strings that are matched by portions of the pattern in parentheses.
472 This provides support for capturing parentheses and back references.
473
474
475 THE DFA MATCHING ALGORITHM
476
477 DFA stands for "deterministic finite automaton", but you do not need to
478 understand the origins of that name. This algorithm conducts a breadth-
479 first search of the tree. Starting from the first matching point in the
480 subject, it scans the subject string from left to right, once, charac-
481 ter by character, and as it does this, it remembers all the paths
482 through the tree that represent valid matches.
483
484 The scan continues until either the end of the subject is reached, or
485 there are no more unterminated paths. At this point, terminated paths
486 represent the different matching possibilities (if there are none, the
487 match has failed). Thus, if there is more than one possible match,
488 this algorithm finds all of them, and in particular, it finds the long-
489 est. In PCRE, there is an option to stop the algorithm after the first
490 match (which is necessarily the shortest) has been found.
491
492 Note that all the matches that are found start at the same point in the
493 subject. If the pattern
494
495 cat(er(pillar)?)
496
497 is matched against the string "the caterpillar catchment", the result
498 will be the three strings "cat", "cater", and "caterpillar" that start
499 at the fourth character of the subject. The algorithm does not automat-
500 ically move on to find matches that start at later positions.
501
502 There are a number of features of PCRE regular expressions that are not
503 supported by the DFA matching algorithm. They are as follows:
504
505 1. Because the algorithm finds all possible matches, the greedy or
506 ungreedy nature of repetition quantifiers is not relevant. Greedy and
507 ungreedy quantifiers are treated in exactly the same way.
508
509 2. When dealing with multiple paths through the tree simultaneously, it
510 is not straightforward to keep track of captured substrings for the
511 different matching possibilities, and PCRE's implementation of this
512 algorithm does not attempt to do this. This means that no captured sub-
513 strings are available.
514
515 3. Because no substrings are captured, back references within the pat-
516 tern are not supported, and cause errors if encountered.
517
518 4. For the same reason, conditional expressions that use a backrefer-
519 ence as the condition are not supported.
520
521 5. Callouts are supported, but the value of the capture_top field is
522 always 1, and the value of the capture_last field is always -1.
523
524 6. The \C escape sequence, which (in the standard algorithm) matches a
525 single byte, even in UTF-8 mode, is not supported because the DFA algo-
526 rithm moves through the subject string one character at a time, for all
527 active paths through the tree.
528
529
530 ADVANTAGES OF THE DFA ALGORITHM
531
532 Using the DFA matching algorithm provides the following advantages:
533
534 1. All possible matches (at a single point in the subject) are automat-
535 ically found, and in particular, the longest match is found. To find
536 more than one match using the standard algorithm, you have to do kludgy
537 things with callouts.
538
539 2. There is much better support for partial matching. The restrictions
540 on the content of the pattern that apply when using the standard algo-
541 rithm for partial matching do not apply to the DFA algorithm. For non-
542 anchored patterns, the starting position of a partial match is avail-
543 able.
544
545 3. Because the DFA algorithm scans the subject string just once, and
546 never needs to backtrack, it is possible to pass very long subject
547 strings to the matching function in several pieces, checking for par-
548 tial matching each time.
549
550
551 DISADVANTAGES OF THE DFA ALGORITHM
552
553 The DFA algorithm suffers from a number of disadvantages:
554
555 1. It is substantially slower than the standard algorithm. This is
556 partly because it has to search for all possible matches, but is also
557 because it is less susceptible to optimization.
558
559 2. Capturing parentheses and back references are not supported.
560
561 3. The "atomic group" feature of PCRE regular expressions is supported,
562 but does not provide the advantage that it does for the standard algo-
563 rithm.
564
565 Last updated: 28 February 2005
566 Copyright (c) 1997-2005 University of Cambridge.
567 ------------------------------------------------------------------------------
568
569
570 PCREAPI(3) PCREAPI(3)
571
572
573 NAME
574 PCRE - Perl-compatible regular expressions
575
576
577 PCRE NATIVE API
578
579 #include <pcre.h>
580
581 pcre *pcre_compile(const char *pattern, int options,
582 const char **errptr, int *erroffset,
583 const unsigned char *tableptr);
584
585 pcre *pcre_compile2(const char *pattern, int options,
586 int *errorcodeptr,
587 const char **errptr, int *erroffset,
588 const unsigned char *tableptr);
589
590 pcre_extra *pcre_study(const pcre *code, int options,
591 const char **errptr);
592
593 int pcre_exec(const pcre *code, const pcre_extra *extra,
594 const char *subject, int length, int startoffset,
595 int options, int *ovector, int ovecsize);
596
597 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
598 const char *subject, int length, int startoffset,
599 int options, int *ovector, int ovecsize,
600 int *workspace, int wscount);
601
602 int pcre_copy_named_substring(const pcre *code,
603 const char *subject, int *ovector,
604 int stringcount, const char *stringname,
605 char *buffer, int buffersize);
606
607 int pcre_copy_substring(const char *subject, int *ovector,
608 int stringcount, int stringnumber, char *buffer,
609 int buffersize);
610
611 int pcre_get_named_substring(const pcre *code,
612 const char *subject, int *ovector,
613 int stringcount, const char *stringname,
614 const char **stringptr);
615
616 int pcre_get_stringnumber(const pcre *code,
617 const char *name);
618
619 int pcre_get_substring(const char *subject, int *ovector,
620 int stringcount, int stringnumber,
621 const char **stringptr);
622
623 int pcre_get_substring_list(const char *subject,
624 int *ovector, int stringcount, const char ***listptr);
625
626 void pcre_free_substring(const char *stringptr);
627
628 void pcre_free_substring_list(const char **stringptr);
629
630 const unsigned char *pcre_maketables(void);
631
632 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
633 int what, void *where);
634
635 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
636
637 int pcre_refcount(pcre *code, int adjust);
638
639 int pcre_config(int what, void *where);
640
641 char *pcre_version(void);
642
643 void *(*pcre_malloc)(size_t);
644
645 void (*pcre_free)(void *);
646
647 void *(*pcre_stack_malloc)(size_t);
648
649 void (*pcre_stack_free)(void *);
650
651 int (*pcre_callout)(pcre_callout_block *);
652
653
654 PCRE API OVERVIEW
655
656 PCRE has its own native API, which is described in this document. There
657 is also a set of wrapper functions that correspond to the POSIX regular
658 expression API. These are described in the pcreposix documentation.
659 Both of these APIs define a set of C function calls. A C++ wrapper is
660 distributed with PCRE. It is documented in the pcrecpp page.
661
662 The native API C function prototypes are defined in the header file
663 pcre.h, and on Unix systems the library itself is called libpcre. It
664 can normally be accessed by adding -lpcre to the command for linking an
665 application that uses PCRE. The header file defines the macros
666 PCRE_MAJOR and PCRE_MINOR to contain the major and minor release num-
667 bers for the library. Applications can use these to include support
668 for different releases of PCRE.
669
670 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
671 pcre_exec() are used for compiling and matching regular expressions in
672 a Perl-compatible manner. A sample program that demonstrates the sim-
673 plest way of using them is provided in the file called pcredemo.c in
674 the source distribution. The pcresample documentation describes how to
675 run it.
676
677 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
678 ble, is also provided. This uses a different algorithm for the match-
679 ing. This allows it to find all possible matches (at a given point in
680 the subject), not just one. However, this algorithm does not return
681 captured substrings. A description of the two matching algorithms and
682 their advantages and disadvantages is given in the pcrematching docu-
683 mentation.
684
685 In addition to the main compiling and matching functions, there are
686 convenience functions for extracting captured substrings from a subject
687 string that is matched by pcre_exec(). They are:
688
689 pcre_copy_substring()
690 pcre_copy_named_substring()
691 pcre_get_substring()
692 pcre_get_named_substring()
693 pcre_get_substring_list()
694 pcre_get_stringnumber()
695
696 pcre_free_substring() and pcre_free_substring_list() are also provided,
697 to free the memory used for extracted strings.
698
699 The function pcre_maketables() is used to build a set of character
700 tables in the current locale for passing to pcre_compile(),
701 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
702 provided for specialist use. Most commonly, no special tables are
703 passed, in which case internal tables that are generated when PCRE is
704 built are used.
705
706 The function pcre_fullinfo() is used to find out information about a
707 compiled pattern; pcre_info() is an obsolete version that returns only
708 some of the available information, but is retained for backwards com-
709 patibility. The function pcre_version() returns a pointer to a string
710 containing the version of PCRE and its date of release.
711
712 The function pcre_refcount() maintains a reference count in a data
713 block containing a compiled pattern. This is provided for the benefit
714 of object-oriented applications.
715
716 The global variables pcre_malloc and pcre_free initially contain the
717 entry points of the standard malloc() and free() functions, respec-
718 tively. PCRE calls the memory management functions via these variables,
719 so a calling program can replace them if it wishes to intercept the
720 calls. This should be done before calling any PCRE functions.
721
722 The global variables pcre_stack_malloc and pcre_stack_free are also
723 indirections to memory management functions. These special functions
724 are used only when PCRE is compiled to use the heap for remembering
725 data, instead of recursive function calls, when running the pcre_exec()
726 function. This is a non-standard way of building PCRE, for use in envi-
727 ronments that have limited stacks. Because of the greater use of memory
728 management, it runs more slowly. Separate functions are provided so
729 that special-purpose external code can be used for this case. When
730 used, these functions are always called in a stack-like manner (last
731 obtained, first freed), and always for memory blocks of the same size.
732
733 The global variable pcre_callout initially contains NULL. It can be set
734 by the caller to a "callout" function, which PCRE will then call at
735 specified points during a matching operation. Details are given in the
736 pcrecallout documentation.
737
738
739 MULTITHREADING
740
741 The PCRE functions can be used in multi-threading applications, with
742 the proviso that the memory management functions pointed to by
743 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
744 callout function pointed to by pcre_callout, are shared by all threads.
745
746 The compiled form of a regular expression is not altered during match-
747 ing, so the same compiled pattern can safely be used by several threads
748 at once.
749
750
751 SAVING PRECOMPILED PATTERNS FOR LATER USE
752
753 The compiled form of a regular expression can be saved and re-used at a
754 later time, possibly by a different program, and even on a host other
755 than the one on which it was compiled. Details are given in the
756 pcreprecompile documentation.
757
758
759 CHECKING BUILD-TIME OPTIONS
760
761 int pcre_config(int what, void *where);
762
763 The function pcre_config() makes it possible for a PCRE client to dis-
764 cover which optional features have been compiled into the PCRE library.
765 The pcrebuild documentation has more details about these optional fea-
766 tures.
767
768 The first argument for pcre_config() is an integer, specifying which
769 information is required; the second argument is a pointer to a variable
770 into which the information is placed. The following information is
771 available:
772
773 PCRE_CONFIG_UTF8
774
775 The output is an integer that is set to one if UTF-8 support is avail-
776 able; otherwise it is set to zero.
777
778 PCRE_CONFIG_UNICODE_PROPERTIES
779
780 The output is an integer that is set to one if support for Unicode
781 character properties is available; otherwise it is set to zero.
782
783 PCRE_CONFIG_NEWLINE
784
785 The output is an integer that is set to the value of the code that is
786 used for the newline character. It is either linefeed (10) or carriage
787 return (13), and should normally be the standard character for your
788 operating system.
789
790 PCRE_CONFIG_LINK_SIZE
791
792 The output is an integer that contains the number of bytes used for
793 internal linkage in compiled regular expressions. The value is 2, 3, or
794 4. Larger values allow larger regular expressions to be compiled, at
795 the expense of slower matching. The default value of 2 is sufficient
796 for all but the most massive patterns, since it allows the compiled
797 pattern to be up to 64K in size.
798
799 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
800
801 The output is an integer that contains the threshold above which the
802 POSIX interface uses malloc() for output vectors. Further details are
803 given in the pcreposix documentation.
804
805 PCRE_CONFIG_MATCH_LIMIT
806
807 The output is an integer that gives the default limit for the number of
808 internal matching function calls in a pcre_exec() execution. Further
809 details are given with pcre_exec() below.
810
811 PCRE_CONFIG_STACKRECURSE
812
813 The output is an integer that is set to one if internal recursion when
814 running pcre_exec() is implemented by recursive function calls that use
815 the stack to remember their state. This is the usual way that PCRE is
816 compiled. The output is zero if PCRE was compiled to use blocks of data
817 on the heap instead of recursive function calls. In this case,
818 pcre_stack_malloc and pcre_stack_free are called to manage memory
819 blocks on the heap, thus avoiding the use of the stack.
820
821
822 COMPILING A PATTERN
823
824 pcre *pcre_compile(const char *pattern, int options,
825 const char **errptr, int *erroffset,
826 const unsigned char *tableptr);
827
828 pcre *pcre_compile2(const char *pattern, int options,
829 int *errorcodeptr,
830 const char **errptr, int *erroffset,
831 const unsigned char *tableptr);
832
833 Either of the functions pcre_compile() or pcre_compile2() can be called
834 to compile a pattern into an internal form. The only difference between
835 the two interfaces is that pcre_compile2() has an additional argument,
836 errorcodeptr, via which a numerical error code can be returned.
837
838 The pattern is a C string terminated by a binary zero, and is passed in
839 the pattern argument. A pointer to a single block of memory that is
840 obtained via pcre_malloc is returned. This contains the compiled code
841 and related data. The pcre type is defined for the returned block; this
842 is a typedef for a structure whose contents are not externally defined.
843 It is up to the caller to free the memory when it is no longer
844 required.
845
846 Although the compiled code of a PCRE regex is relocatable, that is, it
847 does not depend on memory location, the complete pcre data block is not
848 fully relocatable, because it may contain a copy of the tableptr argu-
849 ment, which is an address (see below).
850
851 The options argument contains independent bits that affect the compila-
852 tion. It should be zero if no options are required. The available
853 options are described below. Some of them, in particular, those that
854 are compatible with Perl, can also be set and unset from within the
855 pattern (see the detailed description in the pcrepattern documenta-
856 tion). For these options, the contents of the options argument speci-
857 fies their initial settings at the start of compilation and execution.
858 The PCRE_ANCHORED option can be set at the time of matching as well as
859 at compile time.
860
861 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
862 if compilation of a pattern fails, pcre_compile() returns NULL, and
863 sets the variable pointed to by errptr to point to a textual error mes-
864 sage. The offset from the start of the pattern to the character where
865 the error was discovered is placed in the variable pointed to by
866 erroffset, which must not be NULL. If it is, an immediate error is
867 given.
868
869 If pcre_compile2() is used instead of pcre_compile(), and the error-
870 codeptr argument is not NULL, a non-zero error code number is returned
871 via this argument in the event of an error. This is in addition to the
872 textual error message. Error codes and messages are listed below.
873
874 If the final argument, tableptr, is NULL, PCRE uses a default set of
875 character tables that are built when PCRE is compiled, using the
876 default C locale. Otherwise, tableptr must be an address that is the
877 result of a call to pcre_maketables(). This value is stored with the
878 compiled pattern, and used again by pcre_exec(), unless another table
879 pointer is passed to it. For more discussion, see the section on locale
880 support below.
881
882 This code fragment shows a typical straightforward call to pcre_com-
883 pile():
884
885 pcre *re;
886 const char *error;
887 int erroffset;
888 re = pcre_compile(
889 "^A.*Z", /* the pattern */
890 0, /* default options */
891 &error, /* for error message */
892 &erroffset, /* for error offset */
893 NULL); /* use default character tables */
894
895 The following names for option bits are defined in the pcre.h header
896 file:
897
898 PCRE_ANCHORED
899
900 If this bit is set, the pattern is forced to be "anchored", that is, it
901 is constrained to match only at the first matching point in the string
902 that is being searched (the "subject string"). This effect can also be
903 achieved by appropriate constructs in the pattern itself, which is the
904 only way to do it in Perl.
905
906 PCRE_AUTO_CALLOUT
907
908 If this bit is set, pcre_compile() automatically inserts callout items,
909 all with number 255, before each pattern item. For discussion of the
910 callout facility, see the pcrecallout documentation.
911
912 PCRE_CASELESS
913
914 If this bit is set, letters in the pattern match both upper and lower
915 case letters. It is equivalent to Perl's /i option, and it can be
916 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
917 always understands the concept of case for characters whose values are
918 less than 128, so caseless matching is always possible. For characters
919 with higher values, the concept of case is supported if PCRE is com-
920 piled with Unicode property support, but not otherwise. If you want to
921 use caseless matching for characters 128 and above, you must ensure
922 that PCRE is compiled with Unicode property support as well as with
923 UTF-8 support.
924
925 PCRE_DOLLAR_ENDONLY
926
927 If this bit is set, a dollar metacharacter in the pattern matches only
928 at the end of the subject string. Without this option, a dollar also
929 matches immediately before the final character if it is a newline (but
930 not before any other newlines). The PCRE_DOLLAR_ENDONLY option is
931 ignored if PCRE_MULTILINE is set. There is no equivalent to this option
932 in Perl, and no way to set it within a pattern.
933
934 PCRE_DOTALL
935
936 If this bit is set, a dot metacharater in the pattern matches all char-
937 acters, including newlines. Without it, newlines are excluded. This
938 option is equivalent to Perl's /s option, and it can be changed within
939 a pattern by a (?s) option setting. A negative class such as [^a]
940 always matches a newline character, independent of the setting of this
941 option.
942
943 PCRE_EXTENDED
944
945 If this bit is set, whitespace data characters in the pattern are
946 totally ignored except when escaped or inside a character class. White-
947 space does not include the VT character (code 11). In addition, charac-
948 ters between an unescaped # outside a character class and the next new-
949 line character, inclusive, are also ignored. This is equivalent to
950 Perl's /x option, and it can be changed within a pattern by a (?x)
951 option setting.
952
953 This option makes it possible to include comments inside complicated
954 patterns. Note, however, that this applies only to data characters.
955 Whitespace characters may never appear within special character
956 sequences in a pattern, for example within the sequence (?( which
957 introduces a conditional subpattern.
958
959 PCRE_EXTRA
960
961 This option was invented in order to turn on additional functionality
962 of PCRE that is incompatible with Perl, but it is currently of very
963 little use. When set, any backslash in a pattern that is followed by a
964 letter that has no special meaning causes an error, thus reserving
965 these combinations for future expansion. By default, as in Perl, a
966 backslash followed by a letter with no special meaning is treated as a
967 literal. There are at present no other features controlled by this
968 option. It can also be set by a (?X) option setting within a pattern.
969
970 PCRE_FIRSTLINE
971
972 If this option is set, an unanchored pattern is required to match
973 before or at the first newline character in the subject string, though
974 the matched text may continue over the newline.
975
976 PCRE_MULTILINE
977
978 By default, PCRE treats the subject string as consisting of a single
979 line of characters (even if it actually contains newlines). The "start
980 of line" metacharacter (^) matches only at the start of the string,
981 while the "end of line" metacharacter ($) matches only at the end of
982 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
983 is set). This is the same as Perl.
984
985 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
986 constructs match immediately following or immediately before any new-
987 line in the subject string, respectively, as well as at the very start
988 and end. This is equivalent to Perl's /m option, and it can be changed
989 within a pattern by a (?m) option setting. If there are no "\n" charac-
990 ters in a subject string, or no occurrences of ^ or $ in a pattern,
991 setting PCRE_MULTILINE has no effect.
992
993 PCRE_NO_AUTO_CAPTURE
994
995 If this option is set, it disables the use of numbered capturing paren-
996 theses in the pattern. Any opening parenthesis that is not followed by
997 ? behaves as if it were followed by ?: but named parentheses can still
998 be used for capturing (and they acquire numbers in the usual way).
999 There is no equivalent of this option in Perl.
1000
1001 PCRE_UNGREEDY
1002
1003 This option inverts the "greediness" of the quantifiers so that they
1004 are not greedy by default, but become greedy if followed by "?". It is
1005 not compatible with Perl. It can also be set by a (?U) option setting
1006 within the pattern.
1007
1008 PCRE_UTF8
1009
1010 This option causes PCRE to regard both the pattern and the subject as
1011 strings of UTF-8 characters instead of single-byte character strings.
1012 However, it is available only when PCRE is built to include UTF-8 sup-
1013 port. If not, the use of this option provokes an error. Details of how
1014 this option changes the behaviour of PCRE are given in the section on
1015 UTF-8 support in the main pcre page.
1016
1017 PCRE_NO_UTF8_CHECK
1018
1019 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1020 automatically checked. If an invalid UTF-8 sequence of bytes is found,
1021 pcre_compile() returns an error. If you already know that your pattern
1022 is valid, and you want to skip this check for performance reasons, you
1023 can set the PCRE_NO_UTF8_CHECK option. When it is set, the effect of
1024 passing an invalid UTF-8 string as a pattern is undefined. It may cause
1025 your program to crash. Note that this option can also be passed to
1026 pcre_exec() and pcre_dfa_exec(), to suppress the UTF-8 validity check-
1027 ing of subject strings.
1028
1029
1030 COMPILATION ERROR CODES
1031
1032 The following table lists the error codes than may be returned by
1033 pcre_compile2(), along with the error messages that may be returned by
1034 both compiling functions.
