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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 261 - (show annotations)
Fri Sep 21 08:37:48 2007 UTC (13 years, 1 month ago) by ph10
File MIME type: text/plain
File size: 302499 byte(s)
Final doc updates and file tidies for 7.4.
1 -----------------------------------------------------------------------------
2 This file contains a concatenation of the PCRE man pages, converted to plain
3 text format for ease of searching with a text editor, or for use on systems
4 that do not have a man page processor. The small individual files that give
5 synopses of each function in the library have not been included. There are
6 separate text files for the pcregrep and pcretest commands.
7 -----------------------------------------------------------------------------
8
9
10 PCRE(3) PCRE(3)
11
12
13 NAME
14 PCRE - Perl-compatible regular expressions
15
16
17 INTRODUCTION
18
19 The PCRE library is a set of functions that implement regular expres-
20 sion pattern matching using the same syntax and semantics as Perl, with
21 just a few differences. (Certain features that appeared in Python and
22 PCRE before they appeared in Perl are also available using the Python
23 syntax.)
24
25 The current implementation of PCRE (release 7.x) corresponds approxi-
26 mately with Perl 5.10, including support for UTF-8 encoded strings and
27 Unicode general category properties. However, UTF-8 and Unicode support
28 has to be explicitly enabled; it is not the default. The Unicode tables
29 correspond to Unicode release 5.0.0.
30
31 In addition to the Perl-compatible matching function, PCRE contains an
32 alternative matching function that matches the same compiled patterns
33 in a different way. In certain circumstances, the alternative function
34 has some advantages. For a discussion of the two matching algorithms,
35 see the pcrematching page.
36
37 PCRE is written in C and released as a C library. A number of people
38 have written wrappers and interfaces of various kinds. In particular,
39 Google Inc. have provided a comprehensive C++ wrapper. This is now
40 included as part of the PCRE distribution. The pcrecpp page has details
41 of this interface. Other people's contributions can be found in the
42 Contrib directory at the primary FTP site, which is:
43
44 ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre
45
46 Details of exactly which Perl regular expression features are and are
47 not supported by PCRE are given in separate documents. See the pcrepat-
48 tern and pcrecompat pages. There is a syntax summary in the pcresyntax
49 page.
50
51 Some features of PCRE can be included, excluded, or changed when the
52 library is built. The pcre_config() function makes it possible for a
53 client to discover which features are available. The features them-
54 selves are described in the pcrebuild page. Documentation about build-
55 ing PCRE for various operating systems can be found in the README file
56 in the source distribution.
57
58 The library contains a number of undocumented internal functions and
59 data tables that are used by more than one of the exported external
60 functions, but which are not intended for use by external callers.
61 Their names all begin with "_pcre_", which hopefully will not provoke
62 any name clashes. In some environments, it is possible to control which
63 external symbols are exported when a shared library is built, and in
64 these cases the undocumented symbols are not exported.
65
66
67 USER DOCUMENTATION
68
69 The user documentation for PCRE comprises a number of different sec-
70 tions. In the "man" format, each of these is a separate "man page". In
71 the HTML format, each is a separate page, linked from the index page.
72 In the plain text format, all the sections are concatenated, for ease
73 of searching. The sections are as follows:
74
75 pcre this document
76 pcre-config show PCRE installation configuration information
77 pcreapi details of PCRE's native C API
78 pcrebuild options for building PCRE
79 pcrecallout details of the callout feature
80 pcrecompat discussion of Perl compatibility
81 pcrecpp details of the C++ wrapper
82 pcregrep description of the pcregrep command
83 pcrematching discussion of the two matching algorithms
84 pcrepartial details of the partial matching facility
85 pcrepattern syntax and semantics of supported
86 regular expressions
87 pcresyntax quick syntax reference
88 pcreperform discussion of performance issues
89 pcreposix the POSIX-compatible C API
90 pcreprecompile details of saving and re-using precompiled patterns
91 pcresample discussion of the sample program
92 pcrestack discussion of stack usage
93 pcretest description of the pcretest testing command
94
95 In addition, in the "man" and HTML formats, there is a short page for
96 each C library function, listing its arguments and results.
97
98
99 LIMITATIONS
100
101 There are some size limitations in PCRE but it is hoped that they will
102 never in practice be relevant.
103
104 The maximum length of a compiled pattern is 65539 (sic) bytes if PCRE
105 is compiled with the default internal linkage size of 2. If you want to
106 process regular expressions that are truly enormous, you can compile
107 PCRE with an internal linkage size of 3 or 4 (see the README file in
108 the source distribution and the pcrebuild documentation for details).
109 In these cases the limit is substantially larger. However, the speed
110 of execution is slower.
111
112 All values in repeating quantifiers must be less than 65536.
113
114 There is no limit to the number of parenthesized subpatterns, but there
115 can be no more than 65535 capturing subpatterns.
116
117 The maximum length of name for a named subpattern is 32 characters, and
118 the maximum number of named subpatterns is 10000.
119
120 The maximum length of a subject string is the largest positive number
121 that an integer variable can hold. However, when using the traditional
122 matching function, PCRE uses recursion to handle subpatterns and indef-
123 inite repetition. This means that the available stack space may limit
124 the size of a subject string that can be processed by certain patterns.
125 For a discussion of stack issues, see the pcrestack documentation.
126
127
128 UTF-8 AND UNICODE PROPERTY SUPPORT
129
130 From release 3.3, PCRE has had some support for character strings
131 encoded in the UTF-8 format. For release 4.0 this was greatly extended
132 to cover most common requirements, and in release 5.0 additional sup-
133 port for Unicode general category properties was added.
134
135 In order process UTF-8 strings, you must build PCRE to include UTF-8
136 support in the code, and, in addition, you must call pcre_compile()
137 with the PCRE_UTF8 option flag. When you do this, both the pattern and
138 any subject strings that are matched against it are treated as UTF-8
139 strings instead of just strings of bytes.
140
141 If you compile PCRE with UTF-8 support, but do not use it at run time,
142 the library will be a bit bigger, but the additional run time overhead
143 is limited to testing the PCRE_UTF8 flag occasionally, so should not be
144 very big.
145
146 If PCRE is built with Unicode character property support (which implies
147 UTF-8 support), the escape sequences \p{..}, \P{..}, and \X are sup-
148 ported. The available properties that can be tested are limited to the
149 general category properties such as Lu for an upper case letter or Nd
150 for a decimal number, the Unicode script names such as Arabic or Han,
151 and the derived properties Any and L&. A full list is given in the
152 pcrepattern documentation. Only the short names for properties are sup-
153 ported. For example, \p{L} matches a letter. Its Perl synonym, \p{Let-
154 ter}, is not supported. Furthermore, in Perl, many properties may
155 optionally be prefixed by "Is", for compatibility with Perl 5.6. PCRE
156 does not support this.
157
158 Validity of UTF-8 strings
159
160 When you set the PCRE_UTF8 flag, the strings passed as patterns and
161 subjects are (by default) checked for validity on entry to the relevant
162 functions. From release 7.3 of PCRE, the check is according the rules
163 of RFC 3629, which are themselves derived from the Unicode specifica-
164 tion. Earlier releases of PCRE followed the rules of RFC 2279, which
165 allows the full range of 31-bit values (0 to 0x7FFFFFFF). The current
166 check allows only values in the range U+0 to U+10FFFF, excluding U+D800
167 to U+DFFF.
168
169 The excluded code points are the "Low Surrogate Area" of Unicode, of
170 which the Unicode Standard says this: "The Low Surrogate Area does not
171 contain any character assignments, consequently no character code
172 charts or namelists are provided for this area. Surrogates are reserved
173 for use with UTF-16 and then must be used in pairs." The code points
174 that are encoded by UTF-16 pairs are available as independent code
175 points in the UTF-8 encoding. (In other words, the whole surrogate
176 thing is a fudge for UTF-16 which unfortunately messes up UTF-8.)
177
178 If an invalid UTF-8 string is passed to PCRE, an error return
179 (PCRE_ERROR_BADUTF8) is given. In some situations, you may already know
180 that your strings are valid, and therefore want to skip these checks in
181 order to improve performance. If you set the PCRE_NO_UTF8_CHECK flag at
182 compile time or at run time, PCRE assumes that the pattern or subject
183 it is given (respectively) contains only valid UTF-8 codes. In this
184 case, it does not diagnose an invalid UTF-8 string.
185
186 If you pass an invalid UTF-8 string when PCRE_NO_UTF8_CHECK is set,
187 what happens depends on why the string is invalid. If the string con-
188 forms to the "old" definition of UTF-8 (RFC 2279), it is processed as a
189 string of characters in the range 0 to 0x7FFFFFFF. In other words,
190 apart from the initial validity test, PCRE (when in UTF-8 mode) handles
191 strings according to the more liberal rules of RFC 2279. However, if
192 the string does not even conform to RFC 2279, the result is undefined.
193 Your program may crash.
194
195 If you want to process strings of values in the full range 0 to
196 0x7FFFFFFF, encoded in a UTF-8-like manner as per the old RFC, you can
197 set PCRE_NO_UTF8_CHECK to bypass the more restrictive test. However, in
198 this situation, you will have to apply your own validity check.
199
200 General comments about UTF-8 mode
201
202 1. An unbraced hexadecimal escape sequence (such as \xb3) matches a
203 two-byte UTF-8 character if the value is greater than 127.
204
205 2. Octal numbers up to \777 are recognized, and match two-byte UTF-8
206 characters for values greater than \177.
207
208 3. Repeat quantifiers apply to complete UTF-8 characters, not to indi-
209 vidual bytes, for example: \x{100}{3}.
210
211 4. The dot metacharacter matches one UTF-8 character instead of a sin-
212 gle byte.
213
214 5. The escape sequence \C can be used to match a single byte in UTF-8
215 mode, but its use can lead to some strange effects. This facility is
216 not available in the alternative matching function, pcre_dfa_exec().
217
218 6. The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly
219 test characters of any code value, but the characters that PCRE recog-
220 nizes as digits, spaces, or word characters remain the same set as
221 before, all with values less than 256. This remains true even when PCRE
222 includes Unicode property support, because to do otherwise would slow
223 down PCRE in many common cases. If you really want to test for a wider
224 sense of, say, "digit", you must use Unicode property tests such as
225 \p{Nd}.
226
227 7. Similarly, characters that match the POSIX named character classes
228 are all low-valued characters.
229
230 8. However, the Perl 5.10 horizontal and vertical whitespace matching
231 escapes (\h, \H, \v, and \V) do match all the appropriate Unicode char-
232 acters.
233
234 9. Case-insensitive matching applies only to characters whose values
235 are less than 128, unless PCRE is built with Unicode property support.
236 Even when Unicode property support is available, PCRE still uses its
237 own character tables when checking the case of low-valued characters,
238 so as not to degrade performance. The Unicode property information is
239 used only for characters with higher values. Even when Unicode property
240 support is available, PCRE supports case-insensitive matching only when
241 there is a one-to-one mapping between a letter's cases. There are a
242 small number of many-to-one mappings in Unicode; these are not sup-
243 ported by PCRE.
244
245
246 AUTHOR
247
248 Philip Hazel
249 University Computing Service
250 Cambridge CB2 3QH, England.
251
252 Putting an actual email address here seems to have been a spam magnet,
253 so I've taken it away. If you want to email me, use my two initials,
254 followed by the two digits 10, at the domain cam.ac.uk.
255
256
257 REVISION
258
259 Last updated: 09 August 2007
260 Copyright (c) 1997-2007 University of Cambridge.
261 ------------------------------------------------------------------------------
262
263
264 PCREBUILD(3) PCREBUILD(3)
265
266
267 NAME
268 PCRE - Perl-compatible regular expressions
269
270
271 PCRE BUILD-TIME OPTIONS
272
273 This document describes the optional features of PCRE that can be
274 selected when the library is compiled. It assumes use of the configure
275 script, where the optional features are selected or deselected by pro-
276 viding options to configure before running the make command. However,
277 the same options can be selected in both Unix-like and non-Unix-like
278 environments using the GUI facility of CMakeSetup if you are using
279 CMake instead of configure to build PCRE.
280
281 The complete list of options for configure (which includes the standard
282 ones such as the selection of the installation directory) can be
283 obtained by running
284
285 ./configure --help
286
287 The following sections include descriptions of options whose names
288 begin with --enable or --disable. These settings specify changes to the
289 defaults for the configure command. Because of the way that configure
290 works, --enable and --disable always come in pairs, so the complemen-
291 tary option always exists as well, but as it specifies the default, it
292 is not described.
293
294
295 C++ SUPPORT
296
297 By default, the configure script will search for a C++ compiler and C++
298 header files. If it finds them, it automatically builds the C++ wrapper
299 library for PCRE. You can disable this by adding
300
301 --disable-cpp
302
303 to the configure command.
304
305
306 UTF-8 SUPPORT
307
308 To build PCRE with support for UTF-8 character strings, add
309
310 --enable-utf8
311
312 to the configure command. Of itself, this does not make PCRE treat
313 strings as UTF-8. As well as compiling PCRE with this option, you also
314 have have to set the PCRE_UTF8 option when you call the pcre_compile()
315 function.
316
317
318 UNICODE CHARACTER PROPERTY SUPPORT
319
320 UTF-8 support allows PCRE to process character values greater than 255
321 in the strings that it handles. On its own, however, it does not pro-
322 vide any facilities for accessing the properties of such characters. If
323 you want to be able to use the pattern escapes \P, \p, and \X, which
324 refer to Unicode character properties, you must add
325
326 --enable-unicode-properties
327
328 to the configure command. This implies UTF-8 support, even if you have
329 not explicitly requested it.
330
331 Including Unicode property support adds around 30K of tables to the
332 PCRE library. Only the general category properties such as Lu and Nd
333 are supported. Details are given in the pcrepattern documentation.