1035
1036 0 no error
1037 1 \ at end of pattern
1038 2 \c at end of pattern
1039 3 unrecognized character follows \
1040 4 numbers out of order in {} quantifier
1041 5 number too big in {} quantifier
1042 6 missing terminating ] for character class
1043 7 invalid escape sequence in character class
1044 8 range out of order in character class
1045 9 nothing to repeat
1046 10 operand of unlimited repeat could match the empty string
1047 11 internal error: unexpected repeat
1048 12 unrecognized character after (?
1049 13 POSIX named classes are supported only within a class
1050 14 missing )
1051 15 reference to non-existent subpattern
1052 16 erroffset passed as NULL
1053 17 unknown option bit(s) set
1054 18 missing ) after comment
1055 19 parentheses nested too deeply
1056 20 regular expression too large
1057 21 failed to get memory
1058 22 unmatched parentheses
1059 23 internal error: code overflow
1060 24 unrecognized character after (?<
1061 25 lookbehind assertion is not fixed length
1062 26 malformed number after (?(
1063 27 conditional group contains more than two branches
1064 28 assertion expected after (?(
1065 29 (?R or (?digits must be followed by )
1066 30 unknown POSIX class name
1067 31 POSIX collating elements are not supported
1068 32 this version of PCRE is not compiled with PCRE_UTF8 support
1069 33 spare error
1070 34 character value in \x{...} sequence is too large
1071 35 invalid condition (?(0)
1072 36 \C not allowed in lookbehind assertion
1073 37 PCRE does not support \L, \l, \N, \U, or \u
1074 38 number after (?C is > 255
1075 39 closing ) for (?C expected
1076 40 recursive call could loop indefinitely
1077 41 unrecognized character after (?P
1078 42 syntax error after (?P
1079 43 two named groups have the same name
1080 44 invalid UTF-8 string
1081 45 support for \P, \p, and \X has not been compiled
1082 46 malformed \P or \p sequence
1083 47 unknown property name after \P or \p
1084
1085
1086 STUDYING A PATTERN
1087
1088 pcre_extra *pcre_study(const pcre *code, int options
1089 const char **errptr);
1090
1091 If a compiled pattern is going to be used several times, it is worth
1092 spending more time analyzing it in order to speed up the time taken for
1093 matching. The function pcre_study() takes a pointer to a compiled pat-
1094 tern as its first argument. If studying the pattern produces additional
1095 information that will help speed up matching, pcre_study() returns a
1096 pointer to a pcre_extra block, in which the study_data field points to
1097 the results of the study.
1098
1099 The returned value from pcre_study() can be passed directly to
1100 pcre_exec(). However, a pcre_extra block also contains other fields
1101 that can be set by the caller before the block is passed; these are
1102 described below in the section on matching a pattern.
1103
1104 If studying the pattern does not produce any additional information
1105 pcre_study() returns NULL. In that circumstance, if the calling program
1106 wants to pass any of the other fields to pcre_exec(), it must set up
1107 its own pcre_extra block.
1108
1109 The second argument of pcre_study() contains option bits. At present,
1110 no options are defined, and this argument should always be zero.
1111
1112 The third argument for pcre_study() is a pointer for an error message.
1113 If studying succeeds (even if no data is returned), the variable it
1114 points to is set to NULL. Otherwise it points to a textual error mes-
1115 sage. You should therefore test the error pointer for NULL after call-
1116 ing pcre_study(), to be sure that it has run successfully.
1117
1118 This is a typical call to pcre_study():
1119
1120 pcre_extra *pe;
1121 pe = pcre_study(
1122 re, /* result of pcre_compile() */
1123 0, /* no options exist */
1124 &error); /* set to NULL or points to a message */
1125
1126 At present, studying a pattern is useful only for non-anchored patterns
1127 that do not have a single fixed starting character. A bitmap of possi-
1128 ble starting bytes is created.
1129
1130
1131 LOCALE SUPPORT
1132
1133 PCRE handles caseless matching, and determines whether characters are
1134 letters digits, or whatever, by reference to a set of tables, indexed
1135 by character value. When running in UTF-8 mode, this applies only to
1136 characters with codes less than 128. Higher-valued codes never match
1137 escapes such as \w or \d, but can be tested with \p if PCRE is built
1138 with Unicode character property support.
1139
1140 An internal set of tables is created in the default C locale when PCRE
1141 is built. This is used when the final argument of pcre_compile() is
1142 NULL, and is sufficient for many applications. An alternative set of
1143 tables can, however, be supplied. These may be created in a different
1144 locale from the default. As more and more applications change to using
1145 Unicode, the need for this locale support is expected to die away.
1146
1147 External tables are built by calling the pcre_maketables() function,
1148 which has no arguments, in the relevant locale. The result can then be
1149 passed to pcre_compile() or pcre_exec() as often as necessary. For
1150 example, to build and use tables that are appropriate for the French
1151 locale (where accented characters with values greater than 128 are
1152 treated as letters), the following code could be used:
1153
1154 setlocale(LC_CTYPE, "fr_FR");
1155 tables = pcre_maketables();
1156 re = pcre_compile(..., tables);
1157
1158 When pcre_maketables() runs, the tables are built in memory that is
1159 obtained via pcre_malloc. It is the caller's responsibility to ensure
1160 that the memory containing the tables remains available for as long as
1161 it is needed.
1162
1163 The pointer that is passed to pcre_compile() is saved with the compiled
1164 pattern, and the same tables are used via this pointer by pcre_study()
1165 and normally also by pcre_exec(). Thus, by default, for any single pat-
1166 tern, compilation, studying and matching all happen in the same locale,
1167 but different patterns can be compiled in different locales.
1168
1169 It is possible to pass a table pointer or NULL (indicating the use of
1170 the internal tables) to pcre_exec(). Although not intended for this
1171 purpose, this facility could be used to match a pattern in a different
1172 locale from the one in which it was compiled. Passing table pointers at
1173 run time is discussed below in the section on matching a pattern.
1174
1175
1176 INFORMATION ABOUT A PATTERN
1177
1178 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1179 int what, void *where);
1180
1181 The pcre_fullinfo() function returns information about a compiled pat-
1182 tern. It replaces the obsolete pcre_info() function, which is neverthe-
1183 less retained for backwards compability (and is documented below).
1184
1185 The first argument for pcre_fullinfo() is a pointer to the compiled
1186 pattern. The second argument is the result of pcre_study(), or NULL if
1187 the pattern was not studied. The third argument specifies which piece
1188 of information is required, and the fourth argument is a pointer to a
1189 variable to receive the data. The yield of the function is zero for
1190 success, or one of the following negative numbers:
1191
1192 PCRE_ERROR_NULL the argument code was NULL
1193 the argument where was NULL
1194 PCRE_ERROR_BADMAGIC the "magic number" was not found
1195 PCRE_ERROR_BADOPTION the value of what was invalid
1196
1197 The "magic number" is placed at the start of each compiled pattern as
1198 an simple check against passing an arbitrary memory pointer. Here is a
1199 typical call of pcre_fullinfo(), to obtain the length of the compiled
1200 pattern:
1201
1202 int rc;
1203 unsigned long int length;
1204 rc = pcre_fullinfo(
1205 re, /* result of pcre_compile() */
1206 pe, /* result of pcre_study(), or NULL */
1207 PCRE_INFO_SIZE, /* what is required */
1208 &length); /* where to put the data */
1209
1210 The possible values for the third argument are defined in pcre.h, and
1211 are as follows:
1212
1213 PCRE_INFO_BACKREFMAX
1214
1215 Return the number of the highest back reference in the pattern. The
1216 fourth argument should point to an int variable. Zero is returned if
1217 there are no back references.
1218
1219 PCRE_INFO_CAPTURECOUNT
1220
1221 Return the number of capturing subpatterns in the pattern. The fourth
1222 argument should point to an int variable.
1223
1224 PCRE_INFO_DEFAULT_TABLES
1225
1226 Return a pointer to the internal default character tables within PCRE.
1227 The fourth argument should point to an unsigned char * variable. This
1228 information call is provided for internal use by the pcre_study() func-
1229 tion. External callers can cause PCRE to use its internal tables by
1230 passing a NULL table pointer.
1231
1232 PCRE_INFO_FIRSTBYTE
1233
1234 Return information about the first byte of any matched string, for a
1235 non-anchored pattern. (This option used to be called
1236 PCRE_INFO_FIRSTCHAR; the old name is still recognized for backwards
1237 compatibility.)
1238
1239 If there is a fixed first byte, for example, from a pattern such as
1240 (cat|cow|coyote), it is returned in the integer pointed to by where.
1241 Otherwise, if either
1242
1243 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
1244 branch starts with "^", or
1245
1246 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
1247 set (if it were set, the pattern would be anchored),
1248
1249 -1 is returned, indicating that the pattern matches only at the start
1250 of a subject string or after any newline within the string. Otherwise
1251 -2 is returned. For anchored patterns, -2 is returned.
1252
1253 PCRE_INFO_FIRSTTABLE
1254
1255 If the pattern was studied, and this resulted in the construction of a
1256 256-bit table indicating a fixed set of bytes for the first byte in any
1257 matching string, a pointer to the table is returned. Otherwise NULL is
1258 returned. The fourth argument should point to an unsigned char * vari-
1259 able.
1260
1261 PCRE_INFO_LASTLITERAL
1262
1263 Return the value of the rightmost literal byte that must exist in any
1264 matched string, other than at its start, if such a byte has been
1265 recorded. The fourth argument should point to an int variable. If there
1266 is no such byte, -1 is returned. For anchored patterns, a last literal
1267 byte is recorded only if it follows something of variable length. For
1268 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
1269 /^a\dz\d/ the returned value is -1.
1270
1271 PCRE_INFO_NAMECOUNT
1272 PCRE_INFO_NAMEENTRYSIZE
1273 PCRE_INFO_NAMETABLE
1274
1275 PCRE supports the use of named as well as numbered capturing parenthe-
1276 ses. The names are just an additional way of identifying the parenthe-
1277 ses, which still acquire numbers. A convenience function called
1278 pcre_get_named_substring() is provided for extracting an individual
1279 captured substring by name. It is also possible to extract the data
1280 directly, by first converting the name to a number in order to access
1281 the correct pointers in the output vector (described with pcre_exec()
1282 below). To do the conversion, you need to use the name-to-number map,
1283 which is described by these three values.
1284
1285 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
1286 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
1287 of each entry; both of these return an int value. The entry size
1288 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
1289 a pointer to the first entry of the table (a pointer to char). The
1290 first two bytes of each entry are the number of the capturing parenthe-
1291 sis, most significant byte first. The rest of the entry is the corre-
1292 sponding name, zero terminated. The names are in alphabetical order.
1293 For example, consider the following pattern (assume PCRE_EXTENDED is
1294 set, so white space - including newlines - is ignored):
1295
1296 (?P<date> (?P<year>(\d\d)?\d\d) -
1297 (?P<month>\d\d) - (?P<day>\d\d) )
1298
1299 There are four named subpatterns, so the table has four entries, and
1300 each entry in the table is eight bytes long. The table is as follows,
1301 with non-printing bytes shows in hexadecimal, and undefined bytes shown
1302 as ??:
1303
1304 00 01 d a t e 00 ??
1305 00 05 d a y 00 ?? ??
1306 00 04 m o n t h 00
1307 00 02 y e a r 00 ??
1308
1309 When writing code to extract data from named subpatterns using the
1310 name-to-number map, remember that the length of each entry is likely to
1311 be different for each compiled pattern.
1312
1313 PCRE_INFO_OPTIONS
1314
1315 Return a copy of the options with which the pattern was compiled. The
1316 fourth argument should point to an unsigned long int variable. These
1317 option bits are those specified in the call to pcre_compile(), modified
1318 by any top-level option settings within the pattern itself.
1319
1320 A pattern is automatically anchored by PCRE if all of its top-level
1321 alternatives begin with one of the following:
1322
1323 ^ unless PCRE_MULTILINE is set
1324 \A always
1325 \G always
1326 .* if PCRE_DOTALL is set and there are no back
1327 references to the subpattern in which .* appears
1328
1329 For such patterns, the PCRE_ANCHORED bit is set in the options returned
1330 by pcre_fullinfo().
1331
1332 PCRE_INFO_SIZE
1333
1334 Return the size of the compiled pattern, that is, the value that was
1335 passed as the argument to pcre_malloc() when PCRE was getting memory in
1336 which to place the compiled data. The fourth argument should point to a
1337 size_t variable.
1338
1339 PCRE_INFO_STUDYSIZE
1340
1341 Return the size of the data block pointed to by the study_data field in
1342 a pcre_extra block. That is, it is the value that was passed to
1343 pcre_malloc() when PCRE was getting memory into which to place the data
1344 created by pcre_study(). The fourth argument should point to a size_t
1345 variable.
1346
1347
1348 OBSOLETE INFO FUNCTION
1349
1350 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
1351
1352 The pcre_info() function is now obsolete because its interface is too
1353 restrictive to return all the available data about a compiled pattern.
1354 New programs should use pcre_fullinfo() instead. The yield of
1355 pcre_info() is the number of capturing subpatterns, or one of the fol-
1356 lowing negative numbers:
1357
1358 PCRE_ERROR_NULL the argument code was NULL
1359 PCRE_ERROR_BADMAGIC the "magic number" was not found
1360
1361 If the optptr argument is not NULL, a copy of the options with which
1362 the pattern was compiled is placed in the integer it points to (see
1363 PCRE_INFO_OPTIONS above).
1364
1365 If the pattern is not anchored and the firstcharptr argument is not
1366 NULL, it is used to pass back information about the first character of
1367 any matched string (see PCRE_INFO_FIRSTBYTE above).
1368
1369
1370 REFERENCE COUNTS
1371
1372 int pcre_refcount(pcre *code, int adjust);
1373
1374 The pcre_refcount() function is used to maintain a reference count in
1375 the data block that contains a compiled pattern. It is provided for the
1376 benefit of applications that operate in an object-oriented manner,
1377 where different parts of the application may be using the same compiled
1378 pattern, but you want to free the block when they are all done.
1379
1380 When a pattern is compiled, the reference count field is initialized to
1381 zero. It is changed only by calling this function, whose action is to
1382 add the adjust value (which may be positive or negative) to it. The
1383 yield of the function is the new value. However, the value of the count
1384 is constrained to lie between 0 and 65535, inclusive. If the new value
1385 is outside these limits, it is forced to the appropriate limit value.
1386
1387 Except when it is zero, the reference count is not correctly preserved
1388 if a pattern is compiled on one host and then transferred to a host
1389 whose byte-order is different. (This seems a highly unlikely scenario.)
1390
1391
1392 MATCHING A PATTERN: THE TRADITIONAL FUNCTION
1393
1394 int pcre_exec(const pcre *code, const pcre_extra *extra,
1395 const char *subject, int length, int startoffset,
1396 int options, int *ovector, int ovecsize);
1397
1398 The function pcre_exec() is called to match a subject string against a
1399 compiled pattern, which is passed in the code argument. If the pattern
1400 has been studied, the result of the study should be passed in the extra
1401 argument. This function is the main matching facility of the library,
1402 and it operates in a Perl-like manner. For specialist use there is also
1403 an alternative matching function, which is described below in the sec-
1404 tion about the pcre_dfa_exec() function.
1405
1406 In most applications, the pattern will have been compiled (and option-
1407 ally studied) in the same process that calls pcre_exec(). However, it
1408 is possible to save compiled patterns and study data, and then use them
1409 later in different processes, possibly even on different hosts. For a
1410 discussion about this, see the pcreprecompile documentation.
1411
1412 Here is an example of a simple call to pcre_exec():
1413
1414 int rc;
1415 int ovector[30];
1416 rc = pcre_exec(
1417 re, /* result of pcre_compile() */
1418 NULL, /* we didn't study the pattern */
1419 "some string", /* the subject string */
1420 11, /* the length of the subject string */
1421 0, /* start at offset 0 in the subject */
1422 0, /* default options */
1423 ovector, /* vector of integers for substring information */
1424 30); /* number of elements (NOT size in bytes) */
1425
1426 Extra data for pcre_exec()
1427
1428 If the extra argument is not NULL, it must point to a pcre_extra data
1429 block. The pcre_study() function returns such a block (when it doesn't
1430 return NULL), but you can also create one for yourself, and pass addi-
1431 tional information in it. The fields in a pcre_extra block are as fol-
1432 lows:
1433
1434 unsigned long int flags;
1435 void *study_data;
1436 unsigned long int match_limit;
1437 void *callout_data;
1438 const unsigned char *tables;
1439
1440 The flags field is a bitmap that specifies which of the other fields
1441 are set. The flag bits are:
1442
1443 PCRE_EXTRA_STUDY_DATA
1444 PCRE_EXTRA_MATCH_LIMIT
1445 PCRE_EXTRA_CALLOUT_DATA
1446 PCRE_EXTRA_TABLES
1447
1448 Other flag bits should be set to zero. The study_data field is set in
1449 the pcre_extra block that is returned by pcre_study(), together with
1450 the appropriate flag bit. You should not set this yourself, but you may
1451 add to the block by setting the other fields and their corresponding
1452 flag bits.
1453
1454 The match_limit field provides a means of preventing PCRE from using up
1455 a vast amount of resources when running patterns that are not going to
1456 match, but which have a very large number of possibilities in their
1457 search trees. The classic example is the use of nested unlimited
1458 repeats.
1459
1460 Internally, PCRE uses a function called match() which it calls repeat-
1461 edly (sometimes recursively). The limit is imposed on the number of
1462 times this function is called during a match, which has the effect of
1463 limiting the amount of recursion and backtracking that can take place.
1464 For patterns that are not anchored, the count starts from zero for each
1465 position in the subject string.
1466
1467 The default limit for the library can be set when PCRE is built; the
1468 default default is 10 million, which handles all but the most extreme
1469 cases. You can reduce the default by suppling pcre_exec() with a
1470 pcre_extra block in which match_limit is set to a smaller value, and
1471 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
1472 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
1473
1474 The pcre_callout field is used in conjunction with the "callout" fea-
1475 ture, which is described in the pcrecallout documentation.
1476
1477 The tables field is used to pass a character tables pointer to
1478 pcre_exec(); this overrides the value that is stored with the compiled
1479 pattern. A non-NULL value is stored with the compiled pattern only if
1480 custom tables were supplied to pcre_compile() via its tableptr argu-
1481 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
1482 PCRE's internal tables to be used. This facility is helpful when re-
1483 using patterns that have been saved after compiling with an external
1484 set of tables, because the external tables might be at a different
1485 address when pcre_exec() is called. See the pcreprecompile documenta-
1486 tion for a discussion of saving compiled patterns for later use.
1487
1488 Option bits for pcre_exec()
1489
1490 The unused bits of the options argument for pcre_exec() must be zero.
1491 The only bits that may be set are PCRE_ANCHORED, PCRE_NOTBOL,
1492 PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK and PCRE_PARTIAL.
1493
1494 PCRE_ANCHORED
1495
1496 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
1497 matching position. If a pattern was compiled with PCRE_ANCHORED, or
1498 turned out to be anchored by virtue of its contents, it cannot be made
1499 unachored at matching time.
1500
1501 PCRE_NOTBOL
1502
1503 This option specifies that first character of the subject string is not
1504 the beginning of a line, so the circumflex metacharacter should not
1505 match before it. Setting this without PCRE_MULTILINE (at compile time)
1506 causes circumflex never to match. This option affects only the behav-
1507 iour of the circumflex metacharacter. It does not affect \A.
1508
1509 PCRE_NOTEOL
1510
1511 This option specifies that the end of the subject string is not the end
1512 of a line, so the dollar metacharacter should not match it nor (except
1513 in multiline mode) a newline immediately before it. Setting this with-
1514 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
1515 option affects only the behaviour of the dollar metacharacter. It does
1516 not affect \Z or \z.
1517
1518 PCRE_NOTEMPTY
1519
1520 An empty string is not considered to be a valid match if this option is
1521 set. If there are alternatives in the pattern, they are tried. If all
1522 the alternatives match the empty string, the entire match fails. For
1523 example, if the pattern
1524
1525 a?b?
1526
1527 is applied to a string not beginning with "a" or "b", it matches the
1528 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
1529 match is not valid, so PCRE searches further into the string for occur-
1530 rences of "a" or "b".
1531
1532 Perl has no direct equivalent of PCRE_NOTEMPTY, but it does make a spe-
1533 cial case of a pattern match of the empty string within its split()
1534 function, and when using the /g modifier. It is possible to emulate
1535 Perl's behaviour after matching a null string by first trying the match
1536 again at the same offset with PCRE_NOTEMPTY and PCRE_ANCHORED, and then
1537 if that fails by advancing the starting offset (see below) and trying
1538 an ordinary match again. There is some code that demonstrates how to do
1539 this in the pcredemo.c sample program.
1540
1541 PCRE_NO_UTF8_CHECK
1542
1543 When PCRE_UTF8 is set at compile time, the validity of the subject as a
1544 UTF-8 string is automatically checked when pcre_exec() is subsequently
1545 called. The value of startoffset is also checked to ensure that it
1546 points to the start of a UTF-8 character. If an invalid UTF-8 sequence
1547 of bytes is found, pcre_exec() returns the error PCRE_ERROR_BADUTF8. If
1548 startoffset contains an invalid value, PCRE_ERROR_BADUTF8_OFFSET is
1549 returned.
1550
1551 If you already know that your subject is valid, and you want to skip
1552 these checks for performance reasons, you can set the
1553 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
1554 do this for the second and subsequent calls to pcre_exec() if you are
1555 making repeated calls to find all the matches in a single subject
1556 string. However, you should be sure that the value of startoffset
1557 points to the start of a UTF-8 character. When PCRE_NO_UTF8_CHECK is
1558 set, the effect of passing an invalid UTF-8 string as a subject, or a
1559 value of startoffset that does not point to the start of a UTF-8 char-
1560 acter, is undefined. Your program may crash.
1561
1562 PCRE_PARTIAL
1563
1564 This option turns on the partial matching feature. If the subject
1565 string fails to match the pattern, but at some point during the match-
1566 ing process the end of the subject was reached (that is, the subject
1567 partially matches the pattern and the failure to match occurred only
1568 because there were not enough subject characters), pcre_exec() returns
1569 PCRE_ERROR_PARTIAL instead of PCRE_ERROR_NOMATCH. When PCRE_PARTIAL is
1570 used, there are restrictions on what may appear in the pattern. These
1571 are discussed in the pcrepartial documentation.
1572
1573 The string to be matched by pcre_exec()
1574
1575 The subject string is passed to pcre_exec() as a pointer in subject, a
1576 length in length, and a starting byte offset in startoffset. In UTF-8
1577 mode, the byte offset must point to the start of a UTF-8 character.
1578 Unlike the pattern string, the subject may contain binary zero bytes.
1579 When the starting offset is zero, the search for a match starts at the
1580 beginning of the subject, and this is by far the most common case.
1581
1582 A non-zero starting offset is useful when searching for another match
1583 in the same subject by calling pcre_exec() again after a previous suc-
1584 cess. Setting startoffset differs from just passing over a shortened
1585 string and setting PCRE_NOTBOL in the case of a pattern that begins
1586 with any kind of lookbehind. For example, consider the pattern
1587
1588 \Biss\B
1589
1590 which finds occurrences of "iss" in the middle of words. (\B matches
1591 only if the current position in the subject is not a word boundary.)
1592 When applied to the string "Mississipi" the first call to pcre_exec()
1593 finds the first occurrence. If pcre_exec() is called again with just
1594 the remainder of the subject, namely "issipi", it does not match,
1595 because \B is always false at the start of the subject, which is deemed
1596 to be a word boundary. However, if pcre_exec() is passed the entire
1597 string again, but with startoffset set to 4, it finds the second occur-
1598 rence of "iss" because it is able to look behind the starting point to
1599 discover that it is preceded by a letter.
1600
1601 If a non-zero starting offset is passed when the pattern is anchored,
1602 one attempt to match at the given offset is made. This can only succeed
1603 if the pattern does not require the match to be at the start of the
1604 subject.
1605
1606 How pcre_exec() returns captured substrings
1607
1608 In general, a pattern matches a certain portion of the subject, and in
1609 addition, further substrings from the subject may be picked out by
1610 parts of the pattern. Following the usage in Jeffrey Friedl's book,
1611 this is called "capturing" in what follows, and the phrase "capturing
1612 subpattern" is used for a fragment of a pattern that picks out a sub-
1613 string. PCRE supports several other kinds of parenthesized subpattern
1614 that do not cause substrings to be captured.
1615
1616 Captured substrings are returned to the caller via a vector of integer
1617 offsets whose address is passed in ovector. The number of elements in
1618 the vector is passed in ovecsize, which must be a non-negative number.
1619 Note: this argument is NOT the size of ovector in bytes.
1620
1621 The first two-thirds of the vector is used to pass back captured sub-
1622 strings, each substring using a pair of integers. The remaining third
1623 of the vector is used as workspace by pcre_exec() while matching cap-
1624 turing subpatterns, and is not available for passing back information.
1625 The length passed in ovecsize should always be a multiple of three. If
1626 it is not, it is rounded down.
1627
1628 When a match is successful, information about captured substrings is
1629 returned in pairs of integers, starting at the beginning of ovector,
1630 and continuing up to two-thirds of its length at the most. The first
1631 element of a pair is set to the offset of the first character in a sub-
1632 string, and the second is set to the offset of the first character
1633 after the end of a substring. The first pair, ovector[0] and ovec-
1634 tor[1], identify the portion of the subject string matched by the
1635 entire pattern. The next pair is used for the first capturing subpat-
1636 tern, and so on. The value returned by pcre_exec() is the number of
1637 pairs that have been set. If there are no capturing subpatterns, the
1638 return value from a successful match is 1, indicating that just the
1639 first pair of offsets has been set.
1640
1641 Some convenience functions are provided for extracting the captured
1642 substrings as separate strings. These are described in the following
1643 section.
1644
1645 It is possible for an capturing subpattern number n+1 to match some
1646 part of the subject when subpattern n has not been used at all. For
1647 example, if the string "abc" is matched against the pattern (a|(z))(bc)
1648 subpatterns 1 and 3 are matched, but 2 is not. When this happens, both
1649 offset values corresponding to the unused subpattern are set to -1.
1650
1651 If a capturing subpattern is matched repeatedly, it is the last portion
1652 of the string that it matched that is returned.
1653
1654 If the vector is too small to hold all the captured substring offsets,
1655 it is used as far as possible (up to two-thirds of its length), and the
1656 function returns a value of zero. In particular, if the substring off-
1657 sets are not of interest, pcre_exec() may be called with ovector passed
1658 as NULL and ovecsize as zero. However, if the pattern contains back
1659 references and the ovector is not big enough to remember the related
1660 substrings, PCRE has to get additional memory for use during matching.
1661 Thus it is usually advisable to supply an ovector.
1662
1663 Note that pcre_info() can be used to find out how many capturing sub-
1664 patterns there are in a compiled pattern. The smallest size for ovector
1665 that will allow for n captured substrings, in addition to the offsets
1666 of the substring matched by the whole pattern, is (n+1)*3.
1667
1668 Return values from pcre_exec()
1669
1670 If pcre_exec() fails, it returns a negative number. The following are
1671 defined in the header file:
1672
1673 PCRE_ERROR_NOMATCH (-1)
1674
1675 The subject string did not match the pattern.
1676
1677 PCRE_ERROR_NULL (-2)
1678
1679 Either code or subject was passed as NULL, or ovector was NULL and
1680 ovecsize was not zero.
1681
1682 PCRE_ERROR_BADOPTION (-3)
1683
1684 An unrecognized bit was set in the options argument.
1685
1686 PCRE_ERROR_BADMAGIC (-4)
1687
1688 PCRE stores a 4-byte "magic number" at the start of the compiled code,
1689 to catch the case when it is passed a junk pointer and to detect when a
1690 pattern that was compiled in an environment of one endianness is run in
1691 an environment with the other endianness. This is the error that PCRE
1692 gives when the magic number is not present.
1693
1694 PCRE_ERROR_UNKNOWN_NODE (-5)
1695
1696 While running the pattern match, an unknown item was encountered in the
1697 compiled pattern. This error could be caused by a bug in PCRE or by
1698 overwriting of the compiled pattern.
1699
1700 PCRE_ERROR_NOMEMORY (-6)
1701
1702 If a pattern contains back references, but the ovector that is passed
1703 to pcre_exec() is not big enough to remember the referenced substrings,
1704 PCRE gets a block of memory at the start of matching to use for this
1705 purpose. If the call via pcre_malloc() fails, this error is given. The
1706 memory is automatically freed at the end of matching.
1707
1708 PCRE_ERROR_NOSUBSTRING (-7)
1709
1710 This error is used by the pcre_copy_substring(), pcre_get_substring(),
1711 and pcre_get_substring_list() functions (see below). It is never
1712 returned by pcre_exec().
1713
1714 PCRE_ERROR_MATCHLIMIT (-8)
1715
1716 The recursion and backtracking limit, as specified by the match_limit
1717 field in a pcre_extra structure (or defaulted) was reached. See the
1718 description above.
1719
1720 PCRE_ERROR_CALLOUT (-9)
1721
1722 This error is never generated by pcre_exec() itself. It is provided for
1723 use by callout functions that want to yield a distinctive error code.
1724 See the pcrecallout documentation for details.
1725
1726 PCRE_ERROR_BADUTF8 (-10)
1727
1728 A string that contains an invalid UTF-8 byte sequence was passed as a
1729 subject.
1730
1731 PCRE_ERROR_BADUTF8_OFFSET (-11)
1732
1733 The UTF-8 byte sequence that was passed as a subject was valid, but the
1734 value of startoffset did not point to the beginning of a UTF-8 charac-
1735 ter.
1736
1737 PCRE_ERROR_PARTIAL (-12)
1738
1739 The subject string did not match, but it did match partially. See the
1740 pcrepartial documentation for details of partial matching.
1741
1742 PCRE_ERROR_BADPARTIAL (-13)
1743
1744 The PCRE_PARTIAL option was used with a compiled pattern containing
1745 items that are not supported for partial matching. See the pcrepartial
1746 documentation for details of partial matching.
1747
1748 PCRE_ERROR_INTERNAL (-14)
1749
1750 An unexpected internal error has occurred. This error could be caused
1751 by a bug in PCRE or by overwriting of the compiled pattern.
1752
1753 PCRE_ERROR_BADCOUNT (-15)
1754
1755 This error is given if the value of the ovecsize argument is negative.
1756
1757
1758 EXTRACTING CAPTURED SUBSTRINGS BY NUMBER
1759
1760 int pcre_copy_substring(const char *subject, int *ovector,
1761 int stringcount, int stringnumber, char *buffer,
1762 int buffersize);
1763
1764 int pcre_get_substring(const char *subject, int *ovector,
1765 int stringcount, int stringnumber,
1766 const char **stringptr);
1767
1768 int pcre_get_substring_list(const char *subject,
1769 int *ovector, int stringcount, const char ***listptr);
1770
1771 Captured substrings can be accessed directly by using the offsets
1772 returned by pcre_exec() in ovector. For convenience, the functions
1773 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
1774 string_list() are provided for extracting captured substrings as new,
1775 separate, zero-terminated strings. These functions identify substrings
1776 by number. The next section describes functions for extracting named
1777 substrings. A substring that contains a binary zero is correctly
1778 extracted and has a further zero added on the end, but the result is
1779 not, of course, a C string.
1780
1781 The first three arguments are the same for all three of these func-
1782 tions: subject is the subject string that has just been successfully
1783 matched, ovector is a pointer to the vector of integer offsets that was
1784 passed to pcre_exec(), and stringcount is the number of substrings that
1785 were captured by the match, including the substring that matched the
1786 entire regular expression. This is the value returned by pcre_exec() if
1787 it is greater than zero. If pcre_exec() returned zero, indicating that
1788 it ran out of space in ovector, the value passed as stringcount should
1789 be the number of elements in the vector divided by three.
1790
1791 The functions pcre_copy_substring() and pcre_get_substring() extract a
1792 single substring, whose number is given as stringnumber. A value of
1793 zero extracts the substring that matched the entire pattern, whereas
1794 higher values extract the captured substrings. For pcre_copy_sub-
1795 string(), the string is placed in buffer, whose length is given by
1796 buffersize, while for pcre_get_substring() a new block of memory is
1797 obtained via pcre_malloc, and its address is returned via stringptr.
1798 The yield of the function is the length of the string, not including
1799 the terminating zero, or one of
1800
1801 PCRE_ERROR_NOMEMORY (-6)
1802
1803 The buffer was too small for pcre_copy_substring(), or the attempt to
1804 get memory failed for pcre_get_substring().
1805
1806 PCRE_ERROR_NOSUBSTRING (-7)
1807
1808 There is no substring whose number is stringnumber.
1809
1810 The pcre_get_substring_list() function extracts all available sub-
1811 strings and builds a list of pointers to them. All this is done in a
1812 single block of memory that is obtained via pcre_malloc. The address of
1813 the memory block is returned via listptr, which is also the start of
1814 the list of string pointers. The end of the list is marked by a NULL
1815 pointer. The yield of the function is zero if all went well, or
1816
1817 PCRE_ERROR_NOMEMORY (-6)
1818
1819 if the attempt to get the memory block failed.
1820
1821 When any of these functions encounter a substring that is unset, which
1822 can happen when capturing subpattern number n+1 matches some part of
1823 the subject, but subpattern n has not been used at all, they return an
1824 empty string. This can be distinguished from a genuine zero-length sub-
1825 string by inspecting the appropriate offset in ovector, which is nega-
1826 tive for unset substrings.
1827
1828 The two convenience functions pcre_free_substring() and pcre_free_sub-
1829 string_list() can be used to free the memory returned by a previous
1830 call of pcre_get_substring() or pcre_get_substring_list(), respec-
1831 tively. They do nothing more than call the function pointed to by
1832 pcre_free, which of course could be called directly from a C program.
1833 However, PCRE is used in some situations where it is linked via a spe-
1834 cial interface to another programming language which cannot use
1835 pcre_free directly; it is for these cases that the functions are pro-
1836 vided.
1837
1838
1839 EXTRACTING CAPTURED SUBSTRINGS BY NAME
1840
1841 int pcre_get_stringnumber(const pcre *code,
1842 const char *name);
1843
1844 int pcre_copy_named_substring(const pcre *code,
1845 const char *subject, int *ovector,
1846 int stringcount, const char *stringname,
1847 char *buffer, int buffersize);
1848
1849 int pcre_get_named_substring(const pcre *code,
1850 const char *subject, int *ovector,
1851 int stringcount, const char *stringname,
1852 const char **stringptr);
1853
1854 To extract a substring by name, you first have to find associated num-
1855 ber. For example, for this pattern
1856
1857 (a+)b(?P<xxx>\d+)...
1858
1859 the number of the subpattern called "xxx" is 2. You can find the number
1860 from the name by calling pcre_get_stringnumber(). The first argument is
1861 the compiled pattern, and the second is the name. The yield of the
1862 function is the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if
1863 there is no subpattern of that name.
1864
1865 Given the number, you can extract the substring directly, or use one of
1866 the functions described in the previous section. For convenience, there
1867 are also two functions that do the whole job.
1868
1869 Most of the arguments of pcre_copy_named_substring() and
1870 pcre_get_named_substring() are the same as those for the similarly
1871 named functions that extract by number. As these are described in the
1872 previous section, they are not re-described here. There are just two
1873 differences:
1874
1875 First, instead of a substring number, a substring name is given. Sec-
1876 ond, there is an extra argument, given at the start, which is a pointer
1877 to the compiled pattern. This is needed in order to gain access to the
1878 name-to-number translation table.
1879
1880 These functions call pcre_get_stringnumber(), and if it succeeds, they
1881 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
1882 ate.
1883
1884
1885 FINDING ALL POSSIBLE MATCHES
1886
1887 The traditional matching function uses a similar algorithm to Perl,
1888 which stops when it finds the first match, starting at a given point in
1889 the subject. If you want to find all possible matches, or the longest
1890 possible match, consider using the alternative matching function (see
1891 below) instead. If you cannot use the alternative function, but still
1892 need to find all possible matches, you can kludge it up by making use
1893 of the callout facility, which is described in the pcrecallout documen-
1894 tation.
1895
1896 What you have to do is to insert a callout right at the end of the pat-
1897 tern. When your callout function is called, extract and save the cur-
1898 rent matched substring. Then return 1, which forces pcre_exec() to
1899 backtrack and try other alternatives. Ultimately, when it runs out of
1900 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
1901
1902
1903 MATCHING A PATTERN: THE ALTERNATIVE FUNCTION
1904
1905 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
1906 const char *subject, int length, int startoffset,
1907 int options, int *ovector, int ovecsize,
1908 int *workspace, int wscount);
1909
1910 The function pcre_dfa_exec() is called to match a subject string
1911 against a compiled pattern, using a "DFA" matching algorithm. This has
1912 different characteristics to the normal algorithm, and is not compati-
1913 ble with Perl. Some of the features of PCRE patterns are not supported.
1914 Nevertheless, there are times when this kind of matching can be useful.
1915 For a discussion of the two matching algorithms, see the pcrematching
1916 documentation.
1917
1918 The arguments for the pcre_dfa_exec() function are the same as for
1919 pcre_exec(), plus two extras. The ovector argument is used in a differ-
1920 ent way, and this is described below. The other common arguments are
1921 used in the same way as for pcre_exec(), so their description is not
1922 repeated here.
1923
1924 The two additional arguments provide workspace for the function. The
1925 workspace vector should contain at least 20 elements. It is used for
1926 keeping track of multiple paths through the pattern tree. More
1927 workspace will be needed for patterns and subjects where there are a
1928 lot of possible matches.
1929
1930 Here is an example of a simple call to pcre_exec():
1931
1932 int rc;
1933 int ovector[10];
1934 int wspace[20];
1935 rc = pcre_exec(
1936 re, /* result of pcre_compile() */
1937 NULL, /* we didn't study the pattern */
1938 "some string", /* the subject string */
1939 11, /* the length of the subject string */
1940 0, /* start at offset 0 in the subject */
1941 0, /* default options */
1942 ovector, /* vector of integers for substring information */
1943 10, /* number of elements (NOT size in bytes) */
1944 wspace, /* working space vector */
1945 20); /* number of elements (NOT size in bytes) */
1946
1947 Option bits for pcre_dfa_exec()
1948
1949 The unused bits of the options argument for pcre_dfa_exec() must be
1950 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NOTBOL,
1951 PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK, PCRE_PARTIAL,
1952 PCRE_DFA_SHORTEST, and PCRE_DFA_RESTART. All but the last three of
1953 these are the same as for pcre_exec(), so their description is not
1954 repeated here.
1955
1956 PCRE_PARTIAL
1957
1958 This has the same general effect as it does for pcre_exec(), but the
1959 details are slightly different. When PCRE_PARTIAL is set for
1960 pcre_dfa_exec(), the return code PCRE_ERROR_NOMATCH is converted into
1961 PCRE_ERROR_PARTIAL if the end of the subject is reached, there have
1962 been no complete matches, but there is still at least one matching pos-
1963 sibility. The portion of the string that provided the partial match is
1964 set as the first matching string.
1965
1966 PCRE_DFA_SHORTEST
1967
1968 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
1969 stop as soon as it has found one match. Because of the way the DFA
1970 algorithm works, this is necessarily the shortest possible match at the
1971 first possible matching point in the subject string.
1972
1973 PCRE_DFA_RESTART
1974
1975 When pcre_dfa_exec() is called with the PCRE_PARTIAL option, and
1976 returns a partial match, it is possible to call it again, with addi-
1977 tional subject characters, and have it continue with the same match.
1978 The PCRE_DFA_RESTART option requests this action; when it is set, the
1979 workspace and wscount options must reference the same vector as before
1980 because data about the match so far is left in them after a partial
1981 match. There is more discussion of this facility in the pcrepartial
1982 documentation.
1983
1984 Successful returns from pcre_dfa_exec()
1985
1986 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
1987 string in the subject. Note, however, that all the matches from one run
1988 of the function start at the same point in the subject. The shorter
1989 matches are all initial substrings of the longer matches. For example,
1990 if the pattern
1991
1992 <.*>
1993
1994 is matched against the string
1995
1996 This is <something> <something else> <something further> no more
1997
1998 the three matched strings are
1999
2000 <something>
2001 <something> <something else>
2002 <something> <something else> <something further>
2003
2004 On success, the yield of the function is a number greater than zero,
2005 which is the number of matched substrings. The substrings themselves
2006 are returned in ovector. Each string uses two elements; the first is
2007 the offset to the start, and the second is the offset to the end. All
2008 the strings have the same start offset. (Space could have been saved by
2009 giving this only once, but it was decided to retain some compatibility
2010 with the way pcre_exec() returns data, even though the meaning of the
2011 strings is different.)
2012
2013 The strings are returned in reverse order of length; that is, the long-
2014 est matching string is given first. If there were too many matches to
2015 fit into ovector, the yield of the function is zero, and the vector is
2016 filled with the longest matches.
2017
2018 Error returns from pcre_dfa_exec()
2019
2020 The pcre_dfa_exec() function returns a negative number when it fails.
2021 Many of the errors are the same as for pcre_exec(), and these are
2022 described above. There are in addition the following errors that are
2023 specific to pcre_dfa_exec():
2024
2025 PCRE_ERROR_DFA_UITEM (-16)
2026
2027 This return is given if pcre_dfa_exec() encounters an item in the pat-
2028 tern that it does not support, for instance, the use of \C or a back
2029 reference.
2030
2031 PCRE_ERROR_DFA_UCOND (-17)
2032
2033 This return is given if pcre_dfa_exec() encounters a condition item in
2034 a pattern that uses a back reference for the condition. This is not
2035 supported.
2036
2037 PCRE_ERROR_DFA_UMLIMIT (-18)
2038
2039 This return is given if pcre_dfa_exec() is called with an extra block
2040 that contains a setting of the match_limit field. This is not supported
2041 (it is meaningless).
2042
2043 PCRE_ERROR_DFA_WSSIZE (-19)
2044
2045 This return is given if pcre_dfa_exec() runs out of space in the
2046 workspace vector.
2047
2048 PCRE_ERROR_DFA_RECURSE (-20)
2049
2050 When a recursive subpattern is processed, the matching function calls
2051 itself recursively, using private vectors for ovector and workspace.
2052 This error is given if the output vector is not large enough. This
2053 should be extremely rare, as a vector of size 1000 is used.