334
335
336 CODE VALUE OF NEWLINE
337
338 By default, PCRE interprets character 10 (linefeed, LF) as indicating
339 the end of a line. This is the normal newline character on Unix-like
340 systems. You can compile PCRE to use character 13 (carriage return, CR)
341 instead, by adding
342
343 --enable-newline-is-cr
344
345 to the configure command. There is also a --enable-newline-is-lf
346 option, which explicitly specifies linefeed as the newline character.
347
348 Alternatively, you can specify that line endings are to be indicated by
349 the two character sequence CRLF. If you want this, add
350
351 --enable-newline-is-crlf
352
353 to the configure command. There is a fourth option, specified by
354
355 --enable-newline-is-anycrlf
356
357 which causes PCRE to recognize any of the three sequences CR, LF, or
358 CRLF as indicating a line ending. Finally, a fifth option, specified by
359
360 --enable-newline-is-any
361
362 causes PCRE to recognize any Unicode newline sequence.
363
364 Whatever line ending convention is selected when PCRE is built can be
365 overridden when the library functions are called. At build time it is
366 conventional to use the standard for your operating system.
367
368
369 WHAT \R MATCHES
370
371 By default, the sequence \R in a pattern matches any Unicode newline
372 sequence, whatever has been selected as the line ending sequence. If
373 you specify
374
375 --enable-bsr-anycrlf
376
377 the default is changed so that \R matches only CR, LF, or CRLF. What-
378 ever is selected when PCRE is built can be overridden when the library
379 functions are called.
380
381
382 BUILDING SHARED AND STATIC LIBRARIES
383
384 The PCRE building process uses libtool to build both shared and static
385 Unix libraries by default. You can suppress one of these by adding one
386 of
387
388 --disable-shared
389 --disable-static
390
391 to the configure command, as required.
392
393
394 POSIX MALLOC USAGE
395
396 When PCRE is called through the POSIX interface (see the pcreposix doc-
397 umentation), additional working storage is required for holding the
398 pointers to capturing substrings, because PCRE requires three integers
399 per substring, whereas the POSIX interface provides only two. If the
400 number of expected substrings is small, the wrapper function uses space
401 on the stack, because this is faster than using malloc() for each call.
402 The default threshold above which the stack is no longer used is 10; it
403 can be changed by adding a setting such as
404
405 --with-posix-malloc-threshold=20
406
407 to the configure command.
408
409
410 HANDLING VERY LARGE PATTERNS
411
412 Within a compiled pattern, offset values are used to point from one
413 part to another (for example, from an opening parenthesis to an alter-
414 nation metacharacter). By default, two-byte values are used for these
415 offsets, leading to a maximum size for a compiled pattern of around
416 64K. This is sufficient to handle all but the most gigantic patterns.
417 Nevertheless, some people do want to process enormous patterns, so it
418 is possible to compile PCRE to use three-byte or four-byte offsets by
419 adding a setting such as
420
421 --with-link-size=3
422
423 to the configure command. The value given must be 2, 3, or 4. Using
424 longer offsets slows down the operation of PCRE because it has to load
425 additional bytes when handling them.
426
427
428 AVOIDING EXCESSIVE STACK USAGE
429
430 When matching with the pcre_exec() function, PCRE implements backtrack-
431 ing by making recursive calls to an internal function called match().
432 In environments where the size of the stack is limited, this can se-
433 verely limit PCRE's operation. (The Unix environment does not usually
434 suffer from this problem, but it may sometimes be necessary to increase
435 the maximum stack size. There is a discussion in the pcrestack docu-
436 mentation.) An alternative approach to recursion that uses memory from
437 the heap to remember data, instead of using recursive function calls,
438 has been implemented to work round the problem of limited stack size.
439 If you want to build a version of PCRE that works this way, add
440
441 --disable-stack-for-recursion
442
443 to the configure command. With this configuration, PCRE will use the
444 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
445 ment functions. By default these point to malloc() and free(), but you
446 can replace the pointers so that your own functions are used.
447
448 Separate functions are provided rather than using pcre_malloc and
449 pcre_free because the usage is very predictable: the block sizes
450 requested are always the same, and the blocks are always freed in
451 reverse order. A calling program might be able to implement optimized
452 functions that perform better than malloc() and free(). PCRE runs
453 noticeably more slowly when built in this way. This option affects only
454 the pcre_exec() function; it is not relevant for the the
455 pcre_dfa_exec() function.
456
457
458 LIMITING PCRE RESOURCE USAGE
459
460 Internally, PCRE has a function called match(), which it calls repeat-
461 edly (sometimes recursively) when matching a pattern with the
462 pcre_exec() function. By controlling the maximum number of times this
463 function may be called during a single matching operation, a limit can
464 be placed on the resources used by a single call to pcre_exec(). The
465 limit can be changed at run time, as described in the pcreapi documen-
466 tation. The default is 10 million, but this can be changed by adding a
467 setting such as
468
469 --with-match-limit=500000
470
471 to the configure command. This setting has no effect on the
472 pcre_dfa_exec() matching function.
473
474 In some environments it is desirable to limit the depth of recursive
475 calls of match() more strictly than the total number of calls, in order
476 to restrict the maximum amount of stack (or heap, if --disable-stack-
477 for-recursion is specified) that is used. A second limit controls this;
478 it defaults to the value that is set for --with-match-limit, which
479 imposes no additional constraints. However, you can set a lower limit
480 by adding, for example,
481
482 --with-match-limit-recursion=10000
483
484 to the configure command. This value can also be overridden at run
485 time.
486
487
488 CREATING CHARACTER TABLES AT BUILD TIME
489
490 PCRE uses fixed tables for processing characters whose code values are
491 less than 256. By default, PCRE is built with a set of tables that are
492 distributed in the file pcre_chartables.c.dist. These tables are for
493 ASCII codes only. If you add
494
495 --enable-rebuild-chartables
496
497 to the configure command, the distributed tables are no longer used.
498 Instead, a program called dftables is compiled and run. This outputs
499 the source for new set of tables, created in the default locale of your
500 C runtime system. (This method of replacing the tables does not work if
501 you are cross compiling, because dftables is run on the local host. If
502 you need to create alternative tables when cross compiling, you will
503 have to do so "by hand".)
504
505
506 USING EBCDIC CODE
507
508 PCRE assumes by default that it will run in an environment where the
509 character code is ASCII (or Unicode, which is a superset of ASCII).
510 This is the case for most computer operating systems. PCRE can, how-
511 ever, be compiled to run in an EBCDIC environment by adding
512
513 --enable-ebcdic
514
515 to the configure command. This setting implies --enable-rebuild-charta-
516 bles. You should only use it if you know that you are in an EBCDIC
517 environment (for example, an IBM mainframe operating system).
518
519
520 SEE ALSO
521
522 pcreapi(3), pcre_config(3).
523
524
525 AUTHOR
526
527 Philip Hazel
528 University Computing Service
529 Cambridge CB2 3QH, England.
530
531
532 REVISION
533
534 Last updated: 21 September 2007
535 Copyright (c) 1997-2007 University of Cambridge.
536 ------------------------------------------------------------------------------
537
538
539 PCREMATCHING(3) PCREMATCHING(3)
540
541
542 NAME
543 PCRE - Perl-compatible regular expressions
544
545
546 PCRE MATCHING ALGORITHMS
547
548 This document describes the two different algorithms that are available
549 in PCRE for matching a compiled regular expression against a given sub-
550 ject string. The "standard" algorithm is the one provided by the
551 pcre_exec() function. This works in the same was as Perl's matching
552 function, and provides a Perl-compatible matching operation.
553
554 An alternative algorithm is provided by the pcre_dfa_exec() function;
555 this operates in a different way, and is not Perl-compatible. It has
556 advantages and disadvantages compared with the standard algorithm, and
557 these are described below.
558
559 When there is only one possible way in which a given subject string can
560 match a pattern, the two algorithms give the same answer. A difference
561 arises, however, when there are multiple possibilities. For example, if
562 the pattern
563
564 ^<.*>
565
566 is matched against the string
567
568 <something> <something else> <something further>
569
570 there are three possible answers. The standard algorithm finds only one
571 of them, whereas the alternative algorithm finds all three.
572
573
574 REGULAR EXPRESSIONS AS TREES
575
576 The set of strings that are matched by a regular expression can be rep-
577 resented as a tree structure. An unlimited repetition in the pattern
578 makes the tree of infinite size, but it is still a tree. Matching the
579 pattern to a given subject string (from a given starting point) can be
580 thought of as a search of the tree. There are two ways to search a
581 tree: depth-first and breadth-first, and these correspond to the two
582 matching algorithms provided by PCRE.
583
584
585 THE STANDARD MATCHING ALGORITHM
586
587 In the terminology of Jeffrey Friedl's book "Mastering Regular Expres-
588 sions", the standard algorithm is an "NFA algorithm". It conducts a
589 depth-first search of the pattern tree. That is, it proceeds along a
590 single path through the tree, checking that the subject matches what is
591 required. When there is a mismatch, the algorithm tries any alterna-
592 tives at the current point, and if they all fail, it backs up to the
593 previous branch point in the tree, and tries the next alternative
594 branch at that level. This often involves backing up (moving to the
595 left) in the subject string as well. The order in which repetition
596 branches are tried is controlled by the greedy or ungreedy nature of
597 the quantifier.
598
599 If a leaf node is reached, a matching string has been found, and at
600 that point the algorithm stops. Thus, if there is more than one possi-
601 ble match, this algorithm returns the first one that it finds. Whether
602 this is the shortest, the longest, or some intermediate length depends
603 on the way the greedy and ungreedy repetition quantifiers are specified
604 in the pattern.
605
606 Because it ends up with a single path through the tree, it is rela-
607 tively straightforward for this algorithm to keep track of the sub-
608 strings that are matched by portions of the pattern in parentheses.
609 This provides support for capturing parentheses and back references.
610
611
612 THE ALTERNATIVE MATCHING ALGORITHM
613
614 This algorithm conducts a breadth-first search of the tree. Starting
615 from the first matching point in the subject, it scans the subject
616 string from left to right, once, character by character, and as it does
617 this, it remembers all the paths through the tree that represent valid
618 matches. In Friedl's terminology, this is a kind of "DFA algorithm",
619 though it is not implemented as a traditional finite state machine (it
620 keeps multiple states active simultaneously).
621
622 The scan continues until either the end of the subject is reached, or
623 there are no more unterminated paths. At this point, terminated paths
624 represent the different matching possibilities (if there are none, the
625 match has failed). Thus, if there is more than one possible match,
626 this algorithm finds all of them, and in particular, it finds the long-
627 est. In PCRE, there is an option to stop the algorithm after the first
628 match (which is necessarily the shortest) has been found.
629
630 Note that all the matches that are found start at the same point in the
631 subject. If the pattern
632
633 cat(er(pillar)?)
634
635 is matched against the string "the caterpillar catchment", the result
636 will be the three strings "cat", "cater", and "caterpillar" that start
637 at the fourth character of the subject. The algorithm does not automat-
638 ically move on to find matches that start at later positions.
639
640 There are a number of features of PCRE regular expressions that are not
641 supported by the alternative matching algorithm. They are as follows:
642
643 1. Because the algorithm finds all possible matches, the greedy or
644 ungreedy nature of repetition quantifiers is not relevant. Greedy and
645 ungreedy quantifiers are treated in exactly the same way. However, pos-
646 sessive quantifiers can make a difference when what follows could also
647 match what is quantified, for example in a pattern like this:
648
649 ^a++\w!
650
651 This pattern matches "aaab!" but not "aaa!", which would be matched by
652 a non-possessive quantifier. Similarly, if an atomic group is present,
653 it is matched as if it were a standalone pattern at the current point,
654 and the longest match is then "locked in" for the rest of the overall
655 pattern.
656
657 2. When dealing with multiple paths through the tree simultaneously, it
658 is not straightforward to keep track of captured substrings for the
659 different matching possibilities, and PCRE's implementation of this
660 algorithm does not attempt to do this. This means that no captured sub-
661 strings are available.
662
663 3. Because no substrings are captured, back references within the pat-
664 tern are not supported, and cause errors if encountered.
665
666 4. For the same reason, conditional expressions that use a backrefer-
667 ence as the condition or test for a specific group recursion are not
668 supported.
669
670 5. Because many paths through the tree may be active, the \K escape
671 sequence, which resets the start of the match when encountered (but may
672 be on some paths and not on others), is not supported. It causes an
673 error if encountered.
674
675 6. Callouts are supported, but the value of the capture_top field is
676 always 1, and the value of the capture_last field is always -1.
677
678 7. The \C escape sequence, which (in the standard algorithm) matches a
679 single byte, even in UTF-8 mode, is not supported because the alterna-
680 tive algorithm moves through the subject string one character at a
681 time, for all active paths through the tree.
682
683 8. None of the backtracking control verbs such as (*PRUNE) are sup-
684 ported.
685
686
687 ADVANTAGES OF THE ALTERNATIVE ALGORITHM
688
689 Using the alternative matching algorithm provides the following advan-
690 tages:
691
692 1. All possible matches (at a single point in the subject) are automat-
693 ically found, and in particular, the longest match is found. To find
694 more than one match using the standard algorithm, you have to do kludgy
695 things with callouts.
696
697 2. There is much better support for partial matching. The restrictions
698 on the content of the pattern that apply when using the standard algo-
699 rithm for partial matching do not apply to the alternative algorithm.
700 For non-anchored patterns, the starting position of a partial match is
701 available.
702
703 3. Because the alternative algorithm scans the subject string just
704 once, and never needs to backtrack, it is possible to pass very long
705 subject strings to the matching function in several pieces, checking
706 for partial matching each time.
707
708
709 DISADVANTAGES OF THE ALTERNATIVE ALGORITHM
710
711 The alternative algorithm suffers from a number of disadvantages:
712
713 1. It is substantially slower than the standard algorithm. This is
714 partly because it has to search for all possible matches, but is also
715 because it is less susceptible to optimization.