2054
2055 Last updated: 16 May 2005
2056 Copyright (c) 1997-2005 University of Cambridge.
2057 ------------------------------------------------------------------------------
2058
2059
2060 PCRECALLOUT(3) PCRECALLOUT(3)
2061
2062
2063 NAME
2064 PCRE - Perl-compatible regular expressions
2065
2066
2067 PCRE CALLOUTS
2068
2069 int (*pcre_callout)(pcre_callout_block *);
2070
2071 PCRE provides a feature called "callout", which is a means of temporar-
2072 ily passing control to the caller of PCRE in the middle of pattern
2073 matching. The caller of PCRE provides an external function by putting
2074 its entry point in the global variable pcre_callout. By default, this
2075 variable contains NULL, which disables all calling out.
2076
2077 Within a regular expression, (?C) indicates the points at which the
2078 external function is to be called. Different callout points can be
2079 identified by putting a number less than 256 after the letter C. The
2080 default value is zero. For example, this pattern has two callout
2081 points:
2082
2083 (?C1)eabc(?C2)def
2084
2085 If the PCRE_AUTO_CALLOUT option bit is set when pcre_compile() is
2086 called, PCRE automatically inserts callouts, all with number 255,
2087 before each item in the pattern. For example, if PCRE_AUTO_CALLOUT is
2088 used with the pattern
2089
2090 A(\d{2}|--)
2091
2092 it is processed as if it were
2093
2094 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
2095
2096 Notice that there is a callout before and after each parenthesis and
2097 alternation bar. Automatic callouts can be used for tracking the
2098 progress of pattern matching. The pcretest command has an option that
2099 sets automatic callouts; when it is used, the output indicates how the
2100 pattern is matched. This is useful information when you are trying to
2101 optimize the performance of a particular pattern.
2102
2103
2104 MISSING CALLOUTS
2105
2106 You should be aware that, because of optimizations in the way PCRE
2107 matches patterns, callouts sometimes do not happen. For example, if the
2108 pattern is
2109
2110 ab(?C4)cd
2111
2112 PCRE knows that any matching string must contain the letter "d". If the
2113 subject string is "abyz", the lack of "d" means that matching doesn't
2114 ever start, and the callout is never reached. However, with "abyd",
2115 though the result is still no match, the callout is obeyed.
2116
2117
2118 THE CALLOUT INTERFACE
2119
2120 During matching, when PCRE reaches a callout point, the external func-
2121 tion defined by pcre_callout is called (if it is set). This applies to
2122 both the pcre_exec() and the pcre_dfa_exec() matching functions. The
2123 only argument to the callout function is a pointer to a pcre_callout
2124 block. This structure contains the following fields:
2125
2126 int version;
2127 int callout_number;
2128 int *offset_vector;
2129 const char *subject;
2130 int subject_length;
2131 int start_match;
2132 int current_position;
2133 int capture_top;
2134 int capture_last;
2135 void *callout_data;
2136 int pattern_position;
2137 int next_item_length;
2138
2139 The version field is an integer containing the version number of the
2140 block format. The initial version was 0; the current version is 1. The
2141 version number will change again in future if additional fields are
2142 added, but the intention is never to remove any of the existing fields.
2143
2144 The callout_number field contains the number of the callout, as com-
2145 piled into the pattern (that is, the number after ?C for manual call-
2146 outs, and 255 for automatically generated callouts).
2147
2148 The offset_vector field is a pointer to the vector of offsets that was
2149 passed by the caller to pcre_exec() or pcre_dfa_exec(). When
2150 pcre_exec() is used, the contents can be inspected in order to extract
2151 substrings that have been matched so far, in the same way as for
2152 extracting substrings after a match has completed. For pcre_dfa_exec()
2153 this field is not useful.
2154
2155 The subject and subject_length fields contain copies of the values that
2156 were passed to pcre_exec().
2157
2158 The start_match field contains the offset within the subject at which
2159 the current match attempt started. If the pattern is not anchored, the
2160 callout function may be called several times from the same point in the
2161 pattern for different starting points in the subject.
2162
2163 The current_position field contains the offset within the subject of
2164 the current match pointer.
2165
2166 When the pcre_exec() function is used, the capture_top field contains
2167 one more than the number of the highest numbered captured substring so
2168 far. If no substrings have been captured, the value of capture_top is
2169 one. This is always the case when pcre_dfa_exec() is used, because it
2170 does not support captured substrings.
2171
2172 The capture_last field contains the number of the most recently cap-
2173 tured substring. If no substrings have been captured, its value is -1.
2174 This is always the case when pcre_dfa_exec() is used.
2175
2176 The callout_data field contains a value that is passed to pcre_exec()
2177 or pcre_dfa_exec() specifically so that it can be passed back in call-
2178 outs. It is passed in the pcre_callout field of the pcre_extra data
2179 structure. If no such data was passed, the value of callout_data in a
2180 pcre_callout block is NULL. There is a description of the pcre_extra
2181 structure in the pcreapi documentation.
2182
2183 The pattern_position field is present from version 1 of the pcre_call-
2184 out structure. It contains the offset to the next item to be matched in
2185 the pattern string.
2186
2187 The next_item_length field is present from version 1 of the pcre_call-
2188 out structure. It contains the length of the next item to be matched in
2189 the pattern string. When the callout immediately precedes an alterna-
2190 tion bar, a closing parenthesis, or the end of the pattern, the length
2191 is zero. When the callout precedes an opening parenthesis, the length
2192 is that of the entire subpattern.
2193
2194 The pattern_position and next_item_length fields are intended to help
2195 in distinguishing between different automatic callouts, which all have
2196 the same callout number. However, they are set for all callouts.
2197
2198
2199 RETURN VALUES
2200
2201 The external callout function returns an integer to PCRE. If the value
2202 is zero, matching proceeds as normal. If the value is greater than
2203 zero, matching fails at the current point, but the testing of other
2204 matching possibilities goes ahead, just as if a lookahead assertion had
2205 failed. If the value is less than zero, the match is abandoned, and
2206 pcre_exec() (or pcre_dfa_exec()) returns the negative value.
2207
2208 Negative values should normally be chosen from the set of
2209 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
2210 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
2211 reserved for use by callout functions; it will never be used by PCRE
2212 itself.
2213
2214 Last updated: 28 February 2005
2215 Copyright (c) 1997-2005 University of Cambridge.
2216 ------------------------------------------------------------------------------
2217
2218
2219 PCRECOMPAT(3) PCRECOMPAT(3)
2220
2221
2222 NAME
2223 PCRE - Perl-compatible regular expressions
2224
2225
2226 DIFFERENCES BETWEEN PCRE AND PERL
2227
2228 This document describes the differences in the ways that PCRE and Perl
2229 handle regular expressions. The differences described here are with
2230 respect to Perl 5.8.
2231
2232 1. PCRE does not have full UTF-8 support. Details of what it does have
2233 are given in the section on UTF-8 support in the main pcre page.
2234
2235 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl
2236 permits them, but they do not mean what you might think. For example,
2237 (?!a){3} does not assert that the next three characters are not "a". It
2238 just asserts that the next character is not "a" three times.
2239
2240 3. Capturing subpatterns that occur inside negative lookahead asser-
2241 tions are counted, but their entries in the offsets vector are never
2242 set. Perl sets its numerical variables from any such patterns that are
2243 matched before the assertion fails to match something (thereby succeed-
2244 ing), but only if the negative lookahead assertion contains just one
2245 branch.
2246
2247 4. Though binary zero characters are supported in the subject string,
2248 they are not allowed in a pattern string because it is passed as a nor-
2249 mal C string, terminated by zero. The escape sequence \0 can be used in
2250 the pattern to represent a binary zero.
2251
2252 5. The following Perl escape sequences are not supported: \l, \u, \L,
2253 \U, and \N. In fact these are implemented by Perl's general string-han-
2254 dling and are not part of its pattern matching engine. If any of these
2255 are encountered by PCRE, an error is generated.
2256
2257 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
2258 is built with Unicode character property support. The properties that
2259 can be tested with \p and \P are limited to the general category prop-
2260 erties such as Lu and Nd.
2261
2262 7. PCRE does support the \Q...\E escape for quoting substrings. Charac-
2263 ters in between are treated as literals. This is slightly different
2264 from Perl in that $ and @ are also handled as literals inside the
2265 quotes. In Perl, they cause variable interpolation (but of course PCRE
2266 does not have variables). Note the following examples:
2267
2268 Pattern PCRE matches Perl matches
2269
2270 \Qabc$xyz\E abc$xyz abc followed by the
2271 contents of $xyz
2272 \Qabc\$xyz\E abc\$xyz abc\$xyz
2273 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2274
2275 The \Q...\E sequence is recognized both inside and outside character
2276 classes.
2277
2278 8. Fairly obviously, PCRE does not support the (?{code}) and (?p{code})
2279 constructions. However, there is support for recursive patterns using
2280 the non-Perl items (?R), (?number), and (?P>name). Also, the PCRE
2281 "callout" feature allows an external function to be called during pat-
2282 tern matching. See the pcrecallout documentation for details.
2283
2284 9. There are some differences that are concerned with the settings of
2285 captured strings when part of a pattern is repeated. For example,
2286 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
2287 unset, but in PCRE it is set to "b".
2288
2289 10. PCRE provides some extensions to the Perl regular expression facil-
2290 ities:
2291
2292 (a) Although lookbehind assertions must match fixed length strings,
2293 each alternative branch of a lookbehind assertion can match a different
2294 length of string. Perl requires them all to have the same length.
2295
2296 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
2297 meta-character matches only at the very end of the string.
2298
2299 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
2300 cial meaning is faulted.
2301
2302 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
2303 fiers is inverted, that is, by default they are not greedy, but if fol-
2304 lowed by a question mark they are.
2305
2306 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
2307 tried only at the first matching position in the subject string.
2308
2309 (f) The PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, and PCRE_NO_AUTO_CAP-
2310 TURE options for pcre_exec() have no Perl equivalents.
2311
2312 (g) The (?R), (?number), and (?P>name) constructs allows for recursive
2313 pattern matching (Perl can do this using the (?p{code}) construct,
2314 which PCRE cannot support.)
2315
2316 (h) PCRE supports named capturing substrings, using the Python syntax.
2317
2318 (i) PCRE supports the possessive quantifier "++" syntax, taken from
2319 Sun's Java package.
2320
2321 (j) The (R) condition, for testing recursion, is a PCRE extension.
2322
2323 (k) The callout facility is PCRE-specific.
2324
2325 (l) The partial matching facility is PCRE-specific.
2326
2327 (m) Patterns compiled by PCRE can be saved and re-used at a later time,
2328 even on different hosts that have the other endianness.
2329
2330 (n) The alternative matching function (pcre_dfa_exec()) matches in a
2331 different way and is not Perl-compatible.
2332
2333 Last updated: 28 February 2005
2334 Copyright (c) 1997-2005 University of Cambridge.
2335 ------------------------------------------------------------------------------
2336
2337
2338 PCREPATTERN(3) PCREPATTERN(3)
2339
2340
2341 NAME
2342 PCRE - Perl-compatible regular expressions
2343
2344
2345 PCRE REGULAR EXPRESSION DETAILS
2346
2347 The syntax and semantics of the regular expressions supported by PCRE
2348 are described below. Regular expressions are also described in the Perl
2349 documentation and in a number of books, some of which have copious
2350 examples. Jeffrey Friedl's "Mastering Regular Expressions", published
2351 by O'Reilly, covers regular expressions in great detail. This descrip-
2352 tion of PCRE's regular expressions is intended as reference material.
2353
2354 The original operation of PCRE was on strings of one-byte characters.
2355 However, there is now also support for UTF-8 character strings. To use
2356 this, you must build PCRE to include UTF-8 support, and then call
2357 pcre_compile() with the PCRE_UTF8 option. How this affects pattern
2358 matching is mentioned in several places below. There is also a summary
2359 of UTF-8 features in the section on UTF-8 support in the main pcre
2360 page.
2361
2362 The remainder of this document discusses the patterns that are sup-
2363 ported by PCRE when its main matching function, pcre_exec(), is used.
2364 From release 6.0, PCRE offers a second matching function,
2365 pcre_dfa_exec(), which matches using a different algorithm that is not
2366 Perl-compatible. The advantages and disadvantages of the alternative
2367 function, and how it differs from the normal function, are discussed in
2368 the pcrematching page.
2369
2370 A regular expression is a pattern that is matched against a subject
2371 string from left to right. Most characters stand for themselves in a
2372 pattern, and match the corresponding characters in the subject. As a
2373 trivial example, the pattern
2374
2375 The quick brown fox
2376
2377 matches a portion of a subject string that is identical to itself. When
2378 caseless matching is specified (the PCRE_CASELESS option), letters are
2379 matched independently of case. In UTF-8 mode, PCRE always understands
2380 the concept of case for characters whose values are less than 128, so
2381 caseless matching is always possible. For characters with higher val-
2382 ues, the concept of case is supported if PCRE is compiled with Unicode
2383 property support, but not otherwise. If you want to use caseless
2384 matching for characters 128 and above, you must ensure that PCRE is
2385 compiled with Unicode property support as well as with UTF-8 support.
2386
2387 The power of regular expressions comes from the ability to include
2388 alternatives and repetitions in the pattern. These are encoded in the
2389 pattern by the use of metacharacters, which do not stand for themselves
2390 but instead are interpreted in some special way.
2391
2392 There are two different sets of metacharacters: those that are recog-
2393 nized anywhere in the pattern except within square brackets, and those
2394 that are recognized in square brackets. Outside square brackets, the
2395 metacharacters are as follows:
2396
2397 \ general escape character with several uses
2398 ^ assert start of string (or line, in multiline mode)
2399 $ assert end of string (or line, in multiline mode)
2400 . match any character except newline (by default)
2401 [ start character class definition
2402 | start of alternative branch
2403 ( start subpattern
2404 ) end subpattern
2405 ? extends the meaning of (
2406 also 0 or 1 quantifier
2407 also quantifier minimizer
2408 * 0 or more quantifier
2409 + 1 or more quantifier
2410 also "possessive quantifier"
2411 { start min/max quantifier
2412
2413 Part of a pattern that is in square brackets is called a "character
2414 class". In a character class the only metacharacters are:
2415
2416 \ general escape character
2417 ^ negate the class, but only if the first character
2418 - indicates character range
2419 [ POSIX character class (only if followed by POSIX
2420 syntax)
2421 ] terminates the character class
2422
2423 The following sections describe the use of each of the metacharacters.
2424
2425
2426 BACKSLASH
2427
2428 The backslash character has several uses. Firstly, if it is followed by
2429 a non-alphanumeric character, it takes away any special meaning that
2430 character may have. This use of backslash as an escape character
2431 applies both inside and outside character classes.
2432
2433 For example, if you want to match a * character, you write \* in the
2434 pattern. This escaping action applies whether or not the following
2435 character would otherwise be interpreted as a metacharacter, so it is
2436 always safe to precede a non-alphanumeric with backslash to specify
2437 that it stands for itself. In particular, if you want to match a back-
2438 slash, you write \\.
2439
2440 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
2441 the pattern (other than in a character class) and characters between a
2442 # outside a character class and the next newline character are ignored.
2443 An escaping backslash can be used to include a whitespace or # charac-
2444 ter as part of the pattern.
2445
2446 If you want to remove the special meaning from a sequence of charac-
2447 ters, you can do so by putting them between \Q and \E. This is differ-
2448 ent from Perl in that $ and @ are handled as literals in \Q...\E
2449 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
2450 tion. Note the following examples:
2451
2452 Pattern PCRE matches Perl matches
2453
2454 \Qabc$xyz\E abc$xyz abc followed by the
2455 contents of $xyz
2456 \Qabc\$xyz\E abc\$xyz abc\$xyz
2457 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2458
2459 The \Q...\E sequence is recognized both inside and outside character
2460 classes.
2461
2462 Non-printing characters
2463
2464 A second use of backslash provides a way of encoding non-printing char-
2465 acters in patterns in a visible manner. There is no restriction on the
2466 appearance of non-printing characters, apart from the binary zero that
2467 terminates a pattern, but when a pattern is being prepared by text
2468 editing, it is usually easier to use one of the following escape
2469 sequences than the binary character it represents:
2470
2471 \a alarm, that is, the BEL character (hex 07)
2472 \cx "control-x", where x is any character
2473 \e escape (hex 1B)
2474 \f formfeed (hex 0C)
2475 \n newline (hex 0A)
2476 \r carriage return (hex 0D)
2477 \t tab (hex 09)
2478 \ddd character with octal code ddd, or backreference
2479 \xhh character with hex code hh
2480 \x{hhh..} character with hex code hhh... (UTF-8 mode only)
2481
2482 The precise effect of \cx is as follows: if x is a lower case letter,
2483 it is converted to upper case. Then bit 6 of the character (hex 40) is
2484 inverted. Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c;
2485 becomes hex 7B.
2486
2487 After \x, from zero to two hexadecimal digits are read (letters can be
2488 in upper or lower case). In UTF-8 mode, any number of hexadecimal dig-
2489 its may appear between \x{ and }, but the value of the character code
2490 must be less than 2**31 (that is, the maximum hexadecimal value is
2491 7FFFFFFF). If characters other than hexadecimal digits appear between
2492 \x{ and }, or if there is no terminating }, this form of escape is not
2493 recognized. Instead, the initial \x will be interpreted as a basic
2494 hexadecimal escape, with no following digits, giving a character whose
2495 value is zero.
2496
2497 Characters whose value is less than 256 can be defined by either of the
2498 two syntaxes for \x when PCRE is in UTF-8 mode. There is no difference
2499 in the way they are handled. For example, \xdc is exactly the same as
2500 \x{dc}.
2501
2502 After \0 up to two further octal digits are read. In both cases, if
2503 there are fewer than two digits, just those that are present are used.
2504 Thus the sequence \0\x\07 specifies two binary zeros followed by a BEL
2505 character (code value 7). Make sure you supply two digits after the
2506 initial zero if the pattern character that follows is itself an octal
2507 digit.
2508
2509 The handling of a backslash followed by a digit other than 0 is compli-
2510 cated. Outside a character class, PCRE reads it and any following dig-
2511 its as a decimal number. If the number is less than 10, or if there
2512 have been at least that many previous capturing left parentheses in the
2513 expression, the entire sequence is taken as a back reference. A
2514 description of how this works is given later, following the discussion
2515 of parenthesized subpatterns.
2516
2517 Inside a character class, or if the decimal number is greater than 9
2518 and there have not been that many capturing subpatterns, PCRE re-reads
2519 up to three octal digits following the backslash, and generates a sin-
2520 gle byte from the least significant 8 bits of the value. Any subsequent
2521 digits stand for themselves. For example:
2522
2523 \040 is another way of writing a space
2524 \40 is the same, provided there are fewer than 40
2525 previous capturing subpatterns
2526 \7 is always a back reference
2527 \11 might be a back reference, or another way of
2528 writing a tab
2529 \011 is always a tab
2530 \0113 is a tab followed by the character "3"
2531 \113 might be a back reference, otherwise the
2532 character with octal code 113
2533 \377 might be a back reference, otherwise
2534 the byte consisting entirely of 1 bits
2535 \81 is either a back reference, or a binary zero
2536 followed by the two characters "8" and "1"
2537
2538 Note that octal values of 100 or greater must not be introduced by a
2539 leading zero, because no more than three octal digits are ever read.
2540
2541 All the sequences that define a single byte value or a single UTF-8
2542 character (in UTF-8 mode) can be used both inside and outside character
2543 classes. In addition, inside a character class, the sequence \b is
2544 interpreted as the backspace character (hex 08), and the sequence \X is
2545 interpreted as the character "X". Outside a character class, these
2546 sequences have different meanings (see below).
2547
2548 Generic character types
2549
2550 The third use of backslash is for specifying generic character types.
2551 The following are always recognized:
2552
2553 \d any decimal digit
2554 \D any character that is not a decimal digit
2555 \s any whitespace character
2556 \S any character that is not a whitespace character
2557 \w any "word" character
2558 \W any "non-word" character
2559
2560 Each pair of escape sequences partitions the complete set of characters
2561 into two disjoint sets. Any given character matches one, and only one,
2562 of each pair.
2563
2564 These character type sequences can appear both inside and outside char-
2565 acter classes. They each match one character of the appropriate type.
2566 If the current matching point is at the end of the subject string, all
2567 of them fail, since there is no character to match.
2568
2569 For compatibility with Perl, \s does not match the VT character (code
2570 11). This makes it different from the the POSIX "space" class. The \s
2571 characters are HT (9), LF (10), FF (12), CR (13), and space (32).
2572
2573 A "word" character is an underscore or any character less than 256 that
2574 is a letter or digit. The definition of letters and digits is con-
2575 trolled by PCRE's low-valued character tables, and may vary if locale-
2576 specific matching is taking place (see "Locale support" in the pcreapi
2577 page). For example, in the "fr_FR" (French) locale, some character
2578 codes greater than 128 are used for accented letters, and these are
2579 matched by \w.
2580
2581 In UTF-8 mode, characters with values greater than 128 never match \d,
2582 \s, or \w, and always match \D, \S, and \W. This is true even when Uni-
2583 code character property support is available.