716
717 2. Capturing parentheses and back references are not supported.
718
719 3. Although atomic groups are supported, their use does not provide the
720 performance advantage that it does for the standard algorithm.
721
722
723 AUTHOR
724
725 Philip Hazel
726 University Computing Service
727 Cambridge CB2 3QH, England.
728
729
730 REVISION
731
732 Last updated: 08 August 2007
733 Copyright (c) 1997-2007 University of Cambridge.
734 ------------------------------------------------------------------------------
735
736
737 PCREAPI(3) PCREAPI(3)
738
739
740 NAME
741 PCRE - Perl-compatible regular expressions
742
743
744 PCRE NATIVE API
745
746 #include <pcre.h>
747
748 pcre *pcre_compile(const char *pattern, int options,
749 const char **errptr, int *erroffset,
750 const unsigned char *tableptr);
751
752 pcre *pcre_compile2(const char *pattern, int options,
753 int *errorcodeptr,
754 const char **errptr, int *erroffset,
755 const unsigned char *tableptr);
756
757 pcre_extra *pcre_study(const pcre *code, int options,
758 const char **errptr);
759
760 int pcre_exec(const pcre *code, const pcre_extra *extra,
761 const char *subject, int length, int startoffset,
762 int options, int *ovector, int ovecsize);
763
764 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
765 const char *subject, int length, int startoffset,
766 int options, int *ovector, int ovecsize,
767 int *workspace, int wscount);
768
769 int pcre_copy_named_substring(const pcre *code,
770 const char *subject, int *ovector,
771 int stringcount, const char *stringname,
772 char *buffer, int buffersize);
773
774 int pcre_copy_substring(const char *subject, int *ovector,
775 int stringcount, int stringnumber, char *buffer,
776 int buffersize);
777
778 int pcre_get_named_substring(const pcre *code,
779 const char *subject, int *ovector,
780 int stringcount, const char *stringname,
781 const char **stringptr);
782
783 int pcre_get_stringnumber(const pcre *code,
784 const char *name);
785
786 int pcre_get_stringtable_entries(const pcre *code,
787 const char *name, char **first, char **last);
788
789 int pcre_get_substring(const char *subject, int *ovector,
790 int stringcount, int stringnumber,
791 const char **stringptr);
792
793 int pcre_get_substring_list(const char *subject,
794 int *ovector, int stringcount, const char ***listptr);
795
796 void pcre_free_substring(const char *stringptr);
797
798 void pcre_free_substring_list(const char **stringptr);
799
800 const unsigned char *pcre_maketables(void);
801
802 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
803 int what, void *where);
804
805 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
806
807 int pcre_refcount(pcre *code, int adjust);
808
809 int pcre_config(int what, void *where);
810
811 char *pcre_version(void);
812
813 void *(*pcre_malloc)(size_t);
814
815 void (*pcre_free)(void *);
816
817 void *(*pcre_stack_malloc)(size_t);
818
819 void (*pcre_stack_free)(void *);
820
821 int (*pcre_callout)(pcre_callout_block *);
822
823
824 PCRE API OVERVIEW
825
826 PCRE has its own native API, which is described in this document. There
827 are also some wrapper functions that correspond to the POSIX regular
828 expression API. These are described in the pcreposix documentation.
829 Both of these APIs define a set of C function calls. A C++ wrapper is
830 distributed with PCRE. It is documented in the pcrecpp page.
831
832 The native API C function prototypes are defined in the header file
833 pcre.h, and on Unix systems the library itself is called libpcre. It
834 can normally be accessed by adding -lpcre to the command for linking an
835 application that uses PCRE. The header file defines the macros
836 PCRE_MAJOR and PCRE_MINOR to contain the major and minor release num-
837 bers for the library. Applications can use these to include support
838 for different releases of PCRE.
839
840 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
841 pcre_exec() are used for compiling and matching regular expressions in
842 a Perl-compatible manner. A sample program that demonstrates the sim-
843 plest way of using them is provided in the file called pcredemo.c in
844 the source distribution. The pcresample documentation describes how to
845 run it.
846
847 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
848 ble, is also provided. This uses a different algorithm for the match-
849 ing. The alternative algorithm finds all possible matches (at a given
850 point in the subject), and scans the subject just once. However, this
851 algorithm does not return captured substrings. A description of the two
852 matching algorithms and their advantages and disadvantages is given in
853 the pcrematching documentation.
854
855 In addition to the main compiling and matching functions, there are
856 convenience functions for extracting captured substrings from a subject
857 string that is matched by pcre_exec(). They are:
858
859 pcre_copy_substring()
860 pcre_copy_named_substring()
861 pcre_get_substring()
862 pcre_get_named_substring()
863 pcre_get_substring_list()
864 pcre_get_stringnumber()
865 pcre_get_stringtable_entries()
866
867 pcre_free_substring() and pcre_free_substring_list() are also provided,
868 to free the memory used for extracted strings.
869
870 The function pcre_maketables() is used to build a set of character
871 tables in the current locale for passing to pcre_compile(),
872 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
873 provided for specialist use. Most commonly, no special tables are
874 passed, in which case internal tables that are generated when PCRE is
875 built are used.
876
877 The function pcre_fullinfo() is used to find out information about a
878 compiled pattern; pcre_info() is an obsolete version that returns only
879 some of the available information, but is retained for backwards com-
880 patibility. The function pcre_version() returns a pointer to a string
881 containing the version of PCRE and its date of release.
882
883 The function pcre_refcount() maintains a reference count in a data
884 block containing a compiled pattern. This is provided for the benefit
885 of object-oriented applications.
886
887 The global variables pcre_malloc and pcre_free initially contain the
888 entry points of the standard malloc() and free() functions, respec-
889 tively. PCRE calls the memory management functions via these variables,
890 so a calling program can replace them if it wishes to intercept the
891 calls. This should be done before calling any PCRE functions.
892
893 The global variables pcre_stack_malloc and pcre_stack_free are also
894 indirections to memory management functions. These special functions
895 are used only when PCRE is compiled to use the heap for remembering
896 data, instead of recursive function calls, when running the pcre_exec()
897 function. See the pcrebuild documentation for details of how to do
898 this. It is a non-standard way of building PCRE, for use in environ-
899 ments that have limited stacks. Because of the greater use of memory
900 management, it runs more slowly. Separate functions are provided so
901 that special-purpose external code can be used for this case. When
902 used, these functions are always called in a stack-like manner (last
903 obtained, first freed), and always for memory blocks of the same size.
904 There is a discussion about PCRE's stack usage in the pcrestack docu-
905 mentation.
906
907 The global variable pcre_callout initially contains NULL. It can be set
908 by the caller to a "callout" function, which PCRE will then call at
909 specified points during a matching operation. Details are given in the
910 pcrecallout documentation.
911
912
913 NEWLINES
914
915 PCRE supports five different conventions for indicating line breaks in
916 strings: a single CR (carriage return) character, a single LF (line-
917 feed) character, the two-character sequence CRLF, any of the three pre-
918 ceding, or any Unicode newline sequence. The Unicode newline sequences
919 are the three just mentioned, plus the single characters VT (vertical
920 tab, U+000B), FF (formfeed, U+000C), NEL (next line, U+0085), LS (line
921 separator, U+2028), and PS (paragraph separator, U+2029).
922
923 Each of the first three conventions is used by at least one operating
924 system as its standard newline sequence. When PCRE is built, a default
925 can be specified. The default default is LF, which is the Unix stan-
926 dard. When PCRE is run, the default can be overridden, either when a
927 pattern is compiled, or when it is matched.
928
929 At compile time, the newline convention can be specified by the options
930 argument of pcre_compile(), or it can be specified by special text at
931 the start of the pattern itself; this overrides any other settings. See
932 the pcrepattern page for details of the special character sequences.
933
934 In the PCRE documentation the word "newline" is used to mean "the char-
935 acter or pair of characters that indicate a line break". The choice of
936 newline convention affects the handling of the dot, circumflex, and
937 dollar metacharacters, the handling of #-comments in /x mode, and, when
938 CRLF is a recognized line ending sequence, the match position advance-
939 ment for a non-anchored pattern. There is more detail about this in the
940 section on pcre_exec() options below.
941
942 The choice of newline convention does not affect the interpretation of
943 the \n or \r escape sequences, nor does it affect what \R matches,
944 which is controlled in a similar way, but by separate options.
945
946
947 MULTITHREADING
948
949 The PCRE functions can be used in multi-threading applications, with
950 the proviso that the memory management functions pointed to by
951 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
952 callout function pointed to by pcre_callout, are shared by all threads.
953
954 The compiled form of a regular expression is not altered during match-
955 ing, so the same compiled pattern can safely be used by several threads
956 at once.
957
958
959 SAVING PRECOMPILED PATTERNS FOR LATER USE
960
961 The compiled form of a regular expression can be saved and re-used at a
962 later time, possibly by a different program, and even on a host other
963 than the one on which it was compiled. Details are given in the
964 pcreprecompile documentation. However, compiling a regular expression
965 with one version of PCRE for use with a different version is not guar-
966 anteed to work and may cause crashes.
967
968
969 CHECKING BUILD-TIME OPTIONS
970
971 int pcre_config(int what, void *where);
972
973 The function pcre_config() makes it possible for a PCRE client to dis-
974 cover which optional features have been compiled into the PCRE library.
975 The pcrebuild documentation has more details about these optional fea-
976 tures.
977
978 The first argument for pcre_config() is an integer, specifying which
979 information is required; the second argument is a pointer to a variable
980 into which the information is placed. The following information is
981 available:
982
983 PCRE_CONFIG_UTF8
984
985 The output is an integer that is set to one if UTF-8 support is avail-
986 able; otherwise it is set to zero.
987
988 PCRE_CONFIG_UNICODE_PROPERTIES
989
990 The output is an integer that is set to one if support for Unicode
991 character properties is available; otherwise it is set to zero.
992
993 PCRE_CONFIG_NEWLINE
994
995 The output is an integer whose value specifies the default character
996 sequence that is recognized as meaning "newline". The four values that
997 are supported are: 10 for LF, 13 for CR, 3338 for CRLF, -2 for ANYCRLF,
998 and -1 for ANY. The default should normally be the standard sequence
999 for your operating system.
1000
1001 PCRE_CONFIG_BSR
1002
1003 The output is an integer whose value indicates what character sequences
1004 the \R escape sequence matches by default. A value of 0 means that \R
1005 matches any Unicode line ending sequence; a value of 1 means that \R
1006 matches only CR, LF, or CRLF. The default can be overridden when a pat-
1007 tern is compiled or matched.
1008
1009 PCRE_CONFIG_LINK_SIZE
1010
1011 The output is an integer that contains the number of bytes used for
1012 internal linkage in compiled regular expressions. The value is 2, 3, or
1013 4. Larger values allow larger regular expressions to be compiled, at
1014 the expense of slower matching. The default value of 2 is sufficient
1015 for all but the most massive patterns, since it allows the compiled
1016 pattern to be up to 64K in size.
1017
1018 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
1019
1020 The output is an integer that contains the threshold above which the
1021 POSIX interface uses malloc() for output vectors. Further details are
1022 given in the pcreposix documentation.
1023
1024 PCRE_CONFIG_MATCH_LIMIT
1025
1026 The output is an integer that gives the default limit for the number of
1027 internal matching function calls in a pcre_exec() execution. Further
1028 details are given with pcre_exec() below.
1029
1030 PCRE_CONFIG_MATCH_LIMIT_RECURSION
1031
1032 The output is an integer that gives the default limit for the depth of
1033 recursion when calling the internal matching function in a pcre_exec()
1034 execution. Further details are given with pcre_exec() below.
1035
1036 PCRE_CONFIG_STACKRECURSE
1037
1038 The output is an integer that is set to one if internal recursion when
1039 running pcre_exec() is implemented by recursive function calls that use
1040 the stack to remember their state. This is the usual way that PCRE is
1041 compiled. The output is zero if PCRE was compiled to use blocks of data
1042 on the heap instead of recursive function calls. In this case,
1043 pcre_stack_malloc and pcre_stack_free are called to manage memory
1044 blocks on the heap, thus avoiding the use of the stack.
1045
1046
1047 COMPILING A PATTERN
1048
1049 pcre *pcre_compile(const char *pattern, int options,
1050 const char **errptr, int *erroffset,
1051 const unsigned char *tableptr);
1052
1053 pcre *pcre_compile2(const char *pattern, int options,
1054 int *errorcodeptr,
1055 const char **errptr, int *erroffset,
1056 const unsigned char *tableptr);
1057
1058 Either of the functions pcre_compile() or pcre_compile2() can be called
1059 to compile a pattern into an internal form. The only difference between
1060 the two interfaces is that pcre_compile2() has an additional argument,
1061 errorcodeptr, via which a numerical error code can be returned.
1062
1063 The pattern is a C string terminated by a binary zero, and is passed in
1064 the pattern argument. A pointer to a single block of memory that is
1065 obtained via pcre_malloc is returned. This contains the compiled code
1066 and related data. The pcre type is defined for the returned block; this
1067 is a typedef for a structure whose contents are not externally defined.
1068 It is up to the caller to free the memory (via pcre_free) when it is no
1069 longer required.
1070
1071 Although the compiled code of a PCRE regex is relocatable, that is, it
1072 does not depend on memory location, the complete pcre data block is not
1073 fully relocatable, because it may contain a copy of the tableptr argu-
1074 ment, which is an address (see below).
1075
1076 The options argument contains various bit settings that affect the com-
1077 pilation. It should be zero if no options are required. The available
1078 options are described below. Some of them, in particular, those that
1079 are compatible with Perl, can also be set and unset from within the
1080 pattern (see the detailed description in the pcrepattern documenta-
1081 tion). For these options, the contents of the options argument speci-
1082 fies their initial settings at the start of compilation and execution.
1083 The PCRE_ANCHORED and PCRE_NEWLINE_xxx options can be set at the time
1084 of matching as well as at compile time.
1085
1086 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
1087 if compilation of a pattern fails, pcre_compile() returns NULL, and
1088 sets the variable pointed to by errptr to point to a textual error mes-
1089 sage. This is a static string that is part of the library. You must not
1090 try to free it. The offset from the start of the pattern to the charac-
1091 ter where the error was discovered is placed in the variable pointed to
1092 by erroffset, which must not be NULL. If it is, an immediate error is
1093 given.
1094
1095 If pcre_compile2() is used instead of pcre_compile(), and the error-
1096 codeptr argument is not NULL, a non-zero error code number is returned
1097 via this argument in the event of an error. This is in addition to the
1098 textual error message. Error codes and messages are listed below.
1099
1100 If the final argument, tableptr, is NULL, PCRE uses a default set of
1101 character tables that are built when PCRE is compiled, using the
1102 default C locale. Otherwise, tableptr must be an address that is the
1103 result of a call to pcre_maketables(). This value is stored with the
1104 compiled pattern, and used again by pcre_exec(), unless another table
1105 pointer is passed to it. For more discussion, see the section on locale
1106 support below.
1107
1108 This code fragment shows a typical straightforward call to pcre_com-
1109 pile():
1110
1111 pcre *re;
1112 const char *error;
1113 int erroffset;
1114 re = pcre_compile(
1115 "^A.*Z", /* the pattern */
1116 0, /* default options */
1117 &error, /* for error message */
1118 &erroffset, /* for error offset */
1119 NULL); /* use default character tables */
1120
1121 The following names for option bits are defined in the pcre.h header
1122 file:
1123
1124 PCRE_ANCHORED
1125
1126 If this bit is set, the pattern is forced to be "anchored", that is, it
1127 is constrained to match only at the first matching point in the string
1128 that is being searched (the "subject string"). This effect can also be
1129 achieved by appropriate constructs in the pattern itself, which is the
1130 only way to do it in Perl.
1131
1132 PCRE_AUTO_CALLOUT
1133
1134 If this bit is set, pcre_compile() automatically inserts callout items,
1135 all with number 255, before each pattern item. For discussion of the
1136 callout facility, see the pcrecallout documentation.
1137
1138 PCRE_BSR_ANYCRLF
1139 PCRE_BSR_UNICODE
1140
1141 These options (which are mutually exclusive) control what the \R escape
1142 sequence matches. The choice is either to match only CR, LF, or CRLF,
1143 or to match any Unicode newline sequence. The default is specified when
1144 PCRE is built. It can be overridden from within the pattern, or by set-
1145 ting an option when a compiled pattern is matched.
1146
1147 PCRE_CASELESS
1148
1149 If this bit is set, letters in the pattern match both upper and lower
1150 case letters. It is equivalent to Perl's /i option, and it can be
1151 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
1152 always understands the concept of case for characters whose values are
1153 less than 128, so caseless matching is always possible. For characters
1154 with higher values, the concept of case is supported if PCRE is com-
1155 piled with Unicode property support, but not otherwise. If you want to
1156 use caseless matching for characters 128 and above, you must ensure
1157 that PCRE is compiled with Unicode property support as well as with
1158 UTF-8 support.
1159
1160 PCRE_DOLLAR_ENDONLY
1161
1162 If this bit is set, a dollar metacharacter in the pattern matches only
1163 at the end of the subject string. Without this option, a dollar also
1164 matches immediately before a newline at the end of the string (but not
1165 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
1166 if PCRE_MULTILINE is set. There is no equivalent to this option in
1167 Perl, and no way to set it within a pattern.
1168
1169 PCRE_DOTALL
1170
1171 If this bit is set, a dot metacharater in the pattern matches all char-
1172 acters, including those that indicate newline. Without it, a dot does
1173 not match when the current position is at a newline. This option is
1174 equivalent to Perl's /s option, and it can be changed within a pattern
1175 by a (?s) option setting. A negative class such as [^a] always matches
1176 newline characters, independent of the setting of this option.
1177
1178 PCRE_DUPNAMES
1179
1180 If this bit is set, names used to identify capturing subpatterns need
1181 not be unique. This can be helpful for certain types of pattern when it
1182 is known that only one instance of the named subpattern can ever be
1183 matched. There are more details of named subpatterns below; see also
1184 the pcrepattern documentation.
1185
1186 PCRE_EXTENDED
1187
1188 If this bit is set, whitespace data characters in the pattern are
1189 totally ignored except when escaped or inside a character class. White-
1190 space does not include the VT character (code 11). In addition, charac-
1191 ters between an unescaped # outside a character class and the next new-
1192 line, inclusive, are also ignored. This is equivalent to Perl's /x
1193 option, and it can be changed within a pattern by a (?x) option set-
1194 ting.
1195
1196 This option makes it possible to include comments inside complicated
1197 patterns. Note, however, that this applies only to data characters.
1198 Whitespace characters may never appear within special character
1199 sequences in a pattern, for example within the sequence (?( which
1200 introduces a conditional subpattern.
1201
1202 PCRE_EXTRA
1203
1204 This option was invented in order to turn on additional functionality
1205 of PCRE that is incompatible with Perl, but it is currently of very
1206 little use. When set, any backslash in a pattern that is followed by a
1207 letter that has no special meaning causes an error, thus reserving
1208 these combinations for future expansion. By default, as in Perl, a
1209 backslash followed by a letter with no special meaning is treated as a
1210 literal. (Perl can, however, be persuaded to give a warning for this.)
1211 There are at present no other features controlled by this option. It
1212 can also be set by a (?X) option setting within a pattern.
1213
1214 PCRE_FIRSTLINE
1215
1216 If this option is set, an unanchored pattern is required to match
1217 before or at the first newline in the subject string, though the
1218 matched text may continue over the newline.
1219
1220 PCRE_MULTILINE
1221
1222 By default, PCRE treats the subject string as consisting of a single
1223 line of characters (even if it actually contains newlines). The "start
1224 of line" metacharacter (^) matches only at the start of the string,
1225 while the "end of line" metacharacter ($) matches only at the end of
1226 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1227 is set). This is the same as Perl.
1228
1229 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1230 constructs match immediately following or immediately before internal
1231 newlines in the subject string, respectively, as well as at the very
1232 start and end. This is equivalent to Perl's /m option, and it can be
1233 changed within a pattern by a (?m) option setting. If there are no new-
1234 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1235 setting PCRE_MULTILINE has no effect.
1236
1237 PCRE_NEWLINE_CR
1238 PCRE_NEWLINE_LF
1239 PCRE_NEWLINE_CRLF
1240 PCRE_NEWLINE_ANYCRLF
1241 PCRE_NEWLINE_ANY
1242
1243 These options override the default newline definition that was chosen
1244 when PCRE was built. Setting the first or the second specifies that a
1245 newline is indicated by a single character (CR or LF, respectively).
1246 Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
1247 two-character CRLF sequence. Setting PCRE_NEWLINE_ANYCRLF specifies
1248 that any of the three preceding sequences should be recognized. Setting
1249 PCRE_NEWLINE_ANY specifies that any Unicode newline sequence should be
1250 recognized. The Unicode newline sequences are the three just mentioned,
1251 plus the single characters VT (vertical tab, U+000B), FF (formfeed,
1252 U+000C), NEL (next line, U+0085), LS (line separator, U+2028), and PS
1253 (paragraph separator, U+2029). The last two are recognized only in
1254 UTF-8 mode.
1255
1256 The newline setting in the options word uses three bits that are
1257 treated as a number, giving eight possibilities. Currently only six are
1258 used (default plus the five values above). This means that if you set
1259 more than one newline option, the combination may or may not be sensi-
1260 ble. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equivalent to
1261 PCRE_NEWLINE_CRLF, but other combinations may yield unused numbers and
1262 cause an error.
1263
1264 The only time that a line break is specially recognized when compiling
1265 a pattern is if PCRE_EXTENDED is set, and an unescaped # outside a
1266 character class is encountered. This indicates a comment that lasts
1267 until after the next line break sequence. In other circumstances, line
1268 break sequences are treated as literal data, except that in
1269 PCRE_EXTENDED mode, both CR and LF are treated as whitespace characters
1270 and are therefore ignored.
1271
1272 The newline option that is set at compile time becomes the default that
1273 is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1274
1275 PCRE_NO_AUTO_CAPTURE
1276
1277 If this option is set, it disables the use of numbered capturing paren-
1278 theses in the pattern. Any opening parenthesis that is not followed by
1279 ? behaves as if it were followed by ?: but named parentheses can still
1280 be used for capturing (and they acquire numbers in the usual way).
1281 There is no equivalent of this option in Perl.
1282
1283 PCRE_UNGREEDY
1284
1285 This option inverts the "greediness" of the quantifiers so that they
1286 are not greedy by default, but become greedy if followed by "?". It is
1287 not compatible with Perl. It can also be set by a (?U) option setting
1288 within the pattern.
1289
1290 PCRE_UTF8
1291
1292 This option causes PCRE to regard both the pattern and the subject as
1293 strings of UTF-8 characters instead of single-byte character strings.
1294 However, it is available only when PCRE is built to include UTF-8 sup-
1295 port. If not, the use of this option provokes an error. Details of how
1296 this option changes the behaviour of PCRE are given in the section on
1297 UTF-8 support in the main pcre page.
1298
1299 PCRE_NO_UTF8_CHECK
1300
1301 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1302 automatically checked. There is a discussion about the validity of
1303 UTF-8 strings in the main pcre page. If an invalid UTF-8 sequence of
1304 bytes is found, pcre_compile() returns an error. If you already know
1305 that your pattern is valid, and you want to skip this check for perfor-
1306 mance reasons, you can set the PCRE_NO_UTF8_CHECK option. When it is
1307 set, the effect of passing an invalid UTF-8 string as a pattern is
1308 undefined. It may cause your program to crash. Note that this option
1309 can also be passed to pcre_exec() and pcre_dfa_exec(), to suppress the
1310 UTF-8 validity checking of subject strings.
1311
1312
1313 COMPILATION ERROR CODES
1314
1315 The following table lists the error codes than may be returned by
1316 pcre_compile2(), along with the error messages that may be returned by
1317 both compiling functions. As PCRE has developed, some error codes have
1318 fallen out of use. To avoid confusion, they have not been re-used.
1319
1320 0 no error
1321 1 \ at end of pattern
1322 2 \c at end of pattern
1323 3 unrecognized character follows \
1324 4 numbers out of order in {} quantifier
1325 5 number too big in {} quantifier
1326 6 missing terminating ] for character class
1327 7 invalid escape sequence in character class
1328 8 range out of order in character class
1329 9 nothing to repeat
1330 10 [this code is not in use]
1331 11 internal error: unexpected repeat
1332 12 unrecognized character after (?
1333 13 POSIX named classes are supported only within a class
1334 14 missing )
1335 15 reference to non-existent subpattern
1336 16 erroffset passed as NULL
1337 17 unknown option bit(s) set
1338 18 missing ) after comment
1339 19 [this code is not in use]
1340 20 regular expression too large
1341 21 failed to get memory
1342 22 unmatched parentheses
1343 23 internal error: code overflow
1344 24 unrecognized character after (?<
1345 25 lookbehind assertion is not fixed length
1346 26 malformed number or name after (?(
1347 27 conditional group contains more than two branches
1348 28 assertion expected after (?(
1349 29 (?R or (?[+-]digits must be followed by )
1350 30 unknown POSIX class name
1351 31 POSIX collating elements are not supported
1352 32 this version of PCRE is not compiled with PCRE_UTF8 support
1353 33 [this code is not in use]
1354 34 character value in \x{...} sequence is too large
1355 35 invalid condition (?(0)
1356 36 \C not allowed in lookbehind assertion
1357 37 PCRE does not support \L, \l, \N, \U, or \u
1358 38 number after (?C is > 255
1359 39 closing ) for (?C expected
1360 40 recursive call could loop indefinitely
1361 41 unrecognized character after (?P
1362 42 syntax error in subpattern name (missing terminator)
1363 43 two named subpatterns have the same name
1364 44 invalid UTF-8 string
1365 45 support for \P, \p, and \X has not been compiled
1366 46 malformed \P or \p sequence
1367 47 unknown property name after \P or \p
1368 48 subpattern name is too long (maximum 32 characters)
1369 49 too many named subpatterns (maximum 10,000)
1370 50 [this code is not in use]
1371 51 octal value is greater than \377 (not in UTF-8 mode)
1372 52 internal error: overran compiling workspace
1373 53 internal error: previously-checked referenced subpattern not
1374 found
1375 54 DEFINE group contains more than one branch
1376 55 repeating a DEFINE group is not allowed
1377 56 inconsistent NEWLINE options
1378 57 \g is not followed by a braced name or an optionally braced
1379 non-zero number
1380 58 (?+ or (?- or (?(+ or (?(- must be followed by a non-zero number
1381
1382
1383 STUDYING A PATTERN
1384
1385 pcre_extra *pcre_study(const pcre *code, int options
1386 const char **errptr);
1387
1388 If a compiled pattern is going to be used several times, it is worth
1389 spending more time analyzing it in order to speed up the time taken for
1390 matching. The function pcre_study() takes a pointer to a compiled pat-
1391 tern as its first argument. If studying the pattern produces additional
1392 information that will help speed up matching, pcre_study() returns a
1393 pointer to a pcre_extra block, in which the study_data field points to
1394 the results of the study.
1395
1396 The returned value from pcre_study() can be passed directly to
1397 pcre_exec(). However, a pcre_extra block also contains other fields
1398 that can be set by the caller before the block is passed; these are
1399 described below in the section on matching a pattern.