2584
2585 Unicode character properties
2586
2587 When PCRE is built with Unicode character property support, three addi-
2588 tional escape sequences to match generic character types are available
2589 when UTF-8 mode is selected. They are:
2590
2591 \p{xx} a character with the xx property
2592 \P{xx} a character without the xx property
2593 \X an extended Unicode sequence
2594
2595 The property names represented by xx above are limited to the Unicode
2596 general category properties. Each character has exactly one such prop-
2597 erty, specified by a two-letter abbreviation. For compatibility with
2598 Perl, negation can be specified by including a circumflex between the
2599 opening brace and the property name. For example, \p{^Lu} is the same
2600 as \P{Lu}.
2601
2602 If only one letter is specified with \p or \P, it includes all the
2603 properties that start with that letter. In this case, in the absence of
2604 negation, the curly brackets in the escape sequence are optional; these
2605 two examples have the same effect:
2606
2607 \p{L}
2608 \pL
2609
2610 The following property codes are supported:
2611
2612 C Other
2613 Cc Control
2614 Cf Format
2615 Cn Unassigned
2616 Co Private use
2617 Cs Surrogate
2618
2619 L Letter
2620 Ll Lower case letter
2621 Lm Modifier letter
2622 Lo Other letter
2623 Lt Title case letter
2624 Lu Upper case letter
2625
2626 M Mark
2627 Mc Spacing mark
2628 Me Enclosing mark
2629 Mn Non-spacing mark
2630
2631 N Number
2632 Nd Decimal number
2633 Nl Letter number
2634 No Other number
2635
2636 P Punctuation
2637 Pc Connector punctuation
2638 Pd Dash punctuation
2639 Pe Close punctuation
2640 Pf Final punctuation
2641 Pi Initial punctuation
2642 Po Other punctuation
2643 Ps Open punctuation
2644
2645 S Symbol
2646 Sc Currency symbol
2647 Sk Modifier symbol
2648 Sm Mathematical symbol
2649 So Other symbol
2650
2651 Z Separator
2652 Zl Line separator
2653 Zp Paragraph separator
2654 Zs Space separator
2655
2656 Extended properties such as "Greek" or "InMusicalSymbols" are not sup-
2657 ported by PCRE.
2658
2659 Specifying caseless matching does not affect these escape sequences.
2660 For example, \p{Lu} always matches only upper case letters.
2661
2662 The \X escape matches any number of Unicode characters that form an
2663 extended Unicode sequence. \X is equivalent to
2664
2665 (?>\PM\pM*)
2666
2667 That is, it matches a character without the "mark" property, followed
2668 by zero or more characters with the "mark" property, and treats the
2669 sequence as an atomic group (see below). Characters with the "mark"
2670 property are typically accents that affect the preceding character.
2671
2672 Matching characters by Unicode property is not fast, because PCRE has
2673 to search a structure that contains data for over fifteen thousand
2674 characters. That is why the traditional escape sequences such as \d and
2675 \w do not use Unicode properties in PCRE.
2676
2677 Simple assertions
2678
2679 The fourth use of backslash is for certain simple assertions. An asser-
2680 tion specifies a condition that has to be met at a particular point in
2681 a match, without consuming any characters from the subject string. The
2682 use of subpatterns for more complicated assertions is described below.
2683 The backslashed assertions are:
2684
2685 \b matches at a word boundary
2686 \B matches when not at a word boundary
2687 \A matches at start of subject
2688 \Z matches at end of subject or before newline at end
2689 \z matches at end of subject
2690 \G matches at first matching position in subject
2691
2692 These assertions may not appear in character classes (but note that \b
2693 has a different meaning, namely the backspace character, inside a char-
2694 acter class).
2695
2696 A word boundary is a position in the subject string where the current
2697 character and the previous character do not both match \w or \W (i.e.
2698 one matches \w and the other matches \W), or the start or end of the
2699 string if the first or last character matches \w, respectively.
2700
2701 The \A, \Z, and \z assertions differ from the traditional circumflex
2702 and dollar (described in the next section) in that they only ever match
2703 at the very start and end of the subject string, whatever options are
2704 set. Thus, they are independent of multiline mode. These three asser-
2705 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
2706 affect only the behaviour of the circumflex and dollar metacharacters.
2707 However, if the startoffset argument of pcre_exec() is non-zero, indi-
2708 cating that matching is to start at a point other than the beginning of
2709 the subject, \A can never match. The difference between \Z and \z is
2710 that \Z matches before a newline that is the last character of the
2711 string as well as at the end of the string, whereas \z matches only at
2712 the end.
2713
2714 The \G assertion is true only when the current matching position is at
2715 the start point of the match, as specified by the startoffset argument
2716 of pcre_exec(). It differs from \A when the value of startoffset is
2717 non-zero. By calling pcre_exec() multiple times with appropriate argu-
2718 ments, you can mimic Perl's /g option, and it is in this kind of imple-
2719 mentation where \G can be useful.
2720
2721 Note, however, that PCRE's interpretation of \G, as the start of the
2722 current match, is subtly different from Perl's, which defines it as the
2723 end of the previous match. In Perl, these can be different when the
2724 previously matched string was empty. Because PCRE does just one match
2725 at a time, it cannot reproduce this behaviour.
2726
2727 If all the alternatives of a pattern begin with \G, the expression is
2728 anchored to the starting match position, and the "anchored" flag is set
2729 in the compiled regular expression.
2730
2731
2732 CIRCUMFLEX AND DOLLAR
2733
2734 Outside a character class, in the default matching mode, the circumflex
2735 character is an assertion that is true only if the current matching
2736 point is at the start of the subject string. If the startoffset argu-
2737 ment of pcre_exec() is non-zero, circumflex can never match if the
2738 PCRE_MULTILINE option is unset. Inside a character class, circumflex
2739 has an entirely different meaning (see below).
2740
2741 Circumflex need not be the first character of the pattern if a number
2742 of alternatives are involved, but it should be the first thing in each
2743 alternative in which it appears if the pattern is ever to match that
2744 branch. If all possible alternatives start with a circumflex, that is,
2745 if the pattern is constrained to match only at the start of the sub-
2746 ject, it is said to be an "anchored" pattern. (There are also other
2747 constructs that can cause a pattern to be anchored.)
2748
2749 A dollar character is an assertion that is true only if the current
2750 matching point is at the end of the subject string, or immediately
2751 before a newline character that is the last character in the string (by
2752 default). Dollar need not be the last character of the pattern if a
2753 number of alternatives are involved, but it should be the last item in
2754 any branch in which it appears. Dollar has no special meaning in a
2755 character class.
2756
2757 The meaning of dollar can be changed so that it matches only at the
2758 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
2759 compile time. This does not affect the \Z assertion.
2760
2761 The meanings of the circumflex and dollar characters are changed if the
2762 PCRE_MULTILINE option is set. When this is the case, they match immedi-
2763 ately after and immediately before an internal newline character,
2764 respectively, in addition to matching at the start and end of the sub-
2765 ject string. For example, the pattern /^abc$/ matches the subject
2766 string "def\nabc" (where \n represents a newline character) in multi-
2767 line mode, but not otherwise. Consequently, patterns that are anchored
2768 in single line mode because all branches start with ^ are not anchored
2769 in multiline mode, and a match for circumflex is possible when the
2770 startoffset argument of pcre_exec() is non-zero. The PCRE_DOL-
2771 LAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
2772
2773 Note that the sequences \A, \Z, and \z can be used to match the start
2774 and end of the subject in both modes, and if all branches of a pattern
2775 start with \A it is always anchored, whether PCRE_MULTILINE is set or
2776 not.
2777
2778
2779 FULL STOP (PERIOD, DOT)
2780
2781 Outside a character class, a dot in the pattern matches any one charac-
2782 ter in the subject, including a non-printing character, but not (by
2783 default) newline. In UTF-8 mode, a dot matches any UTF-8 character,
2784 which might be more than one byte long, except (by default) newline. If
2785 the PCRE_DOTALL option is set, dots match newlines as well. The han-
2786 dling of dot is entirely independent of the handling of circumflex and
2787 dollar, the only relationship being that they both involve newline
2788 characters. Dot has no special meaning in a character class.
2789
2790
2791 MATCHING A SINGLE BYTE
2792
2793 Outside a character class, the escape sequence \C matches any one byte,
2794 both in and out of UTF-8 mode. Unlike a dot, it can match a newline.
2795 The feature is provided in Perl in order to match individual bytes in
2796 UTF-8 mode. Because it breaks up UTF-8 characters into individual
2797 bytes, what remains in the string may be a malformed UTF-8 string. For
2798 this reason, the \C escape sequence is best avoided.
2799
2800 PCRE does not allow \C to appear in lookbehind assertions (described
2801 below), because in UTF-8 mode this would make it impossible to calcu-
2802 late the length of the lookbehind.
2803
2804
2805 SQUARE BRACKETS AND CHARACTER CLASSES
2806
2807 An opening square bracket introduces a character class, terminated by a
2808 closing square bracket. A closing square bracket on its own is not spe-
2809 cial. If a closing square bracket is required as a member of the class,
2810 it should be the first data character in the class (after an initial
2811 circumflex, if present) or escaped with a backslash.
2812
2813 A character class matches a single character in the subject. In UTF-8
2814 mode, the character may occupy more than one byte. A matched character
2815 must be in the set of characters defined by the class, unless the first
2816 character in the class definition is a circumflex, in which case the
2817 subject character must not be in the set defined by the class. If a
2818 circumflex is actually required as a member of the class, ensure it is
2819 not the first character, or escape it with a backslash.
2820
2821 For example, the character class [aeiou] matches any lower case vowel,
2822 while [^aeiou] matches any character that is not a lower case vowel.
2823 Note that a circumflex is just a convenient notation for specifying the
2824 characters that are in the class by enumerating those that are not. A
2825 class that starts with a circumflex is not an assertion: it still con-
2826 sumes a character from the subject string, and therefore it fails if
2827 the current pointer is at the end of the string.
2828
2829 In UTF-8 mode, characters with values greater than 255 can be included
2830 in a class as a literal string of bytes, or by using the \x{ escaping
2831 mechanism.
2832
2833 When caseless matching is set, any letters in a class represent both
2834 their upper case and lower case versions, so for example, a caseless
2835 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
2836 match "A", whereas a caseful version would. In UTF-8 mode, PCRE always
2837 understands the concept of case for characters whose values are less
2838 than 128, so caseless matching is always possible. For characters with
2839 higher values, the concept of case is supported if PCRE is compiled
2840 with Unicode property support, but not otherwise. If you want to use
2841 caseless matching for characters 128 and above, you must ensure that
2842 PCRE is compiled with Unicode property support as well as with UTF-8
2843 support.
2844
2845 The newline character is never treated in any special way in character
2846 classes, whatever the setting of the PCRE_DOTALL or PCRE_MULTILINE
2847 options is. A class such as [^a] will always match a newline.
2848
2849 The minus (hyphen) character can be used to specify a range of charac-
2850 ters in a character class. For example, [d-m] matches any letter
2851 between d and m, inclusive. If a minus character is required in a
2852 class, it must be escaped with a backslash or appear in a position
2853 where it cannot be interpreted as indicating a range, typically as the
2854 first or last character in the class.
2855
2856 It is not possible to have the literal character "]" as the end charac-
2857 ter of a range. A pattern such as [W-]46] is interpreted as a class of
2858 two characters ("W" and "-") followed by a literal string "46]", so it
2859 would match "W46]" or "-46]". However, if the "]" is escaped with a
2860 backslash it is interpreted as the end of range, so [W-\]46] is inter-
2861 preted as a class containing a range followed by two other characters.
2862 The octal or hexadecimal representation of "]" can also be used to end
2863 a range.
2864
2865 Ranges operate in the collating sequence of character values. They can
2866 also be used for characters specified numerically, for example
2867 [\000-\037]. In UTF-8 mode, ranges can include characters whose values
2868 are greater than 255, for example [\x{100}-\x{2ff}].
2869
2870 If a range that includes letters is used when caseless matching is set,
2871 it matches the letters in either case. For example, [W-c] is equivalent
2872 to [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if
2873 character tables for the "fr_FR" locale are in use, [\xc8-\xcb] matches
2874 accented E characters in both cases. In UTF-8 mode, PCRE supports the
2875 concept of case for characters with values greater than 128 only when
2876 it is compiled with Unicode property support.
2877
2878 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
2879 in a character class, and add the characters that they match to the
2880 class. For example, [\dABCDEF] matches any hexadecimal digit. A circum-
2881 flex can conveniently be used with the upper case character types to
2882 specify a more restricted set of characters than the matching lower
2883 case type. For example, the class [^\W_] matches any letter or digit,
2884 but not underscore.
2885
2886 The only metacharacters that are recognized in character classes are
2887 backslash, hyphen (only where it can be interpreted as specifying a
2888 range), circumflex (only at the start), opening square bracket (only
2889 when it can be interpreted as introducing a POSIX class name - see the
2890 next section), and the terminating closing square bracket. However,
2891 escaping other non-alphanumeric characters does no harm.
2892
2893
2894 POSIX CHARACTER CLASSES
2895
2896 Perl supports the POSIX notation for character classes. This uses names
2897 enclosed by [: and :] within the enclosing square brackets. PCRE also
2898 supports this notation. For example,
2899
2900 [01[:alpha:]%]
2901
2902 matches "0", "1", any alphabetic character, or "%". The supported class
2903 names are
2904
2905 alnum letters and digits
2906 alpha letters
2907 ascii character codes 0 - 127
2908 blank space or tab only
2909 cntrl control characters
2910 digit decimal digits (same as \d)
2911 graph printing characters, excluding space
2912 lower lower case letters
2913 print printing characters, including space
2914 punct printing characters, excluding letters and digits
2915 space white space (not quite the same as \s)
2916 upper upper case letters
2917 word "word" characters (same as \w)
2918 xdigit hexadecimal digits
2919
2920 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
2921 and space (32). Notice that this list includes the VT character (code
2922 11). This makes "space" different to \s, which does not include VT (for
2923 Perl compatibility).
2924
2925 The name "word" is a Perl extension, and "blank" is a GNU extension
2926 from Perl 5.8. Another Perl extension is negation, which is indicated
2927 by a ^ character after the colon. For example,
2928
2929 [12[:^digit:]]
2930
2931 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
2932 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
2933 these are not supported, and an error is given if they are encountered.
2934
2935 In UTF-8 mode, characters with values greater than 128 do not match any
2936 of the POSIX character classes.
2937
2938
2939 VERTICAL BAR
2940
2941 Vertical bar characters are used to separate alternative patterns. For
2942 example, the pattern
2943
2944 gilbert|sullivan
2945
2946 matches either "gilbert" or "sullivan". Any number of alternatives may
2947 appear, and an empty alternative is permitted (matching the empty
2948 string). The matching process tries each alternative in turn, from
2949 left to right, and the first one that succeeds is used. If the alterna-
2950 tives are within a subpattern (defined below), "succeeds" means match-
2951 ing the rest of the main pattern as well as the alternative in the sub-
2952 pattern.
2953
2954
2955 INTERNAL OPTION SETTING
2956
2957 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
2958 PCRE_EXTENDED options can be changed from within the pattern by a
2959 sequence of Perl option letters enclosed between "(?" and ")". The
2960 option letters are
2961
2962 i for PCRE_CASELESS
2963 m for PCRE_MULTILINE
2964 s for PCRE_DOTALL
2965 x for PCRE_EXTENDED
2966
2967 For example, (?im) sets caseless, multiline matching. It is also possi-
2968 ble to unset these options by preceding the letter with a hyphen, and a
2969 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
2970 LESS and PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED,
2971 is also permitted. If a letter appears both before and after the
2972 hyphen, the option is unset.
2973
2974 When an option change occurs at top level (that is, not inside subpat-
2975 tern parentheses), the change applies to the remainder of the pattern
2976 that follows. If the change is placed right at the start of a pattern,
2977 PCRE extracts it into the global options (and it will therefore show up
2978 in data extracted by the pcre_fullinfo() function).
2979
2980 An option change within a subpattern affects only that part of the cur-
2981 rent pattern that follows it, so
2982
2983 (a(?i)b)c
2984
2985 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
2986 used). By this means, options can be made to have different settings
2987 in different parts of the pattern. Any changes made in one alternative
2988 do carry on into subsequent branches within the same subpattern. For
2989 example,
2990
2991 (a(?i)b|c)
2992
2993 matches "ab", "aB", "c", and "C", even though when matching "C" the
2994 first branch is abandoned before the option setting. This is because
2995 the effects of option settings happen at compile time. There would be
2996 some very weird behaviour otherwise.
2997
2998 The PCRE-specific options PCRE_UNGREEDY and PCRE_EXTRA can be changed
2999 in the same way as the Perl-compatible options by using the characters
3000 U and X respectively. The (?X) flag setting is special in that it must
3001 always occur earlier in the pattern than any of the additional features
3002 it turns on, even when it is at top level. It is best to put it at the
3003 start.
3004
3005
3006 SUBPATTERNS
3007
3008 Subpatterns are delimited by parentheses (round brackets), which can be
3009 nested. Turning part of a pattern into a subpattern does two things:
3010
3011 1. It localizes a set of alternatives. For example, the pattern
3012
3013 cat(aract|erpillar|)
3014
3015 matches one of the words "cat", "cataract", or "caterpillar". Without
3016 the parentheses, it would match "cataract", "erpillar" or the empty
3017 string.
3018
3019 2. It sets up the subpattern as a capturing subpattern. This means
3020 that, when the whole pattern matches, that portion of the subject
3021 string that matched the subpattern is passed back to the caller via the
3022 ovector argument of pcre_exec(). Opening parentheses are counted from
3023 left to right (starting from 1) to obtain numbers for the capturing
3024 subpatterns.
3025
3026 For example, if the string "the red king" is matched against the pat-
3027 tern
3028
3029 the ((red|white) (king|queen))
3030
3031 the captured substrings are "red king", "red", and "king", and are num-
3032 bered 1, 2, and 3, respectively.
3033
3034 The fact that plain parentheses fulfil two functions is not always
3035 helpful. There are often times when a grouping subpattern is required
3036 without a capturing requirement. If an opening parenthesis is followed
3037 by a question mark and a colon, the subpattern does not do any captur-
3038 ing, and is not counted when computing the number of any subsequent
3039 capturing subpatterns. For example, if the string "the white queen" is
3040 matched against the pattern
3041
3042 the ((?:red|white) (king|queen))
3043
3044 the captured substrings are "white queen" and "queen", and are numbered
3045 1 and 2. The maximum number of capturing subpatterns is 65535, and the
3046 maximum depth of nesting of all subpatterns, both capturing and non-
3047 capturing, is 200.
3048
3049 As a convenient shorthand, if any option settings are required at the
3050 start of a non-capturing subpattern, the option letters may appear
3051 between the "?" and the ":". Thus the two patterns
3052
3053 (?i:saturday|sunday)
3054 (?:(?i)saturday|sunday)
3055
3056 match exactly the same set of strings. Because alternative branches are
3057 tried from left to right, and options are not reset until the end of
3058 the subpattern is reached, an option setting in one branch does affect
3059 subsequent branches, so the above patterns match "SUNDAY" as well as
3060 "Saturday".
3061
3062
3063 NAMED SUBPATTERNS
3064
3065 Identifying capturing parentheses by number is simple, but it can be
3066 very hard to keep track of the numbers in complicated regular expres-
3067 sions. Furthermore, if an expression is modified, the numbers may
3068 change. To help with this difficulty, PCRE supports the naming of sub-
3069 patterns, something that Perl does not provide. The Python syntax
3070 (?P<name>...) is used. Names consist of alphanumeric characters and
3071 underscores, and must be unique within a pattern.
3072
3073 Named capturing parentheses are still allocated numbers as well as
3074 names. The PCRE API provides function calls for extracting the name-to-
3075 number translation table from a compiled pattern. There is also a con-
3076 venience function for extracting a captured substring by name. For fur-
3077 ther details see the pcreapi documentation.
3078
3079
3080 REPETITION
3081
3082 Repetition is specified by quantifiers, which can follow any of the
3083 following items:
3084
3085 a literal data character
3086 the . metacharacter
3087 the \C escape sequence
3088 the \X escape sequence (in UTF-8 mode with Unicode properties)
3089 an escape such as \d that matches a single character
3090 a character class
3091 a back reference (see next section)
3092 a parenthesized subpattern (unless it is an assertion)
3093
3094 The general repetition quantifier specifies a minimum and maximum num-
3095 ber of permitted matches, by giving the two numbers in curly brackets
3096 (braces), separated by a comma. The numbers must be less than 65536,
3097 and the first must be less than or equal to the second. For example:
3098
3099 z{2,4}
3100
3101 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
3102 special character. If the second number is omitted, but the comma is
3103 present, there is no upper limit; if the second number and the comma
3104 are both omitted, the quantifier specifies an exact number of required
3105 matches. Thus
3106
3107 [aeiou]{3,}
3108
3109 matches at least 3 successive vowels, but may match many more, while
3110
3111 \d{8}
3112
3113 matches exactly 8 digits. An opening curly bracket that appears in a
3114 position where a quantifier is not allowed, or one that does not match
3115 the syntax of a quantifier, is taken as a literal character. For exam-
3116 ple, {,6} is not a quantifier, but a literal string of four characters.
3117
3118 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to
3119 individual bytes. Thus, for example, \x{100}{2} matches two UTF-8 char-
3120 acters, each of which is represented by a two-byte sequence. Similarly,
3121 when Unicode property support is available, \X{3} matches three Unicode
3122 extended sequences, each of which may be several bytes long (and they
3123 may be of different lengths).