1400
1401 If studying the pattern does not produce any additional information
1402 pcre_study() returns NULL. In that circumstance, if the calling program
1403 wants to pass any of the other fields to pcre_exec(), it must set up
1404 its own pcre_extra block.
1405
1406 The second argument of pcre_study() contains option bits. At present,
1407 no options are defined, and this argument should always be zero.
1408
1409 The third argument for pcre_study() is a pointer for an error message.
1410 If studying succeeds (even if no data is returned), the variable it
1411 points to is set to NULL. Otherwise it is set to point to a textual
1412 error message. This is a static string that is part of the library. You
1413 must not try to free it. You should test the error pointer for NULL
1414 after calling pcre_study(), to be sure that it has run successfully.
1415
1416 This is a typical call to pcre_study():
1417
1418 pcre_extra *pe;
1419 pe = pcre_study(
1420 re, /* result of pcre_compile() */
1421 0, /* no options exist */
1422 &error); /* set to NULL or points to a message */
1423
1424 At present, studying a pattern is useful only for non-anchored patterns
1425 that do not have a single fixed starting character. A bitmap of possi-
1426 ble starting bytes is created.
1427
1428
1429 LOCALE SUPPORT
1430
1431 PCRE handles caseless matching, and determines whether characters are
1432 letters, digits, or whatever, by reference to a set of tables, indexed
1433 by character value. When running in UTF-8 mode, this applies only to
1434 characters with codes less than 128. Higher-valued codes never match
1435 escapes such as \w or \d, but can be tested with \p if PCRE is built
1436 with Unicode character property support. The use of locales with Uni-
1437 code is discouraged. If you are handling characters with codes greater
1438 than 128, you should either use UTF-8 and Unicode, or use locales, but
1439 not try to mix the two.
1440
1441 PCRE contains an internal set of tables that are used when the final
1442 argument of pcre_compile() is NULL. These are sufficient for many
1443 applications. Normally, the internal tables recognize only ASCII char-
1444 acters. However, when PCRE is built, it is possible to cause the inter-
1445 nal tables to be rebuilt in the default "C" locale of the local system,
1446 which may cause them to be different.
1447
1448 The internal tables can always be overridden by tables supplied by the
1449 application that calls PCRE. These may be created in a different locale
1450 from the default. As more and more applications change to using Uni-
1451 code, the need for this locale support is expected to die away.
1452
1453 External tables are built by calling the pcre_maketables() function,
1454 which has no arguments, in the relevant locale. The result can then be
1455 passed to pcre_compile() or pcre_exec() as often as necessary. For
1456 example, to build and use tables that are appropriate for the French
1457 locale (where accented characters with values greater than 128 are
1458 treated as letters), the following code could be used:
1459
1460 setlocale(LC_CTYPE, "fr_FR");
1461 tables = pcre_maketables();
1462 re = pcre_compile(..., tables);
1463
1464 The locale name "fr_FR" is used on Linux and other Unix-like systems;
1465 if you are using Windows, the name for the French locale is "french".
1466
1467 When pcre_maketables() runs, the tables are built in memory that is
1468 obtained via pcre_malloc. It is the caller's responsibility to ensure
1469 that the memory containing the tables remains available for as long as
1470 it is needed.
1471
1472 The pointer that is passed to pcre_compile() is saved with the compiled
1473 pattern, and the same tables are used via this pointer by pcre_study()
1474 and normally also by pcre_exec(). Thus, by default, for any single pat-
1475 tern, compilation, studying and matching all happen in the same locale,
1476 but different patterns can be compiled in different locales.
1477
1478 It is possible to pass a table pointer or NULL (indicating the use of
1479 the internal tables) to pcre_exec(). Although not intended for this
1480 purpose, this facility could be used to match a pattern in a different
1481 locale from the one in which it was compiled. Passing table pointers at
1482 run time is discussed below in the section on matching a pattern.
1483
1484
1485 INFORMATION ABOUT A PATTERN
1486
1487 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1488 int what, void *where);
1489
1490 The pcre_fullinfo() function returns information about a compiled pat-
1491 tern. It replaces the obsolete pcre_info() function, which is neverthe-
1492 less retained for backwards compability (and is documented below).
1493
1494 The first argument for pcre_fullinfo() is a pointer to the compiled
1495 pattern. The second argument is the result of pcre_study(), or NULL if
1496 the pattern was not studied. The third argument specifies which piece
1497 of information is required, and the fourth argument is a pointer to a
1498 variable to receive the data. The yield of the function is zero for
1499 success, or one of the following negative numbers:
1500
1501 PCRE_ERROR_NULL the argument code was NULL
1502 the argument where was NULL
1503 PCRE_ERROR_BADMAGIC the "magic number" was not found
1504 PCRE_ERROR_BADOPTION the value of what was invalid
1505
1506 The "magic number" is placed at the start of each compiled pattern as
1507 an simple check against passing an arbitrary memory pointer. Here is a
1508 typical call of pcre_fullinfo(), to obtain the length of the compiled
1509 pattern:
1510
1511 int rc;
1512 size_t length;
1513 rc = pcre_fullinfo(
1514 re, /* result of pcre_compile() */
1515 pe, /* result of pcre_study(), or NULL */
1516 PCRE_INFO_SIZE, /* what is required */
1517 &length); /* where to put the data */
1518
1519 The possible values for the third argument are defined in pcre.h, and
1520 are as follows:
1521
1522 PCRE_INFO_BACKREFMAX
1523
1524 Return the number of the highest back reference in the pattern. The
1525 fourth argument should point to an int variable. Zero is returned if
1526 there are no back references.
1527
1528 PCRE_INFO_CAPTURECOUNT
1529
1530 Return the number of capturing subpatterns in the pattern. The fourth
1531 argument should point to an int variable.
1532
1533 PCRE_INFO_DEFAULT_TABLES
1534
1535 Return a pointer to the internal default character tables within PCRE.
1536 The fourth argument should point to an unsigned char * variable. This
1537 information call is provided for internal use by the pcre_study() func-
1538 tion. External callers can cause PCRE to use its internal tables by
1539 passing a NULL table pointer.
1540
1541 PCRE_INFO_FIRSTBYTE
1542
1543 Return information about the first byte of any matched string, for a
1544 non-anchored pattern. The fourth argument should point to an int vari-
1545 able. (This option used to be called PCRE_INFO_FIRSTCHAR; the old name
1546 is still recognized for backwards compatibility.)
1547
1548 If there is a fixed first byte, for example, from a pattern such as
1549 (cat|cow|coyote), its value is returned. Otherwise, if either
1550
1551 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
1552 branch starts with "^", or
1553
1554 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
1555 set (if it were set, the pattern would be anchored),
1556
1557 -1 is returned, indicating that the pattern matches only at the start
1558 of a subject string or after any newline within the string. Otherwise
1559 -2 is returned. For anchored patterns, -2 is returned.
1560
1561 PCRE_INFO_FIRSTTABLE
1562
1563 If the pattern was studied, and this resulted in the construction of a
1564 256-bit table indicating a fixed set of bytes for the first byte in any
1565 matching string, a pointer to the table is returned. Otherwise NULL is
1566 returned. The fourth argument should point to an unsigned char * vari-
1567 able.
1568
1569 PCRE_INFO_HASCRORLF
1570
1571 Return 1 if the pattern contains any explicit matches for CR or LF
1572 characters, otherwise 0. The fourth argument should point to an int
1573 variable. An explicit match is either a literal CR or LF character, or
1574 \r or \n.
1575
1576 PCRE_INFO_JCHANGED
1577
1578 Return 1 if the (?J) option setting is used in the pattern, otherwise
1579 0. The fourth argument should point to an int variable. The (?J) inter-
1580 nal option setting changes the local PCRE_DUPNAMES option.
1581
1582 PCRE_INFO_LASTLITERAL
1583
1584 Return the value of the rightmost literal byte that must exist in any
1585 matched string, other than at its start, if such a byte has been
1586 recorded. The fourth argument should point to an int variable. If there
1587 is no such byte, -1 is returned. For anchored patterns, a last literal
1588 byte is recorded only if it follows something of variable length. For
1589 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
1590 /^a\dz\d/ the returned value is -1.
1591
1592 PCRE_INFO_NAMECOUNT
1593 PCRE_INFO_NAMEENTRYSIZE
1594 PCRE_INFO_NAMETABLE
1595
1596 PCRE supports the use of named as well as numbered capturing parenthe-
1597 ses. The names are just an additional way of identifying the parenthe-
1598 ses, which still acquire numbers. Several convenience functions such as
1599 pcre_get_named_substring() are provided for extracting captured sub-
1600 strings by name. It is also possible to extract the data directly, by
1601 first converting the name to a number in order to access the correct
1602 pointers in the output vector (described with pcre_exec() below). To do
1603 the conversion, you need to use the name-to-number map, which is
1604 described by these three values.
1605
1606 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
1607 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
1608 of each entry; both of these return an int value. The entry size
1609 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
1610 a pointer to the first entry of the table (a pointer to char). The
1611 first two bytes of each entry are the number of the capturing parenthe-
1612 sis, most significant byte first. The rest of the entry is the corre-
1613 sponding name, zero terminated. The names are in alphabetical order.
1614 When PCRE_DUPNAMES is set, duplicate names are in order of their paren-
1615 theses numbers. For example, consider the following pattern (assume
1616 PCRE_EXTENDED is set, so white space - including newlines - is
1617 ignored):
1618
1619 (?<date> (?<year>(\d\d)?\d\d) -
1620 (?<month>\d\d) - (?<day>\d\d) )
1621
1622 There are four named subpatterns, so the table has four entries, and
1623 each entry in the table is eight bytes long. The table is as follows,
1624 with non-printing bytes shows in hexadecimal, and undefined bytes shown
1625 as ??:
1626
1627 00 01 d a t e 00 ??
1628 00 05 d a y 00 ?? ??
1629 00 04 m o n t h 00
1630 00 02 y e a r 00 ??
1631
1632 When writing code to extract data from named subpatterns using the
1633 name-to-number map, remember that the length of the entries is likely
1634 to be different for each compiled pattern.
1635
1636 PCRE_INFO_OKPARTIAL
1637
1638 Return 1 if the pattern can be used for partial matching, otherwise 0.
1639 The fourth argument should point to an int variable. The pcrepartial
1640 documentation lists the restrictions that apply to patterns when par-
1641 tial matching is used.
1642
1643 PCRE_INFO_OPTIONS
1644
1645 Return a copy of the options with which the pattern was compiled. The
1646 fourth argument should point to an unsigned long int variable. These
1647 option bits are those specified in the call to pcre_compile(), modified
1648 by any top-level option settings at the start of the pattern itself. In
1649 other words, they are the options that will be in force when matching
1650 starts. For example, if the pattern /(?im)abc(?-i)d/ is compiled with
1651 the PCRE_EXTENDED option, the result is PCRE_CASELESS, PCRE_MULTILINE,
1652 and PCRE_EXTENDED.
1653
1654 A pattern is automatically anchored by PCRE if all of its top-level
1655 alternatives begin with one of the following:
1656
1657 ^ unless PCRE_MULTILINE is set
1658 \A always
1659 \G always
1660 .* if PCRE_DOTALL is set and there are no back
1661 references to the subpattern in which .* appears
1662
1663 For such patterns, the PCRE_ANCHORED bit is set in the options returned
1664 by pcre_fullinfo().
1665
1666 PCRE_INFO_SIZE
1667
1668 Return the size of the compiled pattern, that is, the value that was
1669 passed as the argument to pcre_malloc() when PCRE was getting memory in
1670 which to place the compiled data. The fourth argument should point to a
1671 size_t variable.
1672
1673 PCRE_INFO_STUDYSIZE
1674
1675 Return the size of the data block pointed to by the study_data field in
1676 a pcre_extra block. That is, it is the value that was passed to
1677 pcre_malloc() when PCRE was getting memory into which to place the data
1678 created by pcre_study(). The fourth argument should point to a size_t
1679 variable.
1680
1681
1682 OBSOLETE INFO FUNCTION
1683
1684 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
1685
1686 The pcre_info() function is now obsolete because its interface is too
1687 restrictive to return all the available data about a compiled pattern.
1688 New programs should use pcre_fullinfo() instead. The yield of
1689 pcre_info() is the number of capturing subpatterns, or one of the fol-
1690 lowing negative numbers:
1691
1692 PCRE_ERROR_NULL the argument code was NULL
1693 PCRE_ERROR_BADMAGIC the "magic number" was not found
1694
1695 If the optptr argument is not NULL, a copy of the options with which
1696 the pattern was compiled is placed in the integer it points to (see
1697 PCRE_INFO_OPTIONS above).
1698
1699 If the pattern is not anchored and the firstcharptr argument is not
1700 NULL, it is used to pass back information about the first character of
1701 any matched string (see PCRE_INFO_FIRSTBYTE above).
1702
1703
1704 REFERENCE COUNTS
1705
1706 int pcre_refcount(pcre *code, int adjust);
1707
1708 The pcre_refcount() function is used to maintain a reference count in
1709 the data block that contains a compiled pattern. It is provided for the
1710 benefit of applications that operate in an object-oriented manner,
1711 where different parts of the application may be using the same compiled
1712 pattern, but you want to free the block when they are all done.
1713
1714 When a pattern is compiled, the reference count field is initialized to
1715 zero. It is changed only by calling this function, whose action is to
1716 add the adjust value (which may be positive or negative) to it. The
1717 yield of the function is the new value. However, the value of the count
1718 is constrained to lie between 0 and 65535, inclusive. If the new value
1719 is outside these limits, it is forced to the appropriate limit value.
1720
1721 Except when it is zero, the reference count is not correctly preserved
1722 if a pattern is compiled on one host and then transferred to a host
1723 whose byte-order is different. (This seems a highly unlikely scenario.)
1724
1725
1726 MATCHING A PATTERN: THE TRADITIONAL FUNCTION
1727
1728 int pcre_exec(const pcre *code, const pcre_extra *extra,
1729 const char *subject, int length, int startoffset,
1730 int options, int *ovector, int ovecsize);
1731
1732 The function pcre_exec() is called to match a subject string against a
1733 compiled pattern, which is passed in the code argument. If the pattern
1734 has been studied, the result of the study should be passed in the extra
1735 argument. This function is the main matching facility of the library,
1736 and it operates in a Perl-like manner. For specialist use there is also
1737 an alternative matching function, which is described below in the sec-
1738 tion about the pcre_dfa_exec() function.
1739
1740 In most applications, the pattern will have been compiled (and option-
1741 ally studied) in the same process that calls pcre_exec(). However, it
1742 is possible to save compiled patterns and study data, and then use them
1743 later in different processes, possibly even on different hosts. For a
1744 discussion about this, see the pcreprecompile documentation.
1745
1746 Here is an example of a simple call to pcre_exec():
1747
1748 int rc;
1749 int ovector[30];
1750 rc = pcre_exec(
1751 re, /* result of pcre_compile() */
1752 NULL, /* we didn't study the pattern */
1753 "some string", /* the subject string */
1754 11, /* the length of the subject string */
1755 0, /* start at offset 0 in the subject */
1756 0, /* default options */
1757 ovector, /* vector of integers for substring information */
1758 30); /* number of elements (NOT size in bytes) */
1759
1760 Extra data for pcre_exec()
1761
1762 If the extra argument is not NULL, it must point to a pcre_extra data
1763 block. The pcre_study() function returns such a block (when it doesn't
1764 return NULL), but you can also create one for yourself, and pass addi-
1765 tional information in it. The pcre_extra block contains the following
1766 fields (not necessarily in this order):
1767
1768 unsigned long int flags;
1769 void *study_data;
1770 unsigned long int match_limit;
1771 unsigned long int match_limit_recursion;
1772 void *callout_data;
1773 const unsigned char *tables;
1774
1775 The flags field is a bitmap that specifies which of the other fields
1776 are set. The flag bits are:
1777
1778 PCRE_EXTRA_STUDY_DATA
1779 PCRE_EXTRA_MATCH_LIMIT
1780 PCRE_EXTRA_MATCH_LIMIT_RECURSION
1781 PCRE_EXTRA_CALLOUT_DATA
1782 PCRE_EXTRA_TABLES
1783
1784 Other flag bits should be set to zero. The study_data field is set in
1785 the pcre_extra block that is returned by pcre_study(), together with
1786 the appropriate flag bit. You should not set this yourself, but you may
1787 add to the block by setting the other fields and their corresponding
1788 flag bits.
1789
1790 The match_limit field provides a means of preventing PCRE from using up
1791 a vast amount of resources when running patterns that are not going to
1792 match, but which have a very large number of possibilities in their
1793 search trees. The classic example is the use of nested unlimited
1794 repeats.
1795
1796 Internally, PCRE uses a function called match() which it calls repeat-
1797 edly (sometimes recursively). The limit set by match_limit is imposed
1798 on the number of times this function is called during a match, which
1799 has the effect of limiting the amount of backtracking that can take
1800 place. For patterns that are not anchored, the count restarts from zero
1801 for each position in the subject string.
1802
1803 The default value for the limit can be set when PCRE is built; the
1804 default default is 10 million, which handles all but the most extreme
1805 cases. You can override the default by suppling pcre_exec() with a
1806 pcre_extra block in which match_limit is set, and
1807 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
1808 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
1809
1810 The match_limit_recursion field is similar to match_limit, but instead
1811 of limiting the total number of times that match() is called, it limits
1812 the depth of recursion. The recursion depth is a smaller number than
1813 the total number of calls, because not all calls to match() are recur-
1814 sive. This limit is of use only if it is set smaller than match_limit.
1815
1816 Limiting the recursion depth limits the amount of stack that can be
1817 used, or, when PCRE has been compiled to use memory on the heap instead
1818 of the stack, the amount of heap memory that can be used.
1819
1820 The default value for match_limit_recursion can be set when PCRE is
1821 built; the default default is the same value as the default for
1822 match_limit. You can override the default by suppling pcre_exec() with
1823 a pcre_extra block in which match_limit_recursion is set, and
1824 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
1825 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
1826
1827 The pcre_callout field is used in conjunction with the "callout" fea-
1828 ture, which is described in the pcrecallout documentation.
1829
1830 The tables field is used to pass a character tables pointer to
1831 pcre_exec(); this overrides the value that is stored with the compiled
1832 pattern. A non-NULL value is stored with the compiled pattern only if
1833 custom tables were supplied to pcre_compile() via its tableptr argu-
1834 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
1835 PCRE's internal tables to be used. This facility is helpful when re-
1836 using patterns that have been saved after compiling with an external
1837 set of tables, because the external tables might be at a different
1838 address when pcre_exec() is called. See the pcreprecompile documenta-
1839 tion for a discussion of saving compiled patterns for later use.
1840
1841 Option bits for pcre_exec()
1842
1843 The unused bits of the options argument for pcre_exec() must be zero.
1844 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
1845 PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK and
1846 PCRE_PARTIAL.
1847
1848 PCRE_ANCHORED
1849
1850 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
1851 matching position. If a pattern was compiled with PCRE_ANCHORED, or
1852 turned out to be anchored by virtue of its contents, it cannot be made
1853 unachored at matching time.
1854
1855 PCRE_BSR_ANYCRLF
1856 PCRE_BSR_UNICODE
1857
1858 These options (which are mutually exclusive) control what the \R escape
1859 sequence matches. The choice is either to match only CR, LF, or CRLF,
1860 or to match any Unicode newline sequence. These options override the
1861 choice that was made or defaulted when the pattern was compiled.
1862
1863 PCRE_NEWLINE_CR
1864 PCRE_NEWLINE_LF
1865 PCRE_NEWLINE_CRLF
1866 PCRE_NEWLINE_ANYCRLF
1867 PCRE_NEWLINE_ANY
1868
1869 These options override the newline definition that was chosen or
1870 defaulted when the pattern was compiled. For details, see the descrip-
1871 tion of pcre_compile() above. During matching, the newline choice
1872 affects the behaviour of the dot, circumflex, and dollar metacharac-
1873 ters. It may also alter the way the match position is advanced after a
1874 match failure for an unanchored pattern.
1875
1876 When PCRE_NEWLINE_CRLF, PCRE_NEWLINE_ANYCRLF, or PCRE_NEWLINE_ANY is
1877 set, and a match attempt for an unanchored pattern fails when the cur-
1878 rent position is at a CRLF sequence, and the pattern contains no
1879 explicit matches for CR or LF characters, the match position is
1880 advanced by two characters instead of one, in other words, to after the
1881 CRLF.
1882
1883 The above rule is a compromise that makes the most common cases work as
1884 expected. For example, if the pattern is .+A (and the PCRE_DOTALL
1885 option is not set), it does not match the string "\r\nA" because, after
1886 failing at the start, it skips both the CR and the LF before retrying.
1887 However, the pattern [\r\n]A does match that string, because it con-
1888 tains an explicit CR or LF reference, and so advances only by one char-
1889 acter after the first failure.
1890
1891 An explicit match for CR of LF is either a literal appearance of one of
1892 those characters, or one of the \r or \n escape sequences. Implicit
1893 matches such as [^X] do not count, nor does \s (which includes CR and
1894 LF in the characters that it matches).
1895
1896 Notwithstanding the above, anomalous effects may still occur when CRLF
1897 is a valid newline sequence and explicit \r or \n escapes appear in the
1898 pattern.
1899
1900 PCRE_NOTBOL
1901
1902 This option specifies that first character of the subject string is not
1903 the beginning of a line, so the circumflex metacharacter should not
1904 match before it. Setting this without PCRE_MULTILINE (at compile time)
1905 causes circumflex never to match. This option affects only the behav-
1906 iour of the circumflex metacharacter. It does not affect \A.
1907
1908 PCRE_NOTEOL
1909
1910 This option specifies that the end of the subject string is not the end
1911 of a line, so the dollar metacharacter should not match it nor (except
1912 in multiline mode) a newline immediately before it. Setting this with-
1913 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
1914 option affects only the behaviour of the dollar metacharacter. It does
1915 not affect \Z or \z.
1916
1917 PCRE_NOTEMPTY
1918
1919 An empty string is not considered to be a valid match if this option is
1920 set. If there are alternatives in the pattern, they are tried. If all
1921 the alternatives match the empty string, the entire match fails. For
1922 example, if the pattern
1923
1924 a?b?
1925
1926 is applied to a string not beginning with "a" or "b", it matches the
1927 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
1928 match is not valid, so PCRE searches further into the string for occur-
1929 rences of "a" or "b".
1930
1931 Perl has no direct equivalent of PCRE_NOTEMPTY, but it does make a spe-
1932 cial case of a pattern match of the empty string within its split()
1933 function, and when using the /g modifier. It is possible to emulate
1934 Perl's behaviour after matching a null string by first trying the match
1935 again at the same offset with PCRE_NOTEMPTY and PCRE_ANCHORED, and then
1936 if that fails by advancing the starting offset (see below) and trying
1937 an ordinary match again. There is some code that demonstrates how to do
1938 this in the pcredemo.c sample program.
1939
1940 PCRE_NO_UTF8_CHECK
1941
1942 When PCRE_UTF8 is set at compile time, the validity of the subject as a
1943 UTF-8 string is automatically checked when pcre_exec() is subsequently
1944 called. The value of startoffset is also checked to ensure that it
1945 points to the start of a UTF-8 character. There is a discussion about
1946 the validity of UTF-8 strings in the section on UTF-8 support in the
1947 main pcre page. If an invalid UTF-8 sequence of bytes is found,
1948 pcre_exec() returns the error PCRE_ERROR_BADUTF8. If startoffset con-
1949 tains an invalid value, PCRE_ERROR_BADUTF8_OFFSET is returned.
1950
1951 If you already know that your subject is valid, and you want to skip
1952 these checks for performance reasons, you can set the
1953 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
1954 do this for the second and subsequent calls to pcre_exec() if you are
1955 making repeated calls to find all the matches in a single subject
1956 string. However, you should be sure that the value of startoffset
1957 points to the start of a UTF-8 character. When PCRE_NO_UTF8_CHECK is
1958 set, the effect of passing an invalid UTF-8 string as a subject, or a
1959 value of startoffset that does not point to the start of a UTF-8 char-
1960 acter, is undefined. Your program may crash.
1961
1962 PCRE_PARTIAL
1963
1964 This option turns on the partial matching feature. If the subject
1965 string fails to match the pattern, but at some point during the match-
1966 ing process the end of the subject was reached (that is, the subject
1967 partially matches the pattern and the failure to match occurred only
1968 because there were not enough subject characters), pcre_exec() returns
1969 PCRE_ERROR_PARTIAL instead of PCRE_ERROR_NOMATCH. When PCRE_PARTIAL is
1970 used, there are restrictions on what may appear in the pattern. These
1971 are discussed in the pcrepartial documentation.
1972
1973 The string to be matched by pcre_exec()
1974
1975 The subject string is passed to pcre_exec() as a pointer in subject, a
1976 length in length, and a starting byte offset in startoffset. In UTF-8
1977 mode, the byte offset must point to the start of a UTF-8 character.
1978 Unlike the pattern string, the subject may contain binary zero bytes.
1979 When the starting offset is zero, the search for a match starts at the
1980 beginning of the subject, and this is by far the most common case.
1981
1982 A non-zero starting offset is useful when searching for another match
1983 in the same subject by calling pcre_exec() again after a previous suc-
1984 cess. Setting startoffset differs from just passing over a shortened
1985 string and setting PCRE_NOTBOL in the case of a pattern that begins
1986 with any kind of lookbehind. For example, consider the pattern
1987
1988 \Biss\B
1989
1990 which finds occurrences of "iss" in the middle of words. (\B matches
1991 only if the current position in the subject is not a word boundary.)
1992 When applied to the string "Mississipi" the first call to pcre_exec()
1993 finds the first occurrence. If pcre_exec() is called again with just
1994 the remainder of the subject, namely "issipi", it does not match,
1995 because \B is always false at the start of the subject, which is deemed
1996 to be a word boundary. However, if pcre_exec() is passed the entire
1997 string again, but with startoffset set to 4, it finds the second occur-
1998 rence of "iss" because it is able to look behind the starting point to
1999 discover that it is preceded by a letter.
2000
2001 If a non-zero starting offset is passed when the pattern is anchored,
2002 one attempt to match at the given offset is made. This can only succeed
2003 if the pattern does not require the match to be at the start of the
2004 subject.
2005
2006 How pcre_exec() returns captured substrings
2007
2008 In general, a pattern matches a certain portion of the subject, and in
2009 addition, further substrings from the subject may be picked out by
2010 parts of the pattern. Following the usage in Jeffrey Friedl's book,
2011 this is called "capturing" in what follows, and the phrase "capturing
2012 subpattern" is used for a fragment of a pattern that picks out a sub-
2013 string. PCRE supports several other kinds of parenthesized subpattern
2014 that do not cause substrings to be captured.
2015
2016 Captured substrings are returned to the caller via a vector of integer
2017 offsets whose address is passed in ovector. The number of elements in
2018 the vector is passed in ovecsize, which must be a non-negative number.
2019 Note: this argument is NOT the size of ovector in bytes.
2020
2021 The first two-thirds of the vector is used to pass back captured sub-
2022 strings, each substring using a pair of integers. The remaining third
2023 of the vector is used as workspace by pcre_exec() while matching cap-
2024 turing subpatterns, and is not available for passing back information.
2025 The length passed in ovecsize should always be a multiple of three. If
2026 it is not, it is rounded down.