3124
3125 The quantifier {0} is permitted, causing the expression to behave as if
3126 the previous item and the quantifier were not present.
3127
3128 For convenience (and historical compatibility) the three most common
3129 quantifiers have single-character abbreviations:
3130
3131 * is equivalent to {0,}
3132 + is equivalent to {1,}
3133 ? is equivalent to {0,1}
3134
3135 It is possible to construct infinite loops by following a subpattern
3136 that can match no characters with a quantifier that has no upper limit,
3137 for example:
3138
3139 (a?)*
3140
3141 Earlier versions of Perl and PCRE used to give an error at compile time
3142 for such patterns. However, because there are cases where this can be
3143 useful, such patterns are now accepted, but if any repetition of the
3144 subpattern does in fact match no characters, the loop is forcibly bro-
3145 ken.
3146
3147 By default, the quantifiers are "greedy", that is, they match as much
3148 as possible (up to the maximum number of permitted times), without
3149 causing the rest of the pattern to fail. The classic example of where
3150 this gives problems is in trying to match comments in C programs. These
3151 appear between /* and */ and within the comment, individual * and /
3152 characters may appear. An attempt to match C comments by applying the
3153 pattern
3154
3155 /\*.*\*/
3156
3157 to the string
3158
3159 /* first comment */ not comment /* second comment */
3160
3161 fails, because it matches the entire string owing to the greediness of
3162 the .* item.
3163
3164 However, if a quantifier is followed by a question mark, it ceases to
3165 be greedy, and instead matches the minimum number of times possible, so
3166 the pattern
3167
3168 /\*.*?\*/
3169
3170 does the right thing with the C comments. The meaning of the various
3171 quantifiers is not otherwise changed, just the preferred number of
3172 matches. Do not confuse this use of question mark with its use as a
3173 quantifier in its own right. Because it has two uses, it can sometimes
3174 appear doubled, as in
3175
3176 \d??\d
3177
3178 which matches one digit by preference, but can match two if that is the
3179 only way the rest of the pattern matches.
3180
3181 If the PCRE_UNGREEDY option is set (an option which is not available in
3182 Perl), the quantifiers are not greedy by default, but individual ones
3183 can be made greedy by following them with a question mark. In other
3184 words, it inverts the default behaviour.
3185
3186 When a parenthesized subpattern is quantified with a minimum repeat
3187 count that is greater than 1 or with a limited maximum, more memory is
3188 required for the compiled pattern, in proportion to the size of the
3189 minimum or maximum.
3190
3191 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
3192 alent to Perl's /s) is set, thus allowing the . to match newlines, the
3193 pattern is implicitly anchored, because whatever follows will be tried
3194 against every character position in the subject string, so there is no
3195 point in retrying the overall match at any position after the first.
3196 PCRE normally treats such a pattern as though it were preceded by \A.
3197
3198 In cases where it is known that the subject string contains no new-
3199 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
3200 mization, or alternatively using ^ to indicate anchoring explicitly.
3201
3202 However, there is one situation where the optimization cannot be used.
3203 When .* is inside capturing parentheses that are the subject of a
3204 backreference elsewhere in the pattern, a match at the start may fail,
3205 and a later one succeed. Consider, for example:
3206
3207 (.*)abc\1
3208
3209 If the subject is "xyz123abc123" the match point is the fourth charac-
3210 ter. For this reason, such a pattern is not implicitly anchored.
3211
3212 When a capturing subpattern is repeated, the value captured is the sub-
3213 string that matched the final iteration. For example, after
3214
3215 (tweedle[dume]{3}\s*)+
3216
3217 has matched "tweedledum tweedledee" the value of the captured substring
3218 is "tweedledee". However, if there are nested capturing subpatterns,
3219 the corresponding captured values may have been set in previous itera-
3220 tions. For example, after
3221
3222 /(a|(b))+/
3223
3224 matches "aba" the value of the second captured substring is "b".
3225
3226
3227 ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS
3228
3229 With both maximizing and minimizing repetition, failure of what follows
3230 normally causes the repeated item to be re-evaluated to see if a dif-
3231 ferent number of repeats allows the rest of the pattern to match. Some-
3232 times it is useful to prevent this, either to change the nature of the
3233 match, or to cause it fail earlier than it otherwise might, when the
3234 author of the pattern knows there is no point in carrying on.
3235
3236 Consider, for example, the pattern \d+foo when applied to the subject
3237 line
3238
3239 123456bar
3240
3241 After matching all 6 digits and then failing to match "foo", the normal
3242 action of the matcher is to try again with only 5 digits matching the
3243 \d+ item, and then with 4, and so on, before ultimately failing.
3244 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
3245 the means for specifying that once a subpattern has matched, it is not
3246 to be re-evaluated in this way.
3247
3248 If we use atomic grouping for the previous example, the matcher would
3249 give up immediately on failing to match "foo" the first time. The nota-
3250 tion is a kind of special parenthesis, starting with (?> as in this
3251 example:
3252
3253 (?>\d+)foo
3254
3255 This kind of parenthesis "locks up" the part of the pattern it con-
3256 tains once it has matched, and a failure further into the pattern is
3257 prevented from backtracking into it. Backtracking past it to previous
3258 items, however, works as normal.
3259
3260 An alternative description is that a subpattern of this type matches
3261 the string of characters that an identical standalone pattern would
3262 match, if anchored at the current point in the subject string.
3263
3264 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
3265 such as the above example can be thought of as a maximizing repeat that
3266 must swallow everything it can. So, while both \d+ and \d+? are pre-
3267 pared to adjust the number of digits they match in order to make the
3268 rest of the pattern match, (?>\d+) can only match an entire sequence of
3269 digits.
3270
3271 Atomic groups in general can of course contain arbitrarily complicated
3272 subpatterns, and can be nested. However, when the subpattern for an
3273 atomic group is just a single repeated item, as in the example above, a
3274 simpler notation, called a "possessive quantifier" can be used. This
3275 consists of an additional + character following a quantifier. Using
3276 this notation, the previous example can be rewritten as
3277
3278 \d++foo
3279
3280 Possessive quantifiers are always greedy; the setting of the
3281 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
3282 simpler forms of atomic group. However, there is no difference in the
3283 meaning or processing of a possessive quantifier and the equivalent
3284 atomic group.
3285
3286 The possessive quantifier syntax is an extension to the Perl syntax. It
3287 originates in Sun's Java package.
3288
3289 When a pattern contains an unlimited repeat inside a subpattern that
3290 can itself be repeated an unlimited number of times, the use of an
3291 atomic group is the only way to avoid some failing matches taking a
3292 very long time indeed. The pattern
3293
3294 (\D+|<\d+>)*[!?]
3295
3296 matches an unlimited number of substrings that either consist of non-
3297 digits, or digits enclosed in <>, followed by either ! or ?. When it
3298 matches, it runs quickly. However, if it is applied to
3299
3300 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
3301
3302 it takes a long time before reporting failure. This is because the
3303 string can be divided between the internal \D+ repeat and the external
3304 * repeat in a large number of ways, and all have to be tried. (The
3305 example uses [!?] rather than a single character at the end, because
3306 both PCRE and Perl have an optimization that allows for fast failure
3307 when a single character is used. They remember the last single charac-
3308 ter that is required for a match, and fail early if it is not present
3309 in the string.) If the pattern is changed so that it uses an atomic
3310 group, like this:
3311
3312 ((?>\D+)|<\d+>)*[!?]
3313
3314 sequences of non-digits cannot be broken, and failure happens quickly.
3315
3316
3317 BACK REFERENCES
3318
3319 Outside a character class, a backslash followed by a digit greater than
3320 0 (and possibly further digits) is a back reference to a capturing sub-
3321 pattern earlier (that is, to its left) in the pattern, provided there
3322 have been that many previous capturing left parentheses.
3323
3324 However, if the decimal number following the backslash is less than 10,
3325 it is always taken as a back reference, and causes an error only if
3326 there are not that many capturing left parentheses in the entire pat-
3327 tern. In other words, the parentheses that are referenced need not be
3328 to the left of the reference for numbers less than 10. See the subsec-
3329 tion entitled "Non-printing characters" above for further details of
3330 the handling of digits following a backslash.
3331
3332 A back reference matches whatever actually matched the capturing sub-
3333 pattern in the current subject string, rather than anything matching
3334 the subpattern itself (see "Subpatterns as subroutines" below for a way
3335 of doing that). So the pattern
3336
3337 (sens|respons)e and \1ibility
3338
3339 matches "sense and sensibility" and "response and responsibility", but
3340 not "sense and responsibility". If caseful matching is in force at the
3341 time of the back reference, the case of letters is relevant. For exam-
3342 ple,
3343
3344 ((?i)rah)\s+\1
3345
3346 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
3347 original capturing subpattern is matched caselessly.
3348
3349 Back references to named subpatterns use the Python syntax (?P=name).
3350 We could rewrite the above example as follows:
3351
3352 (?<p1>(?i)rah)\s+(?P=p1)
3353
3354 There may be more than one back reference to the same subpattern. If a
3355 subpattern has not actually been used in a particular match, any back
3356 references to it always fail. For example, the pattern
3357
3358 (a|(bc))\2
3359
3360 always fails if it starts to match "a" rather than "bc". Because there
3361 may be many capturing parentheses in a pattern, all digits following
3362 the backslash are taken as part of a potential back reference number.
3363 If the pattern continues with a digit character, some delimiter must be
3364 used to terminate the back reference. If the PCRE_EXTENDED option is
3365 set, this can be whitespace. Otherwise an empty comment (see "Com-
3366 ments" below) can be used.
3367
3368 A back reference that occurs inside the parentheses to which it refers
3369 fails when the subpattern is first used, so, for example, (a\1) never
3370 matches. However, such references can be useful inside repeated sub-
3371 patterns. For example, the pattern
3372
3373 (a|b\1)+
3374
3375 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
3376 ation of the subpattern, the back reference matches the character
3377 string corresponding to the previous iteration. In order for this to
3378 work, the pattern must be such that the first iteration does not need
3379 to match the back reference. This can be done using alternation, as in
3380 the example above, or by a quantifier with a minimum of zero.
3381
3382
3383 ASSERTIONS
3384
3385 An assertion is a test on the characters following or preceding the
3386 current matching point that does not actually consume any characters.
3387 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
3388 described above.
3389
3390 More complicated assertions are coded as subpatterns. There are two
3391 kinds: those that look ahead of the current position in the subject
3392 string, and those that look behind it. An assertion subpattern is
3393 matched in the normal way, except that it does not cause the current
3394 matching position to be changed.
3395
3396 Assertion subpatterns are not capturing subpatterns, and may not be
3397 repeated, because it makes no sense to assert the same thing several
3398 times. If any kind of assertion contains capturing subpatterns within
3399 it, these are counted for the purposes of numbering the capturing sub-
3400 patterns in the whole pattern. However, substring capturing is carried
3401 out only for positive assertions, because it does not make sense for
3402 negative assertions.
3403
3404 Lookahead assertions
3405
3406 Lookahead assertions start with (?= for positive assertions and (?! for
3407 negative assertions. For example,
3408
3409 \w+(?=;)
3410
3411 matches a word followed by a semicolon, but does not include the semi-
3412 colon in the match, and
3413
3414 foo(?!bar)
3415
3416 matches any occurrence of "foo" that is not followed by "bar". Note
3417 that the apparently similar pattern
3418
3419 (?!foo)bar
3420
3421 does not find an occurrence of "bar" that is preceded by something
3422 other than "foo"; it finds any occurrence of "bar" whatsoever, because
3423 the assertion (?!foo) is always true when the next three characters are
3424 "bar". A lookbehind assertion is needed to achieve the other effect.
3425
3426 If you want to force a matching failure at some point in a pattern, the
3427 most convenient way to do it is with (?!) because an empty string
3428 always matches, so an assertion that requires there not to be an empty
3429 string must always fail.
3430
3431 Lookbehind assertions
3432
3433 Lookbehind assertions start with (?<= for positive assertions and (?<!
3434 for negative assertions. For example,
3435
3436 (?<!foo)bar
3437
3438 does find an occurrence of "bar" that is not preceded by "foo". The
3439 contents of a lookbehind assertion are restricted such that all the
3440 strings it matches must have a fixed length. However, if there are sev-
3441 eral alternatives, they do not all have to have the same fixed length.
3442 Thus
3443
3444 (?<=bullock|donkey)
3445
3446 is permitted, but
3447
3448 (?<!dogs?|cats?)
3449
3450 causes an error at compile time. Branches that match different length
3451 strings are permitted only at the top level of a lookbehind assertion.
3452 This is an extension compared with Perl (at least for 5.8), which
3453 requires all branches to match the same length of string. An assertion
3454 such as
3455
3456 (?<=ab(c|de))
3457
3458 is not permitted, because its single top-level branch can match two
3459 different lengths, but it is acceptable if rewritten to use two top-
3460 level branches:
3461
3462 (?<=abc|abde)
3463
3464 The implementation of lookbehind assertions is, for each alternative,
3465 to temporarily move the current position back by the fixed width and
3466 then try to match. If there are insufficient characters before the cur-
3467 rent position, the match is deemed to fail.
3468
3469 PCRE does not allow the \C escape (which matches a single byte in UTF-8
3470 mode) to appear in lookbehind assertions, because it makes it impossi-
3471 ble to calculate the length of the lookbehind. The \X escape, which can
3472 match different numbers of bytes, is also not permitted.
3473
3474 Atomic groups can be used in conjunction with lookbehind assertions to
3475 specify efficient matching at the end of the subject string. Consider a
3476 simple pattern such as
3477
3478 abcd$
3479
3480 when applied to a long string that does not match. Because matching
3481 proceeds from left to right, PCRE will look for each "a" in the subject
3482 and then see if what follows matches the rest of the pattern. If the
3483 pattern is specified as
3484
3485 ^.*abcd$
3486
3487 the initial .* matches the entire string at first, but when this fails
3488 (because there is no following "a"), it backtracks to match all but the
3489 last character, then all but the last two characters, and so on. Once
3490 again the search for "a" covers the entire string, from right to left,
3491 so we are no better off. However, if the pattern is written as
3492
3493 ^(?>.*)(?<=abcd)
3494
3495 or, equivalently, using the possessive quantifier syntax,
3496
3497 ^.*+(?<=abcd)
3498
3499 there can be no backtracking for the .* item; it can match only the
3500 entire string. The subsequent lookbehind assertion does a single test
3501 on the last four characters. If it fails, the match fails immediately.
3502 For long strings, this approach makes a significant difference to the
3503 processing time.
3504
3505 Using multiple assertions
3506
3507 Several assertions (of any sort) may occur in succession. For example,
3508
3509 (?<=\d{3})(?<!999)foo
3510
3511 matches "foo" preceded by three digits that are not "999". Notice that
3512 each of the assertions is applied independently at the same point in
3513 the subject string. First there is a check that the previous three
3514 characters are all digits, and then there is a check that the same
3515 three characters are not "999". This pattern does not match "foo" pre-
3516 ceded by six characters, the first of which are digits and the last
3517 three of which are not "999". For example, it doesn't match "123abc-
3518 foo". A pattern to do that is
3519
3520 (?<=\d{3}...)(?<!999)foo
3521
3522 This time the first assertion looks at the preceding six characters,
3523 checking that the first three are digits, and then the second assertion
3524 checks that the preceding three characters are not "999".
3525
3526 Assertions can be nested in any combination. For example,
3527
3528 (?<=(?<!foo)bar)baz
3529
3530 matches an occurrence of "baz" that is preceded by "bar" which in turn
3531 is not preceded by "foo", while
3532
3533 (?<=\d{3}(?!999)...)foo
3534
3535 is another pattern that matches "foo" preceded by three digits and any
3536 three characters that are not "999".
3537
3538
3539 CONDITIONAL SUBPATTERNS
3540
3541 It is possible to cause the matching process to obey a subpattern con-
3542 ditionally or to choose between two alternative subpatterns, depending
3543 on the result of an assertion, or whether a previous capturing subpat-
3544 tern matched or not. The two possible forms of conditional subpattern
3545 are
3546
3547 (?(condition)yes-pattern)
3548 (?(condition)yes-pattern|no-pattern)
3549
3550 If the condition is satisfied, the yes-pattern is used; otherwise the
3551 no-pattern (if present) is used. If there are more than two alterna-
3552 tives in the subpattern, a compile-time error occurs.
3553
3554 There are three kinds of condition. If the text between the parentheses
3555 consists of a sequence of digits, the condition is satisfied if the
3556 capturing subpattern of that number has previously matched. The number
3557 must be greater than zero. Consider the following pattern, which con-
3558 tains non-significant white space to make it more readable (assume the
3559 PCRE_EXTENDED option) and to divide it into three parts for ease of
3560 discussion:
3561
3562 ( \( )? [^()]+ (?(1) \) )
3563
3564 The first part matches an optional opening parenthesis, and if that
3565 character is present, sets it as the first captured substring. The sec-
3566 ond part matches one or more characters that are not parentheses. The
3567 third part is a conditional subpattern that tests whether the first set
3568 of parentheses matched or not. If they did, that is, if subject started
3569 with an opening parenthesis, the condition is true, and so the yes-pat-
3570 tern is executed and a closing parenthesis is required. Otherwise,
3571 since no-pattern is not present, the subpattern matches nothing. In
3572 other words, this pattern matches a sequence of non-parentheses,
3573 optionally enclosed in parentheses.
3574
3575 If the condition is the string (R), it is satisfied if a recursive call
3576 to the pattern or subpattern has been made. At "top level", the condi-
3577 tion is false. This is a PCRE extension. Recursive patterns are
3578 described in the next section.
3579
3580 If the condition is not a sequence of digits or (R), it must be an
3581 assertion. This may be a positive or negative lookahead or lookbehind
3582 assertion. Consider this pattern, again containing non-significant
3583 white space, and with the two alternatives on the second line:
3584
3585 (?(?=[^a-z]*[a-z])
3586 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
3587
3588 The condition is a positive lookahead assertion that matches an
3589 optional sequence of non-letters followed by a letter. In other words,
3590 it tests for the presence of at least one letter in the subject. If a
3591 letter is found, the subject is matched against the first alternative;
3592 otherwise it is matched against the second. This pattern matches
3593 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
3594 letters and dd are digits.
3595
3596
3597 COMMENTS
3598
3599 The sequence (?# marks the start of a comment that continues up to the
3600 next closing parenthesis. Nested parentheses are not permitted. The
3601 characters that make up a comment play no part in the pattern matching
3602 at all.
3603
3604 If the PCRE_EXTENDED option is set, an unescaped # character outside a
3605 character class introduces a comment that continues up to the next new-
3606 line character in the pattern.
3607
3608
3609 RECURSIVE PATTERNS
3610
3611 Consider the problem of matching a string in parentheses, allowing for
3612 unlimited nested parentheses. Without the use of recursion, the best
3613 that can be done is to use a pattern that matches up to some fixed
3614 depth of nesting. It is not possible to handle an arbitrary nesting
3615 depth. Perl provides a facility that allows regular expressions to
3616 recurse (amongst other things). It does this by interpolating Perl code
3617 in the expression at run time, and the code can refer to the expression
3618 itself. A Perl pattern to solve the parentheses problem can be created
3619 like this:
3620
3621 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
3622
3623 The (?p{...}) item interpolates Perl code at run time, and in this case
3624 refers recursively to the pattern in which it appears. Obviously, PCRE
3625 cannot support the interpolation of Perl code. Instead, it supports
3626 some special syntax for recursion of the entire pattern, and also for
3627 individual subpattern recursion.
3628
3629 The special item that consists of (? followed by a number greater than
3630 zero and a closing parenthesis is a recursive call of the subpattern of
3631 the given number, provided that it occurs inside that subpattern. (If
3632 not, it is a "subroutine" call, which is described in the next sec-
3633 tion.) The special item (?R) is a recursive call of the entire regular
3634 expression.
3635
3636 For example, this PCRE pattern solves the nested parentheses problem
3637 (assume the PCRE_EXTENDED option is set so that white space is
3638 ignored):
3639
3640 \( ( (?>[^()]+) | (?R) )* \)
3641
3642 First it matches an opening parenthesis. Then it matches any number of
3643 substrings which can either be a sequence of non-parentheses, or a
3644 recursive match of the pattern itself (that is a correctly parenthe-
3645 sized substring). Finally there is a closing parenthesis.
3646
3647 If this were part of a larger pattern, you would not want to recurse
3648 the entire pattern, so instead you could use this:
3649
3650 ( \( ( (?>[^()]+) | (?1) )* \) )
3651
3652 We have put the pattern into parentheses, and caused the recursion to
3653 refer to them instead of the whole pattern. In a larger pattern, keep-
3654 ing track of parenthesis numbers can be tricky. It may be more conve-
3655 nient to use named parentheses instead. For this, PCRE uses (?P>name),
3656 which is an extension to the Python syntax that PCRE uses for named
3657 parentheses (Perl does not provide named parentheses). We could rewrite
3658 the above example as follows:
3659
3660 (?P<pn> \( ( (?>[^()]+) | (?P>pn) )* \) )
3661
3662 This particular example pattern contains nested unlimited repeats, and
3663 so the use of atomic grouping for matching strings of non-parentheses
3664 is important when applying the pattern to strings that do not match.