2027
2028 When a match is successful, information about captured substrings is
2029 returned in pairs of integers, starting at the beginning of ovector,
2030 and continuing up to two-thirds of its length at the most. The first
2031 element of a pair is set to the offset of the first character in a sub-
2032 string, and the second is set to the offset of the first character
2033 after the end of a substring. The first pair, ovector[0] and ovec-
2034 tor[1], identify the portion of the subject string matched by the
2035 entire pattern. The next pair is used for the first capturing subpat-
2036 tern, and so on. The value returned by pcre_exec() is one more than the
2037 highest numbered pair that has been set. For example, if two substrings
2038 have been captured, the returned value is 3. If there are no capturing
2039 subpatterns, the return value from a successful match is 1, indicating
2040 that just the first pair of offsets has been set.
2041
2042 If a capturing subpattern is matched repeatedly, it is the last portion
2043 of the string that it matched that is returned.
2044
2045 If the vector is too small to hold all the captured substring offsets,
2046 it is used as far as possible (up to two-thirds of its length), and the
2047 function returns a value of zero. In particular, if the substring off-
2048 sets are not of interest, pcre_exec() may be called with ovector passed
2049 as NULL and ovecsize as zero. However, if the pattern contains back
2050 references and the ovector is not big enough to remember the related
2051 substrings, PCRE has to get additional memory for use during matching.
2052 Thus it is usually advisable to supply an ovector.
2053
2054 The pcre_info() function can be used to find out how many capturing
2055 subpatterns there are in a compiled pattern. The smallest size for
2056 ovector that will allow for n captured substrings, in addition to the
2057 offsets of the substring matched by the whole pattern, is (n+1)*3.
2058
2059 It is possible for capturing subpattern number n+1 to match some part
2060 of the subject when subpattern n has not been used at all. For example,
2061 if the string "abc" is matched against the pattern (a|(z))(bc) the
2062 return from the function is 4, and subpatterns 1 and 3 are matched, but
2063 2 is not. When this happens, both values in the offset pairs corre-
2064 sponding to unused subpatterns are set to -1.
2065
2066 Offset values that correspond to unused subpatterns at the end of the
2067 expression are also set to -1. For example, if the string "abc" is
2068 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
2069 matched. The return from the function is 2, because the highest used
2070 capturing subpattern number is 1. However, you can refer to the offsets
2071 for the second and third capturing subpatterns if you wish (assuming
2072 the vector is large enough, of course).
2073
2074 Some convenience functions are provided for extracting the captured
2075 substrings as separate strings. These are described below.
2076
2077 Error return values from pcre_exec()
2078
2079 If pcre_exec() fails, it returns a negative number. The following are
2080 defined in the header file:
2081
2082 PCRE_ERROR_NOMATCH (-1)
2083
2084 The subject string did not match the pattern.
2085
2086 PCRE_ERROR_NULL (-2)
2087
2088 Either code or subject was passed as NULL, or ovector was NULL and
2089 ovecsize was not zero.
2090
2091 PCRE_ERROR_BADOPTION (-3)
2092
2093 An unrecognized bit was set in the options argument.
2094
2095 PCRE_ERROR_BADMAGIC (-4)
2096
2097 PCRE stores a 4-byte "magic number" at the start of the compiled code,
2098 to catch the case when it is passed a junk pointer and to detect when a
2099 pattern that was compiled in an environment of one endianness is run in
2100 an environment with the other endianness. This is the error that PCRE
2101 gives when the magic number is not present.
2102
2103 PCRE_ERROR_UNKNOWN_OPCODE (-5)
2104
2105 While running the pattern match, an unknown item was encountered in the
2106 compiled pattern. This error could be caused by a bug in PCRE or by
2107 overwriting of the compiled pattern.
2108
2109 PCRE_ERROR_NOMEMORY (-6)
2110
2111 If a pattern contains back references, but the ovector that is passed
2112 to pcre_exec() is not big enough to remember the referenced substrings,
2113 PCRE gets a block of memory at the start of matching to use for this
2114 purpose. If the call via pcre_malloc() fails, this error is given. The
2115 memory is automatically freed at the end of matching.
2116
2117 PCRE_ERROR_NOSUBSTRING (-7)
2118
2119 This error is used by the pcre_copy_substring(), pcre_get_substring(),
2120 and pcre_get_substring_list() functions (see below). It is never
2121 returned by pcre_exec().
2122
2123 PCRE_ERROR_MATCHLIMIT (-8)
2124
2125 The backtracking limit, as specified by the match_limit field in a
2126 pcre_extra structure (or defaulted) was reached. See the description
2127 above.
2128
2129 PCRE_ERROR_CALLOUT (-9)
2130
2131 This error is never generated by pcre_exec() itself. It is provided for
2132 use by callout functions that want to yield a distinctive error code.
2133 See the pcrecallout documentation for details.
2134
2135 PCRE_ERROR_BADUTF8 (-10)
2136
2137 A string that contains an invalid UTF-8 byte sequence was passed as a
2138 subject.
2139
2140 PCRE_ERROR_BADUTF8_OFFSET (-11)
2141
2142 The UTF-8 byte sequence that was passed as a subject was valid, but the
2143 value of startoffset did not point to the beginning of a UTF-8 charac-
2144 ter.
2145
2146 PCRE_ERROR_PARTIAL (-12)
2147
2148 The subject string did not match, but it did match partially. See the
2149 pcrepartial documentation for details of partial matching.
2150
2151 PCRE_ERROR_BADPARTIAL (-13)
2152
2153 The PCRE_PARTIAL option was used with a compiled pattern containing
2154 items that are not supported for partial matching. See the pcrepartial
2155 documentation for details of partial matching.
2156
2157 PCRE_ERROR_INTERNAL (-14)
2158
2159 An unexpected internal error has occurred. This error could be caused
2160 by a bug in PCRE or by overwriting of the compiled pattern.
2161
2162 PCRE_ERROR_BADCOUNT (-15)
2163
2164 This error is given if the value of the ovecsize argument is negative.
2165
2166 PCRE_ERROR_RECURSIONLIMIT (-21)
2167
2168 The internal recursion limit, as specified by the match_limit_recursion
2169 field in a pcre_extra structure (or defaulted) was reached. See the
2170 description above.
2171
2172 PCRE_ERROR_BADNEWLINE (-23)
2173
2174 An invalid combination of PCRE_NEWLINE_xxx options was given.
2175
2176 Error numbers -16 to -20 and -22 are not used by pcre_exec().
2177
2178
2179 EXTRACTING CAPTURED SUBSTRINGS BY NUMBER
2180
2181 int pcre_copy_substring(const char *subject, int *ovector,
2182 int stringcount, int stringnumber, char *buffer,
2183 int buffersize);
2184
2185 int pcre_get_substring(const char *subject, int *ovector,
2186 int stringcount, int stringnumber,
2187 const char **stringptr);
2188
2189 int pcre_get_substring_list(const char *subject,
2190 int *ovector, int stringcount, const char ***listptr);
2191
2192 Captured substrings can be accessed directly by using the offsets
2193 returned by pcre_exec() in ovector. For convenience, the functions
2194 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
2195 string_list() are provided for extracting captured substrings as new,
2196 separate, zero-terminated strings. These functions identify substrings
2197 by number. The next section describes functions for extracting named
2198 substrings.
2199
2200 A substring that contains a binary zero is correctly extracted and has
2201 a further zero added on the end, but the result is not, of course, a C
2202 string. However, you can process such a string by referring to the
2203 length that is returned by pcre_copy_substring() and pcre_get_sub-
2204 string(). Unfortunately, the interface to pcre_get_substring_list() is
2205 not adequate for handling strings containing binary zeros, because the
2206 end of the final string is not independently indicated.
2207
2208 The first three arguments are the same for all three of these func-
2209 tions: subject is the subject string that has just been successfully
2210 matched, ovector is a pointer to the vector of integer offsets that was
2211 passed to pcre_exec(), and stringcount is the number of substrings that
2212 were captured by the match, including the substring that matched the
2213 entire regular expression. This is the value returned by pcre_exec() if
2214 it is greater than zero. If pcre_exec() returned zero, indicating that
2215 it ran out of space in ovector, the value passed as stringcount should
2216 be the number of elements in the vector divided by three.
2217
2218 The functions pcre_copy_substring() and pcre_get_substring() extract a
2219 single substring, whose number is given as stringnumber. A value of
2220 zero extracts the substring that matched the entire pattern, whereas
2221 higher values extract the captured substrings. For pcre_copy_sub-
2222 string(), the string is placed in buffer, whose length is given by
2223 buffersize, while for pcre_get_substring() a new block of memory is
2224 obtained via pcre_malloc, and its address is returned via stringptr.
2225 The yield of the function is the length of the string, not including
2226 the terminating zero, or one of these error codes:
2227
2228 PCRE_ERROR_NOMEMORY (-6)
2229
2230 The buffer was too small for pcre_copy_substring(), or the attempt to
2231 get memory failed for pcre_get_substring().
2232
2233 PCRE_ERROR_NOSUBSTRING (-7)
2234
2235 There is no substring whose number is stringnumber.
2236
2237 The pcre_get_substring_list() function extracts all available sub-
2238 strings and builds a list of pointers to them. All this is done in a
2239 single block of memory that is obtained via pcre_malloc. The address of
2240 the memory block is returned via listptr, which is also the start of
2241 the list of string pointers. The end of the list is marked by a NULL
2242 pointer. The yield of the function is zero if all went well, or the
2243 error code
2244
2245 PCRE_ERROR_NOMEMORY (-6)
2246
2247 if the attempt to get the memory block failed.
2248
2249 When any of these functions encounter a substring that is unset, which
2250 can happen when capturing subpattern number n+1 matches some part of
2251 the subject, but subpattern n has not been used at all, they return an
2252 empty string. This can be distinguished from a genuine zero-length sub-
2253 string by inspecting the appropriate offset in ovector, which is nega-
2254 tive for unset substrings.
2255
2256 The two convenience functions pcre_free_substring() and pcre_free_sub-
2257 string_list() can be used to free the memory returned by a previous
2258 call of pcre_get_substring() or pcre_get_substring_list(), respec-
2259 tively. They do nothing more than call the function pointed to by
2260 pcre_free, which of course could be called directly from a C program.
2261 However, PCRE is used in some situations where it is linked via a spe-
2262 cial interface to another programming language that cannot use
2263 pcre_free directly; it is for these cases that the functions are pro-
2264 vided.
2265
2266
2267 EXTRACTING CAPTURED SUBSTRINGS BY NAME
2268
2269 int pcre_get_stringnumber(const pcre *code,
2270 const char *name);
2271
2272 int pcre_copy_named_substring(const pcre *code,
2273 const char *subject, int *ovector,
2274 int stringcount, const char *stringname,
2275 char *buffer, int buffersize);
2276
2277 int pcre_get_named_substring(const pcre *code,
2278 const char *subject, int *ovector,
2279 int stringcount, const char *stringname,
2280 const char **stringptr);
2281
2282 To extract a substring by name, you first have to find associated num-
2283 ber. For example, for this pattern
2284
2285 (a+)b(?<xxx>\d+)...
2286
2287 the number of the subpattern called "xxx" is 2. If the name is known to
2288 be unique (PCRE_DUPNAMES was not set), you can find the number from the
2289 name by calling pcre_get_stringnumber(). The first argument is the com-
2290 piled pattern, and the second is the name. The yield of the function is
2291 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
2292 subpattern of that name.
2293
2294 Given the number, you can extract the substring directly, or use one of
2295 the functions described in the previous section. For convenience, there
2296 are also two functions that do the whole job.
2297
2298 Most of the arguments of pcre_copy_named_substring() and
2299 pcre_get_named_substring() are the same as those for the similarly
2300 named functions that extract by number. As these are described in the
2301 previous section, they are not re-described here. There are just two
2302 differences:
2303
2304 First, instead of a substring number, a substring name is given. Sec-
2305 ond, there is an extra argument, given at the start, which is a pointer
2306 to the compiled pattern. This is needed in order to gain access to the
2307 name-to-number translation table.
2308
2309 These functions call pcre_get_stringnumber(), and if it succeeds, they
2310 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
2311 ate. NOTE: If PCRE_DUPNAMES is set and there are duplicate names, the
2312 behaviour may not be what you want (see the next section).
2313
2314
2315 DUPLICATE SUBPATTERN NAMES
2316
2317 int pcre_get_stringtable_entries(const pcre *code,
2318 const char *name, char **first, char **last);
2319
2320 When a pattern is compiled with the PCRE_DUPNAMES option, names for
2321 subpatterns are not required to be unique. Normally, patterns with
2322 duplicate names are such that in any one match, only one of the named
2323 subpatterns participates. An example is shown in the pcrepattern docu-
2324 mentation.
2325
2326 When duplicates are present, pcre_copy_named_substring() and
2327 pcre_get_named_substring() return the first substring corresponding to
2328 the given name that is set. If none are set, PCRE_ERROR_NOSUBSTRING
2329 (-7) is returned; no data is returned. The pcre_get_stringnumber()
2330 function returns one of the numbers that are associated with the name,
2331 but it is not defined which it is.
2332
2333 If you want to get full details of all captured substrings for a given
2334 name, you must use the pcre_get_stringtable_entries() function. The
2335 first argument is the compiled pattern, and the second is the name. The
2336 third and fourth are pointers to variables which are updated by the
2337 function. After it has run, they point to the first and last entries in
2338 the name-to-number table for the given name. The function itself
2339 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
2340 there are none. The format of the table is described above in the sec-
2341 tion entitled Information about a pattern. Given all the relevant
2342 entries for the name, you can extract each of their numbers, and hence
2343 the captured data, if any.
2344
2345
2346 FINDING ALL POSSIBLE MATCHES
2347
2348 The traditional matching function uses a similar algorithm to Perl,
2349 which stops when it finds the first match, starting at a given point in
2350 the subject. If you want to find all possible matches, or the longest
2351 possible match, consider using the alternative matching function (see
2352 below) instead. If you cannot use the alternative function, but still
2353 need to find all possible matches, you can kludge it up by making use
2354 of the callout facility, which is described in the pcrecallout documen-
2355 tation.