3665 For example, when this pattern is applied to
3666
3667 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
3668
3669 it yields "no match" quickly. However, if atomic grouping is not used,
3670 the match runs for a very long time indeed because there are so many
3671 different ways the + and * repeats can carve up the subject, and all
3672 have to be tested before failure can be reported.
3673
3674 At the end of a match, the values set for any capturing subpatterns are
3675 those from the outermost level of the recursion at which the subpattern
3676 value is set. If you want to obtain intermediate values, a callout
3677 function can be used (see the next section and the pcrecallout documen-
3678 tation). If the pattern above is matched against
3679
3680 (ab(cd)ef)
3681
3682 the value for the capturing parentheses is "ef", which is the last
3683 value taken on at the top level. If additional parentheses are added,
3684 giving
3685
3686 \( ( ( (?>[^()]+) | (?R) )* ) \)
3687 ^ ^
3688 ^ ^
3689
3690 the string they capture is "ab(cd)ef", the contents of the top level
3691 parentheses. If there are more than 15 capturing parentheses in a pat-
3692 tern, PCRE has to obtain extra memory to store data during a recursion,
3693 which it does by using pcre_malloc, freeing it via pcre_free after-
3694 wards. If no memory can be obtained, the match fails with the
3695 PCRE_ERROR_NOMEMORY error.
3696
3697 Do not confuse the (?R) item with the condition (R), which tests for
3698 recursion. Consider this pattern, which matches text in angle brack-
3699 ets, allowing for arbitrary nesting. Only digits are allowed in nested
3700 brackets (that is, when recursing), whereas any characters are permit-
3701 ted at the outer level.
3702
3703 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
3704
3705 In this pattern, (?(R) is the start of a conditional subpattern, with
3706 two different alternatives for the recursive and non-recursive cases.
3707 The (?R) item is the actual recursive call.
3708
3709
3710 SUBPATTERNS AS SUBROUTINES
3711
3712 If the syntax for a recursive subpattern reference (either by number or
3713 by name) is used outside the parentheses to which it refers, it oper-
3714 ates like a subroutine in a programming language. An earlier example
3715 pointed out that the pattern
3716
3717 (sens|respons)e and \1ibility
3718
3719 matches "sense and sensibility" and "response and responsibility", but
3720 not "sense and responsibility". If instead the pattern
3721
3722 (sens|respons)e and (?1)ibility
3723
3724 is used, it does match "sense and responsibility" as well as the other
3725 two strings. Such references must, however, follow the subpattern to
3726 which they refer.
3727
3728
3729 CALLOUTS
3730
3731 Perl has a feature whereby using the sequence (?{...}) causes arbitrary
3732 Perl code to be obeyed in the middle of matching a regular expression.
3733 This makes it possible, amongst other things, to extract different sub-
3734 strings that match the same pair of parentheses when there is a repeti-
3735 tion.
3736
3737 PCRE provides a similar feature, but of course it cannot obey arbitrary
3738 Perl code. The feature is called "callout". The caller of PCRE provides
3739 an external function by putting its entry point in the global variable
3740 pcre_callout. By default, this variable contains NULL, which disables
3741 all calling out.
3742
3743 Within a regular expression, (?C) indicates the points at which the
3744 external function is to be called. If you want to identify different
3745 callout points, you can put a number less than 256 after the letter C.
3746 The default value is zero. For example, this pattern has two callout
3747 points:
3748
3749 (?C1)abc(?C2)def
3750
3751 If the PCRE_AUTO_CALLOUT flag is passed to pcre_compile(), callouts are
3752 automatically installed before each item in the pattern. They are all
3753 numbered 255.
3754
3755 During matching, when PCRE reaches a callout point (and pcre_callout is
3756 set), the external function is called. It is provided with the number
3757 of the callout, the position in the pattern, and, optionally, one item
3758 of data originally supplied by the caller of pcre_exec(). The callout
3759 function may cause matching to proceed, to backtrack, or to fail alto-
3760 gether. A complete description of the interface to the callout function
3761 is given in the pcrecallout documentation.
3762
3763 Last updated: 28 February 2005
3764 Copyright (c) 1997-2005 University of Cambridge.
3765 ------------------------------------------------------------------------------
3766
3767
3768 PCREPARTIAL(3) PCREPARTIAL(3)
3769
3770
3771 NAME
3772 PCRE - Perl-compatible regular expressions
3773
3774
3775 PARTIAL MATCHING IN PCRE
3776
3777 In normal use of PCRE, if the subject string that is passed to
3778 pcre_exec() or pcre_dfa_exec() matches as far as it goes, but is too
3779 short to match the entire pattern, PCRE_ERROR_NOMATCH is returned.
3780 There are circumstances where it might be helpful to distinguish this
3781 case from other cases in which there is no match.
3782
3783 Consider, for example, an application where a human is required to type
3784 in data for a field with specific formatting requirements. An example
3785 might be a date in the form ddmmmyy, defined by this pattern:
3786
3787 ^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$
3788
3789 If the application sees the user's keystrokes one by one, and can check
3790 that what has been typed so far is potentially valid, it is able to
3791 raise an error as soon as a mistake is made, possibly beeping and not
3792 reflecting the character that has been typed. This immediate feedback
3793 is likely to be a better user interface than a check that is delayed
3794 until the entire string has been entered.
3795
3796 PCRE supports the concept of partial matching by means of the PCRE_PAR-
3797 TIAL option, which can be set when calling pcre_exec() or
3798 pcre_dfa_exec(). When this flag is set for pcre_exec(), the return code
3799 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if at any time
3800 during the matching process the last part of the subject string matched
3801 part of the pattern. Unfortunately, for non-anchored matching, it is
3802 not possible to obtain the position of the start of the partial match.
3803 No captured data is set when PCRE_ERROR_PARTIAL is returned.
3804
3805 When PCRE_PARTIAL is set for pcre_dfa_exec(), the return code
3806 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end of
3807 the subject is reached, there have been no complete matches, but there
3808 is still at least one matching possibility. The portion of the string
3809 that provided the partial match is set as the first matching string.
3810
3811 Using PCRE_PARTIAL disables one of PCRE's optimizations. PCRE remembers
3812 the last literal byte in a pattern, and abandons matching immediately
3813 if such a byte is not present in the subject string. This optimization
3814 cannot be used for a subject string that might match only partially.
3815
3816
3817 RESTRICTED PATTERNS FOR PCRE_PARTIAL
3818
3819 Because of the way certain internal optimizations are implemented in
3820 the pcre_exec() function, the PCRE_PARTIAL option cannot be used with
3821 all patterns. These restrictions do not apply when pcre_dfa_exec() is
3822 used. For pcre_exec(), repeated single characters such as
3823
3824 a{2,4}
3825
3826 and repeated single metasequences such as
3827
3828 \d+
3829
3830 are not permitted if the maximum number of occurrences is greater than
3831 one. Optional items such as \d? (where the maximum is one) are permit-
3832 ted. Quantifiers with any values are permitted after parentheses, so
3833 the invalid examples above can be coded thus:
3834
3835 (a){2,4}
3836 (\d)+
3837
3838 These constructions run more slowly, but for the kinds of application
3839 that are envisaged for this facility, this is not felt to be a major
3840 restriction.
3841
3842 If PCRE_PARTIAL is set for a pattern that does not conform to the
3843 restrictions, pcre_exec() returns the error code PCRE_ERROR_BADPARTIAL
3844 (-13).
3845
3846
3847 EXAMPLE OF PARTIAL MATCHING USING PCRETEST
3848
3849 If the escape sequence \P is present in a pcretest data line, the
3850 PCRE_PARTIAL flag is used for the match. Here is a run of pcretest that
3851 uses the date example quoted above:
3852
3853 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
3854 data> 25jun04P
3855 0: 25jun04
3856 1: jun
3857 data> 25dec3P
3858 Partial match
3859 data> 3juP
3860 Partial match
3861 data> 3jujP
3862 No match
3863 data> jP
3864 No match
3865
3866 The first data string is matched completely, so pcretest shows the
3867 matched substrings. The remaining four strings do not match the com-
3868 plete pattern, but the first two are partial matches. The same test,
3869 using DFA matching (by means of the \D escape sequence), produces the
3870 following output:
3871
3872 re> /^?(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)$/
3873 data> 25jun04\P\D
3874 0: 25jun04
3875 data> 23dec3\P\D
3876 Partial match: 23dec3
3877 data> 3ju\P\D
3878 Partial match: 3ju
3879 data> 3juj\P\D
3880 No match
3881 data> j\P\D
3882 No match
3883
3884 Notice that in this case the portion of the string that was matched is
3885 made available.
3886
3887
3888 MULTI-SEGMENT MATCHING WITH pcre_dfa_exec()
3889
3890 When a partial match has been found using pcre_dfa_exec(), it is possi-
3891 ble to continue the match by providing additional subject data and
3892 calling pcre_dfa_exec() again with the PCRE_DFA_RESTART option and the
3893 same working space (where details of the previous partial match are
3894 stored). Here is an example using pcretest, where the \R escape
3895 sequence sets the PCRE_DFA_RESTART option and the \D escape sequence
3896 requests the use of pcre_dfa_exec():
3897
3898 re> /^?(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)$/
3899 data> 23ja\P\D
3900 Partial match: 23ja
3901 data> n05\R\D
3902 0: n05
3903
3904 The first call has "23ja" as the subject, and requests partial match-
3905 ing; the second call has "n05" as the subject for the continued
3906 (restarted) match. Notice that when the match is complete, only the
3907 last part is shown; PCRE does not retain the previously partially-
3908 matched string. It is up to the calling program to do that if it needs
3909 to.
3910
3911 This facility can be used to pass very long subject strings to
3912 pcre_dfa_exec(). However, some care is needed for certain types of pat-
3913 tern.
3914
3915 1. If the pattern contains tests for the beginning or end of a line,
3916 you need to pass the PCRE_NOTBOL or PCRE_NOTEOL options, as appropri-
3917 ate, when the subject string for any call does not contain the begin-
3918 ning or end of a line.
3919
3920 2. If the pattern contains backward assertions (including \b or \B),
3921 you need to arrange for some overlap in the subject strings to allow
3922 for this. For example, you could pass the subject in chunks that were
3923 500 bytes long, but in a buffer of 700 bytes, with the starting offset
3924 set to 200 and the previous 200 bytes at the start of the buffer.
3925
3926 3. Matching a subject string that is split into multiple segments does
3927 not always produce exactly the same result as matching over one single
3928 long string. The difference arises when there are multiple matching
3929 possibilities, because a partial match result is given only when there
3930 are no completed matches in a call to fBpcre_dfa_exec(). This means
3931 that as soon as the shortest match has been found, continuation to a
3932 new subject segment is no longer possible. Consider this pcretest
3933 example:
3934
3935 re> /dog(sbody)?/
3936 data> do\P\D
3937 Partial match: do
3938 data> gsb\R\P\D
3939 0: g
3940 data> dogsbody\D
3941 0: dogsbody
3942 1: dog
3943
3944 The pattern matches the words "dog" or "dogsbody". When the subject is
3945 presented in several parts ("do" and "gsb" being the first two) the
3946 match stops when "dog" has been found, and it is not possible to con-
3947 tinue. On the other hand, if "dogsbody" is presented as a single
3948 string, both matches are found.
3949
3950 Because of this phenomenon, it does not usually make sense to end a
3951 pattern that is going to be matched in this way with a variable repeat.
3952
3953 Last updated: 28 February 2005
3954 Copyright (c) 1997-2005 University of Cambridge.
3955 ------------------------------------------------------------------------------
3956
3957
3958 PCREPRECOMPILE(3) PCREPRECOMPILE(3)
3959
3960
3961 NAME
3962 PCRE - Perl-compatible regular expressions
3963
3964
3965 SAVING AND RE-USING PRECOMPILED PCRE PATTERNS
3966
3967 If you are running an application that uses a large number of regular
3968 expression patterns, it may be useful to store them in a precompiled
3969 form instead of having to compile them every time the application is
3970 run. If you are not using any private character tables (see the
3971 pcre_maketables() documentation), this is relatively straightforward.
3972 If you are using private tables, it is a little bit more complicated.
3973
3974 If you save compiled patterns to a file, you can copy them to a differ-
3975 ent host and run them there. This works even if the new host has the
3976 opposite endianness to the one on which the patterns were compiled.
3977 There may be a small performance penalty, but it should be insignifi-
3978 cant.
3979
3980
3981 SAVING A COMPILED PATTERN
3982 The value returned by pcre_compile() points to a single block of memory
3983 that holds the compiled pattern and associated data. You can find the
3984 length of this block in bytes by calling pcre_fullinfo() with an argu-
3985 ment of PCRE_INFO_SIZE. You can then save the data in any appropriate
3986 manner. Here is sample code that compiles a pattern and writes it to a
3987 file. It assumes that the variable fd refers to a file that is open for
3988 output:
3989
3990 int erroroffset, rc, size;
3991 char *error;
3992 pcre *re;
3993
3994 re = pcre_compile("my pattern", 0, &error, &erroroffset, NULL);
3995 if (re == NULL) { ... handle errors ... }
3996 rc = pcre_fullinfo(re, NULL, PCRE_INFO_SIZE, &size);
3997 if (rc < 0) { ... handle errors ... }
3998 rc = fwrite(re, 1, size, fd);
3999 if (rc != size) { ... handle errors ... }
4000
4001 In this example, the bytes that comprise the compiled pattern are
4002 copied exactly. Note that this is binary data that may contain any of
4003 the 256 possible byte values. On systems that make a distinction
4004 between binary and non-binary data, be sure that the file is opened for
4005 binary output.
4006
4007 If you want to write more than one pattern to a file, you will have to
4008 devise a way of separating them. For binary data, preceding each pat-
4009 tern with its length is probably the most straightforward approach.
4010 Another possibility is to write out the data in hexadecimal instead of
4011 binary, one pattern to a line.
4012
4013 Saving compiled patterns in a file is only one possible way of storing
4014 them for later use. They could equally well be saved in a database, or
4015 in the memory of some daemon process that passes them via sockets to
4016 the processes that want them.
4017
4018 If the pattern has been studied, it is also possible to save the study
4019 data in a similar way to the compiled pattern itself. When studying
4020 generates additional information, pcre_study() returns a pointer to a
4021 pcre_extra data block. Its format is defined in the section on matching
4022 a pattern in the pcreapi documentation. The study_data field points to
4023 the binary study data, and this is what you must save (not the
4024 pcre_extra block itself). The length of the study data can be obtained
4025 by calling pcre_fullinfo() with an argument of PCRE_INFO_STUDYSIZE.
4026 Remember to check that pcre_study() did return a non-NULL value before
4027 trying to save the study data.
4028
4029
4030 RE-USING A PRECOMPILED PATTERN
4031
4032 Re-using a precompiled pattern is straightforward. Having reloaded it
4033 into main memory, you pass its pointer to pcre_exec() or
4034 pcre_dfa_exec() in the usual way. This should work even on another
4035 host, and even if that host has the opposite endianness to the one
4036 where the pattern was compiled.
4037
4038 However, if you passed a pointer to custom character tables when the
4039 pattern was compiled (the tableptr argument of pcre_compile()), you
4040 must now pass a similar pointer to pcre_exec() or pcre_dfa_exec(),
4041 because the value saved with the compiled pattern will obviously be
4042 nonsense. A field in a pcre_extra() block is used to pass this data, as
4043 described in the section on matching a pattern in the pcreapi documen-
4044 tation.
4045
4046 If you did not provide custom character tables when the pattern was
4047 compiled, the pointer in the compiled pattern is NULL, which causes
4048 pcre_exec() to use PCRE's internal tables. Thus, you do not need to
4049 take any special action at run time in this case.
4050
4051 If you saved study data with the compiled pattern, you need to create
4052 your own pcre_extra data block and set the study_data field to point to
4053 the reloaded study data. You must also set the PCRE_EXTRA_STUDY_DATA
4054 bit in the flags field to indicate that study data is present. Then
4055 pass the pcre_extra block to pcre_exec() or pcre_dfa_exec() in the
4056 usual way.
4057
4058
4059 COMPATIBILITY WITH DIFFERENT PCRE RELEASES
4060
4061 The layout of the control block that is at the start of the data that
4062 makes up a compiled pattern was changed for release 5.0. If you have
4063 any saved patterns that were compiled with previous releases (not a
4064 facility that was previously advertised), you will have to recompile
4065 them for release 5.0. However, from now on, it should be possible to
4066 make changes in a compatible manner.
4067
4068 Last updated: 28 February 2005
4069 Copyright (c) 1997-2005 University of Cambridge.
4070 ------------------------------------------------------------------------------
4071
4072
4073 PCREPERFORM(3) PCREPERFORM(3)
4074
4075
4076 NAME
4077 PCRE - Perl-compatible regular expressions
4078
4079
4080 PCRE PERFORMANCE
4081
4082 Certain items that may appear in regular expression patterns are more
4083 efficient than others. It is more efficient to use a character class
4084 like [aeiou] than a set of alternatives such as (a|e|i|o|u). In gen-
4085 eral, the simplest construction that provides the required behaviour is
4086 usually the most efficient. Jeffrey Friedl's book contains a lot of
4087 useful general discussion about optimizing regular expressions for
4088 efficient performance. This document contains a few observations about
4089 PCRE.
4090
4091 Using Unicode character properties (the \p, \P, and \X escapes) is
4092 slow, because PCRE has to scan a structure that contains data for over
4093 fifteen thousand characters whenever it needs a character's property.
4094 If you can find an alternative pattern that does not use character
4095 properties, it will probably be faster.
4096
4097 When a pattern begins with .* not in parentheses, or in parentheses
4098 that are not the subject of a backreference, and the PCRE_DOTALL option
4099 is set, the pattern is implicitly anchored by PCRE, since it can match
4100 only at the start of a subject string. However, if PCRE_DOTALL is not
4101 set, PCRE cannot make this optimization, because the . metacharacter
4102 does not then match a newline, and if the subject string contains new-
4103 lines, the pattern may match from the character immediately following
4104 one of them instead of from the very start. For example, the pattern
4105
4106 .*second
4107
4108 matches the subject "first\nand second" (where \n stands for a newline
4109 character), with the match starting at the seventh character. In order
4110 to do this, PCRE has to retry the match starting after every newline in
4111 the subject.
4112
4113 If you are using such a pattern with subject strings that do not con-
4114 tain newlines, the best performance is obtained by setting PCRE_DOTALL,
4115 or starting the pattern with ^.* or ^.*? to indicate explicit anchor-
4116 ing. That saves PCRE from having to scan along the subject looking for
4117 a newline to restart at.
4118
4119 Beware of patterns that contain nested indefinite repeats. These can
4120 take a long time to run when applied to a string that does not match.
4121 Consider the pattern fragment
4122
4123 (a+)*
4124
4125 This can match "aaaa" in 33 different ways, and this number increases
4126 very rapidly as the string gets longer. (The * repeat can match 0, 1,
4127 2, 3, or 4 times, and for each of those cases other than 0, the +
4128 repeats can match different numbers of times.) When the remainder of
4129 the pattern is such that the entire match is going to fail, PCRE has in
4130 principle to try every possible variation, and this can take an
4131 extremely long time.
4132
4133 An optimization catches some of the more simple cases such as
4134
4135 (a+)*b
4136
4137 where a literal character follows. Before embarking on the standard
4138 matching procedure, PCRE checks that there is a "b" later in the sub-
4139 ject string, and if there is not, it fails the match immediately. How-
4140 ever, when there is no following literal this optimization cannot be
4141 used. You can see the difference by comparing the behaviour of
4142
4143 (a+)*\d
4144
4145 with the pattern above. The former gives a failure almost instantly
4146 when applied to a whole line of "a" characters, whereas the latter
4147 takes an appreciable time with strings longer than about 20 characters.
4148
4149 In many cases, the solution to this kind of performance issue is to use
4150 an atomic group or a possessive quantifier.
4151
4152 Last updated: 28 February 2005
4153 Copyright (c) 1997-2005 University of Cambridge.
4154 ------------------------------------------------------------------------------
4155
4156
4157 PCREPOSIX(3) PCREPOSIX(3)
4158
4159
4160 NAME
4161 PCRE - Perl-compatible regular expressions.
4162
4163
4164 SYNOPSIS OF POSIX API
4165
4166 #include <pcreposix.h>
4167
4168 int regcomp(regex_t *preg, const char *pattern,
4169 int cflags);
4170
4171 int regexec(regex_t *preg, const char *string,
4172 size_t nmatch, regmatch_t pmatch[], int eflags);
4173
4174 size_t regerror(int errcode, const regex_t *preg,
4175 char *errbuf, size_t errbuf_size);
4176
4177 void regfree(regex_t *preg);
4178
4179
4180 DESCRIPTION
4181
4182 This set of functions provides a POSIX-style API to the PCRE regular
4183 expression package. See the pcreapi documentation for a description of
4184 PCRE's native API, which contains much additional functionality.
4185
4186 The functions described here are just wrapper functions that ultimately
4187 call the PCRE native API. Their prototypes are defined in the
4188 pcreposix.h header file, and on Unix systems the library itself is
4189 called pcreposix.a, so can be accessed by adding -lpcreposix to the
4190 command for linking an application that uses them. Because the POSIX
4191 functions call the native ones, it is also necessary to add -lpcre.
4192
4193 I have implemented only those option bits that can be reasonably mapped
4194 to PCRE native options. In addition, the options REG_EXTENDED and
4195 REG_NOSUB are defined with the value zero. They have no effect, but
4196 since programs that are written to the POSIX interface often use them,
4197 this makes it easier to slot in PCRE as a replacement library. Other
4198 POSIX options are not even defined.
4199
4200 When PCRE is called via these functions, it is only the API that is
4201 POSIX-like in style. The syntax and semantics of the regular expres-
4202 sions themselves are still those of Perl, subject to the setting of
4203 various PCRE options, as described below. "POSIX-like in style" means
4204 that the API approximates to the POSIX definition; it is not fully
4205 POSIX-compatible, and in multi-byte encoding domains it is probably
4206 even less compatible.
4207
4208 The header for these functions is supplied as pcreposix.h to avoid any
4209 potential clash with other POSIX libraries. It can, of course, be
4210 renamed or aliased as regex.h, which is the "correct" name. It provides
4211 two structure types, regex_t for compiled internal forms, and reg-
4212 match_t for returning captured substrings. It also defines some con-
4213 stants whose names start with "REG_"; these are used for setting
4214 options and identifying error codes.
4215
4216
4217 COMPILING A PATTERN
4218
4219 The function regcomp() is called to compile a pattern into an internal
4220 form. The pattern is a C string terminated by a binary zero, and is
4221 passed in the argument pattern. The preg argument is a pointer to a
4222 regex_t structure that is used as a base for storing information about
4223 the compiled expression.
4224
4225 The argument cflags is either zero, or contains one or more of the bits
4226 defined by the following macros:
4227
4228 REG_DOTALL
4229
4230 The PCRE_DOTALL option is set when the expression is passed for compi-
4231 lation to the native function. Note that REG_DOTALL is not part of the
4232 POSIX standard.
4233
4234 REG_ICASE
4235
4236 The PCRE_CASELESS option is set when the expression is passed for com-
4237 pilation to the native function.
4238
4239 REG_NEWLINE
4240
4241 The PCRE_MULTILINE option is set when the expression is passed for com-
4242 pilation to the native function. Note that this does not mimic the
4243 defined POSIX behaviour for REG_NEWLINE (see the following section).
4244
4245 In the absence of these flags, no options are passed to the native
4246 function. This means the the regex is compiled with PCRE default
4247 semantics. In particular, the way it handles newline characters in the
4248 subject string is the Perl way, not the POSIX way. Note that setting
4249 PCRE_MULTILINE has only some of the effects specified for REG_NEWLINE.
4250 It does not affect the way newlines are matched by . (they aren't) or
4251 by a negative class such as [^a] (they are).
4252
4253 The yield of regcomp() is zero on success, and non-zero otherwise. The
4254 preg structure is filled in on success, and one member of the structure
4255 is public: re_nsub contains the number of capturing subpatterns in the
4256 regular expression. Various error codes are defined in the header file.
4257
4258
4259 MATCHING NEWLINE CHARACTERS
4260
4261 This area is not simple, because POSIX and Perl take different views of
4262 things. It is not possible to get PCRE to obey POSIX semantics, but
4263 then PCRE was never intended to be a POSIX engine. The following table
4264 lists the different possibilities for matching newline characters in
4265 PCRE:
4266
4267 Default Change with
4268
4269 . matches newline no PCRE_DOTALL
4270 newline matches [^a] yes not changeable
4271 $ matches \n at end yes PCRE_DOLLARENDONLY
4272 $ matches \n in middle no PCRE_MULTILINE
4273 ^ matches \n in middle no PCRE_MULTILINE
4274
4275 This is the equivalent table for POSIX:
4276
4277 Default Change with
4278
4279 . matches newline yes REG_NEWLINE
4280 newline matches [^a] yes REG_NEWLINE
4281 $ matches \n at end no REG_NEWLINE
4282 $ matches \n in middle no REG_NEWLINE
4283 ^ matches \n in middle no REG_NEWLINE
4284
4285 PCRE's behaviour is the same as Perl's, except that there is no equiva-
4286 lent for PCRE_DOLLAR_ENDONLY in Perl. In both PCRE and Perl, there is
4287 no way to stop newline from matching [^a].
4288
4289 The default POSIX newline handling can be obtained by setting
4290 PCRE_DOTALL and PCRE_DOLLAR_ENDONLY, but there is no way to make PCRE
4291 behave exactly as for the REG_NEWLINE action.
4292
4293
4294 MATCHING A PATTERN
4295
4296 The function regexec() is called to match a compiled pattern preg
4297 against a given string, which is terminated by a zero byte, subject to
4298 the options in eflags. These can be:
4299
4300 REG_NOTBOL
4301
4302 The PCRE_NOTBOL option is set when calling the underlying PCRE matching
4303 function.
4304
4305 REG_NOTEOL
4306
4307 The PCRE_NOTEOL option is set when calling the underlying PCRE matching
4308 function.
4309
4310 The portion of the string that was matched, and also any captured sub-
4311 strings, are returned via the pmatch argument, which points to an array
4312 of nmatch structures of type regmatch_t, containing the members rm_so
4313 and rm_eo. These contain the offset to the first character of each sub-
4314 string and the offset to the first character after the end of each sub-
4315 string, respectively. The 0th element of the vector relates to the
4316 entire portion of string that was matched; subsequent elements relate
4317 to the capturing subpatterns of the regular expression. Unused entries
4318 in the array have both structure members set to -1.
4319
4320 A successful match yields a zero return; various error codes are
4321 defined in the header file, of which REG_NOMATCH is the "expected"
4322 failure code.
4323
4324
4325 ERROR MESSAGES
4326
4327 The regerror() function maps a non-zero errorcode from either regcomp()
4328 or regexec() to a printable message. If preg is not NULL, the error
4329 should have arisen from the use of that structure. A message terminated
4330 by a binary zero is placed in errbuf. The length of the message,
4331 including the zero, is limited to errbuf_size. The yield of the func-
4332 tion is the size of buffer needed to hold the whole message.
4333
4334
4335 MEMORY USAGE
4336
4337 Compiling a regular expression causes memory to be allocated and asso-
4338 ciated with the preg structure. The function regfree() frees all such
4339 memory, after which preg may no longer be used as a compiled expres-
4340 sion.
4341
4342
4343 AUTHOR
4344
4345 Philip Hazel
4346 University Computing Service,
4347 Cambridge CB2 3QG, England.
4348
4349 Last updated: 28 February 2005
4350 Copyright (c) 1997-2005 University of Cambridge.
4351 ------------------------------------------------------------------------------
4352
4353
4354 PCRECPP(3) PCRECPP(3)
4355
4356
4357 NAME
4358 PCRE - Perl-compatible regular expressions.
4359
4360
4361 SYNOPSIS OF C++ WRAPPER
4362
4363 #include <pcrecpp.h>
4364
4365
4366 DESCRIPTION
4367
4368 The C++ wrapper for PCRE was provided by Google Inc. Some additional
4369 functionality was added by Giuseppe Maxia. This brief man page was con-
4370 structed from the notes in the pcrecpp.h file, which should be con-
4371 sulted for further details.
4372
4373
4374 MATCHING INTERFACE
4375
4376 The "FullMatch" operation checks that supplied text matches a supplied
4377 pattern exactly. If pointer arguments are supplied, it copies matched
4378 sub-strings that match sub-patterns into them.
4379
4380 Example: successful match
4381 pcrecpp::RE re("h.*o");
4382 re.FullMatch("hello");
4383
4384 Example: unsuccessful match (requires full match):
4385 pcrecpp::RE re("e");
4386 !re.FullMatch("hello");
4387
4388 Example: creating a temporary RE object:
4389 pcrecpp::RE("h.*o").FullMatch("hello");
4390
4391 You can pass in a "const char*" or a "string" for "text". The examples
4392 below tend to use a const char*. You can, as in the different examples
4393 above, store the RE object explicitly in a variable or use a temporary
4394 RE object. The examples below use one mode or the other arbitrarily.
4395 Either could correctly be used for any of these examples.
4396
4397 You must supply extra pointer arguments to extract matched subpieces.
4398
4399 Example: extracts "ruby" into "s" and 1234 into "i"
4400 int i;
4401 string s;
4402 pcrecpp::RE re("(\\w+):(\\d+)");
4403 re.FullMatch("ruby:1234", &s, &i);
4404
4405 Example: does not try to extract any extra sub-patterns
4406 re.FullMatch("ruby:1234", &s);
4407
4408 Example: does not try to extract into NULL
4409 re.FullMatch("ruby:1234", NULL, &i);
4410
4411 Example: integer overflow causes failure
4412 !re.FullMatch("ruby:1234567891234", NULL, &i);
4413
4414 Example: fails because there aren't enough sub-patterns:
4415 !pcrecpp::RE("\\w+:\\d+").FullMatch("ruby:1234", &s);
4416
4417 Example: fails because string cannot be stored in integer
4418 !pcrecpp::RE("(.*)").FullMatch("ruby", &i);
4419
4420 The provided pointer arguments can be pointers to any scalar numeric
4421 type, or one of:
4422
4423 string (matched piece is copied to string)
4424 StringPiece (StringPiece is mutated to point to matched piece)
4425 T (where "bool T::ParseFrom(const char*, int)" exists)
4426 NULL (the corresponding matched sub-pattern is not copied)
4427
4428 The function returns true iff all of the following conditions are sat-
4429 isfied:
4430
4431 a. "text" matches "pattern" exactly;
4432
4433 b. The number of matched sub-patterns is >= number of supplied
4434 pointers;
4435
4436 c. The "i"th argument has a suitable type for holding the
4437 string captured as the "i"th sub-pattern. If you pass in
4438 NULL for the "i"th argument, or pass fewer arguments than
4439 number of sub-patterns, "i"th captured sub-pattern is
4440 ignored.
4441
4442 The matching interface supports at most 16 arguments per call. If you
4443 need more, consider using the more general interface
4444 pcrecpp::RE::DoMatch. See pcrecpp.h for the signature for DoMatch.
4445
4446
4447 PARTIAL MATCHES
4448
4449 You can use the "PartialMatch" operation when you want the pattern to
4450 match any substring of the text.
4451
4452 Example: simple search for a string:
4453 pcrecpp::RE("ell").PartialMatch("hello");
4454
4455 Example: find first number in a string:
4456 int number;
4457 pcrecpp::RE re("(\\d+)");
4458 re.PartialMatch("x*100 + 20", &number);
4459 assert(number == 100);
4460
4461
4462 UTF-8 AND THE MATCHING INTERFACE
4463
4464 By default, pattern and text are plain text, one byte per character.
4465 The UTF8 flag, passed to the constructor, causes both pattern and
4466 string to be treated as UTF-8 text, still a byte stream but potentially
4467 multiple bytes per character. In practice, the text is likelier to be
4468 UTF-8 than the pattern, but the match returned may depend on the UTF8
4469 flag, so always use it when matching UTF8 text. For example, "." will
4470 match one byte normally but with UTF8 set may match up to three bytes
4471 of a multi-byte character.
4472
4473 Example:
4474 pcrecpp::RE_Options options;
4475 options.set_utf8();
4476 pcrecpp::RE re(utf8_pattern, options);
4477 re.FullMatch(utf8_string);
4478
4479 Example: using the convenience function UTF8():
4480 pcrecpp::RE re(utf8_pattern, pcrecpp::UTF8());
4481 re.FullMatch(utf8_string);
4482
4483 NOTE: The UTF8 flag is ignored if pcre was not configured with the
4484 --enable-utf8 flag.
4485
4486
4487 PASSING MODIFIERS TO THE REGULAR EXPRESSION ENGINE
4488
4489 PCRE defines some modifiers to change the behavior of the regular
4490 expression engine. The C++ wrapper defines an auxiliary class,
4491 RE_Options, as a vehicle to pass such modifiers to a RE class. Cur-
4492 rently, the following modifiers are supported:
4493
4494 modifier description Perl corresponding
4495
4496 PCRE_CASELESS case insensitive match /i
4497 PCRE_MULTILINE multiple lines match /m
4498 PCRE_DOTALL dot matches newlines /s
4499 PCRE_DOLLAR_ENDONLY $ matches only at end N/A
4500 PCRE_EXTRA strict escape parsing N/A
4501 PCRE_EXTENDED ignore whitespaces /x
4502 PCRE_UTF8 handles UTF8 chars built-in
4503 PCRE_UNGREEDY reverses * and *? N/A
4504 PCRE_NO_AUTO_CAPTURE disables capturing parens N/A (*)
4505
4506 (*) Both Perl and PCRE allow non capturing parentheses by means of the
4507 "?:" modifier within the pattern itself. e.g. (?:ab|cd) does not cap-
4508 ture, while (ab|cd) does.
4509
4510 For a full account on how each modifier works, please check the PCRE
4511 API reference page.
4512
4513 For each modifier, there are two member functions whose name is made
4514 out of the modifier in lowercase, without the "PCRE_" prefix. For
4515 instance, PCRE_CASELESS is handled by
4516
4517 bool caseless()
4518
4519 which returns true if the modifier is set, and
4520
4521 RE_Options & set_caseless(bool)
4522
4523 which sets or unsets the modifier. Moreover, PCRE_CONFIG_MATCH_LIMIT
4524 can be accessed through the set_match_limit() and match_limit() member
4525 functions. Setting match_limit to a non-zero value will limit the exe-
4526 cution of pcre to keep it from doing bad things like blowing the stack
4527 or taking an eternity to return a result. A value of 5000 is good
4528 enough to stop stack blowup in a 2MB thread stack. Setting match_limit
4529 to zero disables match limiting.
4530
4531 Normally, to pass one or more modifiers to a RE class, you declare a
4532 RE_Options object, set the appropriate options, and pass this object to
4533 a RE constructor. Example:
4534
4535 RE_options opt;
4536 opt.set_caseless(true);
4537 if (RE("HELLO", opt).PartialMatch("hello world")) ...
4538
4539 RE_options has two constructors. The default constructor takes no argu-
4540 ments and creates a set of flags that are off by default. The optional
4541 parameter option_flags is to facilitate transfer of legacy code from C
4542 programs. This lets you do
4543
4544 RE(pattern,
4545 RE_Options(PCRE_CASELESS|PCRE_MULTILINE)).PartialMatch(str);
4546
4547 However, new code is better off doing
4548
4549 RE(pattern,
4550 RE_Options().set_caseless(true).set_multiline(true))
4551 .PartialMatch(str);
4552
4553 If you are going to pass one of the most used modifiers, there are some
4554 convenience functions that return a RE_Options class with the appropri-
4555 ate modifier already set: CASELESS(), UTF8(), MULTILINE(), DOTALL(),
4556 and EXTENDED().
4557
4558 If you need to set several options at once, and you don't want to go
4559 through the pains of declaring a RE_Options object and setting several
4560 options, there is a parallel method that give you such ability on the
4561 fly. You can concatenate several set_xxxxx() member functions, since
4562 each of them returns a reference to its class object. For example, to
4563 pass PCRE_CASELESS, PCRE_EXTENDED, and PCRE_MULTILINE to a RE with one
4564 statement, you may write:
4565
4566 RE(" ^ xyz \\s+ .* blah$",
4567 RE_Options()
4568 .set_caseless(true)
4569 .set_extended(true)
4570 .set_multiline(true)).PartialMatch(sometext);
4571
4572
4573 SCANNING TEXT INCREMENTALLY
4574
4575 The "Consume" operation may be useful if you want to repeatedly match
4576 regular expressions at the front of a string and skip over them as they
4577 match. This requires use of the "StringPiece" type, which represents a
4578 sub-range of a real string. Like RE, StringPiece is defined in the
4579 pcrecpp namespace.
4580
4581 Example: read lines of the form "var = value" from a string.
4582 string contents = ...; // Fill string somehow
4583 pcrecpp::StringPiece input(contents); // Wrap in a StringPiece
4584
4585 string var;
4586 int value;
4587 pcrecpp::RE re("(\\w+) = (\\d+)\n");
4588 while (re.Consume(&input, &var, &value)) {
4589 ...;
4590 }
4591
4592 Each successful call to "Consume" will set "var/value", and also
4593 advance "input" so it points past the matched text.
4594
4595 The "FindAndConsume" operation is similar to "Consume" but does not
4596 anchor your match at the beginning of the string. For example, you
4597 could extract all words from a string by repeatedly calling
4598
4599 pcrecpp::RE("(\\w+)").FindAndConsume(&input, &word)
4600
4601
4602 PARSING HEX/OCTAL/C-RADIX NUMBERS
4603
4604 By default, if you pass a pointer to a numeric value, the corresponding
4605 text is interpreted as a base-10 number. You can instead wrap the
4606 pointer with a call to one of the operators Hex(), Octal(), or CRadix()
4607 to interpret the text in another base. The CRadix operator interprets
4608 C-style "0" (base-8) and "0x" (base-16) prefixes, but defaults to
4609 base-10.
4610
4611 Example:
4612 int a, b, c, d;
4613 pcrecpp::RE re("(.*) (.*) (.*) (.*)");
4614 re.FullMatch("100 40 0100 0x40",
4615 pcrecpp::Octal(&a), pcrecpp::Hex(&b),
4616 pcrecpp::CRadix(&c), pcrecpp::CRadix(&d));
4617
4618 will leave 64 in a, b, c, and d.
4619
4620
4621 REPLACING PARTS OF STRINGS
4622
4623 You can replace the first match of "pattern" in "str" with "rewrite".
4624 Within "rewrite", backslash-escaped digits (\1 to \9) can be used to
4625 insert text matching corresponding parenthesized group from the pat-
4626 tern. \0 in "rewrite" refers to the entire matching text. For example:
4627
4628 string s = "yabba dabba doo";
4629 pcrecpp::RE("b+").Replace("d", &s);
4630
4631 will leave "s" containing "yada dabba doo". The result is true if the
4632 pattern matches and a replacement occurs, false otherwise.
4633
4634 GlobalReplace is like Replace except that it replaces all occurrences
4635 of the pattern in the string with the rewrite. Replacements are not
4636 subject to re-matching. For example:
4637
4638 string s = "yabba dabba doo";
4639 pcrecpp::RE("b+").GlobalReplace("d", &s);
4640
4641 will leave "s" containing "yada dada doo". It returns the number of
4642 replacements made.
4643
4644 Extract is like Replace, except that if the pattern matches, "rewrite"
4645 is copied into "out" (an additional argument) with substitutions. The
4646 non-matching portions of "text" are ignored. Returns true iff a match
4647 occurred and the extraction happened successfully; if no match occurs,
4648 the string is left unaffected.
4649
4650
4651 AUTHOR
4652
4653 The C++ wrapper was contributed by Google Inc.
4654 Copyright (c) 2005 Google Inc.
4655 ------------------------------------------------------------------------------
4656
4657
4658 PCRESAMPLE(3) PCRESAMPLE(3)
4659
4660
4661 NAME
4662 PCRE - Perl-compatible regular expressions
4663
4664
4665 PCRE SAMPLE PROGRAM
4666
4667 A simple, complete demonstration program, to get you started with using
4668 PCRE, is supplied in the file pcredemo.c in the PCRE distribution.
4669
4670 The program compiles the regular expression that is its first argument,
4671 and matches it against the subject string in its second argument. No
4672 PCRE options are set, and default character tables are used. If match-
4673 ing succeeds, the program outputs the portion of the subject that
4674 matched, together with the contents of any captured substrings.
4675
4676 If the -g option is given on the command line, the program then goes on
4677 to check for further matches of the same regular expression in the same
4678 subject string. The logic is a little bit tricky because of the possi-
4679 bility of matching an empty string. Comments in the code explain what
4680 is going on.
4681
4682 If PCRE is installed in the standard include and library directories
4683 for your system, you should be able to compile the demonstration pro-
4684 gram using this command:
4685
4686 gcc -o pcredemo pcredemo.c -lpcre
4687
4688 If PCRE is installed elsewhere, you may need to add additional options
4689 to the command line. For example, on a Unix-like system that has PCRE
4690 installed in /usr/local, you can compile the demonstration program
4691 using a command like this:
4692
4693 gcc -o pcredemo -I/usr/local/include pcredemo.c \
4694 -L/usr/local/lib -lpcre
4695
4696 Once you have compiled the demonstration program, you can run simple
4697 tests like this:
4698
4699 ./pcredemo 'cat|dog' 'the cat sat on the mat'
4700 ./pcredemo -g 'cat|dog' 'the dog sat on the cat'
4701
4702 Note that there is a much more comprehensive test program, called
4703 pcretest, which supports many more facilities for testing regular
4704 expressions and the PCRE library. The pcredemo program is provided as a
4705 simple coding example.
4706
4707 On some operating systems (e.g. Solaris), when PCRE is not installed in
4708 the standard library directory, you may get an error like this when you
4709 try to run pcredemo:
4710
4711 ld.so.1: a.out: fatal: libpcre.so.0: open failed: No such file or
4712 directory
4713
4714 This is caused by the way shared library support works on those sys-
4715 tems. You need to add
4716
4717 -R/usr/local/lib
4718
4719 (for example) to the compile command to get round this problem.
4720
4721 Last updated: 09 September 2004
4722 Copyright (c) 1997-2004 University of Cambridge.
4723 ------------------------------------------------------------------------------

  ViewVC Help
Powered by ViewVC 1.1.5