2356
2357 What you have to do is to insert a callout right at the end of the pat-
2358 tern. When your callout function is called, extract and save the cur-
2359 rent matched substring. Then return 1, which forces pcre_exec() to
2360 backtrack and try other alternatives. Ultimately, when it runs out of
2361 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
2362
2363
2364 MATCHING A PATTERN: THE ALTERNATIVE FUNCTION
2365
2366 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
2367 const char *subject, int length, int startoffset,
2368 int options, int *ovector, int ovecsize,
2369 int *workspace, int wscount);
2370
2371 The function pcre_dfa_exec() is called to match a subject string
2372 against a compiled pattern, using a matching algorithm that scans the
2373 subject string just once, and does not backtrack. This has different
2374 characteristics to the normal algorithm, and is not compatible with
2375 Perl. Some of the features of PCRE patterns are not supported. Never-
2376 theless, there are times when this kind of matching can be useful. For
2377 a discussion of the two matching algorithms, see the pcrematching docu-
2378 mentation.
2379
2380 The arguments for the pcre_dfa_exec() function are the same as for
2381 pcre_exec(), plus two extras. The ovector argument is used in a differ-
2382 ent way, and this is described below. The other common arguments are
2383 used in the same way as for pcre_exec(), so their description is not
2384 repeated here.
2385
2386 The two additional arguments provide workspace for the function. The
2387 workspace vector should contain at least 20 elements. It is used for
2388 keeping track of multiple paths through the pattern tree. More
2389 workspace will be needed for patterns and subjects where there are a
2390 lot of potential matches.
2391
2392 Here is an example of a simple call to pcre_dfa_exec():
2393
2394 int rc;
2395 int ovector[10];
2396 int wspace[20];
2397 rc = pcre_dfa_exec(
2398 re, /* result of pcre_compile() */
2399 NULL, /* we didn't study the pattern */
2400 "some string", /* the subject string */
2401 11, /* the length of the subject string */
2402 0, /* start at offset 0 in the subject */
2403 0, /* default options */
2404 ovector, /* vector of integers for substring information */
2405 10, /* number of elements (NOT size in bytes) */
2406 wspace, /* working space vector */
2407 20); /* number of elements (NOT size in bytes) */
2408
2409 Option bits for pcre_dfa_exec()
2410
2411 The unused bits of the options argument for pcre_dfa_exec() must be
2412 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
2413 LINE_xxx, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK,
2414 PCRE_PARTIAL, PCRE_DFA_SHORTEST, and PCRE_DFA_RESTART. All but the last
2415 three of these are the same as for pcre_exec(), so their description is
2416 not repeated here.
2417
2418 PCRE_PARTIAL
2419
2420 This has the same general effect as it does for pcre_exec(), but the
2421 details are slightly different. When PCRE_PARTIAL is set for
2422 pcre_dfa_exec(), the return code PCRE_ERROR_NOMATCH is converted into
2423 PCRE_ERROR_PARTIAL if the end of the subject is reached, there have
2424 been no complete matches, but there is still at least one matching pos-
2425 sibility. The portion of the string that provided the partial match is
2426 set as the first matching string.
2427
2428 PCRE_DFA_SHORTEST
2429
2430 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
2431 stop as soon as it has found one match. Because of the way the alterna-
2432 tive algorithm works, this is necessarily the shortest possible match
2433 at the first possible matching point in the subject string.
2434
2435 PCRE_DFA_RESTART
2436
2437 When pcre_dfa_exec() is called with the PCRE_PARTIAL option, and
2438 returns a partial match, it is possible to call it again, with addi-
2439 tional subject characters, and have it continue with the same match.
2440 The PCRE_DFA_RESTART option requests this action; when it is set, the
2441 workspace and wscount options must reference the same vector as before
2442 because data about the match so far is left in them after a partial
2443 match. There is more discussion of this facility in the pcrepartial
2444 documentation.
2445
2446 Successful returns from pcre_dfa_exec()
2447
2448 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
2449 string in the subject. Note, however, that all the matches from one run
2450 of the function start at the same point in the subject. The shorter
2451 matches are all initial substrings of the longer matches. For example,
2452 if the pattern
2453
2454 <.*>
2455
2456 is matched against the string
2457
2458 This is <something> <something else> <something further> no more
2459
2460 the three matched strings are
2461
2462 <something>
2463 <something> <something else>
2464 <something> <something else> <something further>
2465
2466 On success, the yield of the function is a number greater than zero,
2467 which is the number of matched substrings. The substrings themselves
2468 are returned in ovector. Each string uses two elements; the first is
2469 the offset to the start, and the second is the offset to the end. In
2470 fact, all the strings have the same start offset. (Space could have
2471 been saved by giving this only once, but it was decided to retain some
2472 compatibility with the way pcre_exec() returns data, even though the
2473 meaning of the strings is different.)
2474
2475 The strings are returned in reverse order of length; that is, the long-
2476 est matching string is given first. If there were too many matches to
2477 fit into ovector, the yield of the function is zero, and the vector is
2478 filled with the longest matches.
2479
2480 Error returns from pcre_dfa_exec()
2481
2482 The pcre_dfa_exec() function returns a negative number when it fails.
2483 Many of the errors are the same as for pcre_exec(), and these are
2484 described above. There are in addition the following errors that are
2485 specific to pcre_dfa_exec():
2486
2487 PCRE_ERROR_DFA_UITEM (-16)
2488
2489 This return is given if pcre_dfa_exec() encounters an item in the pat-
2490 tern that it does not support, for instance, the use of \C or a back
2491 reference.
2492
2493 PCRE_ERROR_DFA_UCOND (-17)
2494
2495 This return is given if pcre_dfa_exec() encounters a condition item
2496 that uses a back reference for the condition, or a test for recursion
2497 in a specific group. These are not supported.
2498
2499 PCRE_ERROR_DFA_UMLIMIT (-18)
2500
2501 This return is given if pcre_dfa_exec() is called with an extra block
2502 that contains a setting of the match_limit field. This is not supported
2503 (it is meaningless).
2504
2505 PCRE_ERROR_DFA_WSSIZE (-19)
2506
2507 This return is given if pcre_dfa_exec() runs out of space in the
2508 workspace vector.
2509
2510 PCRE_ERROR_DFA_RECURSE (-20)
2511
2512 When a recursive subpattern is processed, the matching function calls
2513 itself recursively, using private vectors for ovector and workspace.
2514 This error is given if the output vector is not large enough. This
2515 should be extremely rare, as a vector of size 1000 is used.
2516
2517
2518 SEE ALSO
2519
2520 pcrebuild(3), pcrecallout(3), pcrecpp(3)(3), pcrematching(3), pcrepar-
2521 tial(3), pcreposix(3), pcreprecompile(3), pcresample(3), pcrestack(3).
2522
2523
2524 AUTHOR
2525
2526 Philip Hazel
2527 University Computing Service
2528 Cambridge CB2 3QH, England.
2529
2530
2531 REVISION
2532
2533 Last updated: 11 September 2007
2534 Copyright (c) 1997-2007 University of Cambridge.
2535 ------------------------------------------------------------------------------
2536
2537
2538 PCRECALLOUT(3) PCRECALLOUT(3)
2539
2540
2541 NAME
2542 PCRE - Perl-compatible regular expressions
2543
2544
2545 PCRE CALLOUTS
2546
2547 int (*pcre_callout)(pcre_callout_block *);
2548
2549 PCRE provides a feature called "callout", which is a means of temporar-
2550 ily passing control to the caller of PCRE in the middle of pattern
2551 matching. The caller of PCRE provides an external function by putting
2552 its entry point in the global variable pcre_callout. By default, this
2553 variable contains NULL, which disables all calling out.
2554
2555 Within a regular expression, (?C) indicates the points at which the
2556 external function is to be called. Different callout points can be
2557 identified by putting a number less than 256 after the letter C. The
2558 default value is zero. For example, this pattern has two callout
2559 points:
2560
2561 (?C1)abc(?C2)def
2562
2563 If the PCRE_AUTO_CALLOUT option bit is set when pcre_compile() is
2564 called, PCRE automatically inserts callouts, all with number 255,
2565 before each item in the pattern. For example, if PCRE_AUTO_CALLOUT is
2566 used with the pattern
2567
2568 A(\d{2}|--)
2569
2570 it is processed as if it were
2571
2572 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
2573
2574 Notice that there is a callout before and after each parenthesis and
2575 alternation bar. Automatic callouts can be used for tracking the
2576 progress of pattern matching. The pcretest command has an option that
2577 sets automatic callouts; when it is used, the output indicates how the
2578 pattern is matched. This is useful information when you are trying to
2579 optimize the performance of a particular pattern.
2580
2581
2582 MISSING CALLOUTS
2583
2584 You should be aware that, because of optimizations in the way PCRE
2585 matches patterns, callouts sometimes do not happen. For example, if the
2586 pattern is
2587
2588 ab(?C4)cd
2589
2590 PCRE knows that any matching string must contain the letter "d". If the
2591 subject string is "abyz", the lack of "d" means that matching doesn't
2592 ever start, and the callout is never reached. However, with "abyd",
2593 though the result is still no match, the callout is obeyed.
2594
2595
2596 THE CALLOUT INTERFACE
2597
2598 During matching, when PCRE reaches a callout point, the external func-
2599 tion defined by pcre_callout is called (if it is set). This applies to
2600 both the pcre_exec() and the pcre_dfa_exec() matching functions. The
2601 only argument to the callout function is a pointer to a pcre_callout
2602 block. This structure contains the following fields:
2603
2604 int version;
2605 int callout_number;
2606 int *offset_vector;
2607 const char *subject;
2608 int subject_length;
2609 int start_match;
2610 int current_position;
2611 int capture_top;
2612 int capture_last;
2613 void *callout_data;
2614 int pattern_position;
2615 int next_item_length;
2616
2617 The version field is an integer containing the version number of the
2618 block format. The initial version was 0; the current version is 1. The
2619 version number will change again in future if additional fields are
2620 added, but the intention is never to remove any of the existing fields.
2621
2622 The callout_number field contains the number of the callout, as com-
2623 piled into the pattern (that is, the number after ?C for manual call-
2624 outs, and 255 for automatically generated callouts).
2625
2626 The offset_vector field is a pointer to the vector of offsets that was
2627 passed by the caller to pcre_exec() or pcre_dfa_exec(). When
2628 pcre_exec() is used, the contents can be inspected in order to extract
2629 substrings that have been matched so far, in the same way as for
2630 extracting substrings after a match has completed. For pcre_dfa_exec()
2631 this field is not useful.
2632
2633 The subject and subject_length fields contain copies of the values that
2634 were passed to pcre_exec().
2635
2636 The start_match field normally contains the offset within the subject
2637 at which the current match attempt started. However, if the escape
2638 sequence \K has been encountered, this value is changed to reflect the
2639 modified starting point. If the pattern is not anchored, the callout
2640 function may be called several times from the same point in the pattern
2641 for different starting points in the subject.
2642
2643 The current_position field contains the offset within the subject of
2644 the current match pointer.
2645
2646 When the pcre_exec() function is used, the capture_top field contains
2647 one more than the number of the highest numbered captured substring so
2648 far. If no substrings have been captured, the value of capture_top is
2649 one. This is always the case when pcre_dfa_exec() is used, because it
2650 does not support captured substrings.
2651
2652 The capture_last field contains the number of the most recently cap-
2653 tured substring. If no substrings have been captured, its value is -1.
2654 This is always the case when pcre_dfa_exec() is used.
2655
2656 The callout_data field contains a value that is passed to pcre_exec()
2657 or pcre_dfa_exec() specifically so that it can be passed back in call-
2658 outs. It is passed in the pcre_callout field of the pcre_extra data
2659 structure. If no such data was passed, the value of callout_data in a
2660 pcre_callout block is NULL. There is a description of the pcre_extra
2661 structure in the pcreapi documentation.
2662
2663 The pattern_position field is present from version 1 of the pcre_call-
2664 out structure. It contains the offset to the next item to be matched in
2665 the pattern string.
2666
2667 The next_item_length field is present from version 1 of the pcre_call-
2668 out structure. It contains the length of the next item to be matched in
2669 the pattern string. When the callout immediately precedes an alterna-
2670 tion bar, a closing parenthesis, or the end of the pattern, the length
2671 is zero. When the callout precedes an opening parenthesis, the length
2672 is that of the entire subpattern.
2673
2674 The pattern_position and next_item_length fields are intended to help
2675 in distinguishing between different automatic callouts, which all have
2676 the same callout number. However, they are set for all callouts.
2677
2678
2679 RETURN VALUES
2680
2681 The external callout function returns an integer to PCRE. If the value
2682 is zero, matching proceeds as normal. If the value is greater than
2683 zero, matching fails at the current point, but the testing of other
2684 matching possibilities goes ahead, just as if a lookahead assertion had
2685 failed. If the value is less than zero, the match is abandoned, and
2686 pcre_exec() (or pcre_dfa_exec()) returns the negative value.
2687
2688 Negative values should normally be chosen from the set of
2689 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
2690 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
2691 reserved for use by callout functions; it will never be used by PCRE
2692 itself.
2693
2694
2695 AUTHOR
2696
2697 Philip Hazel
2698 University Computing Service
2699 Cambridge CB2 3QH, England.
2700
2701
2702 REVISION
2703
2704 Last updated: 29 May 2007
2705 Copyright (c) 1997-2007 University of Cambridge.
2706 ------------------------------------------------------------------------------
2707
2708
2709 PCRECOMPAT(3) PCRECOMPAT(3)
2710
2711
2712 NAME
2713 PCRE - Perl-compatible regular expressions
2714
2715
2716 DIFFERENCES BETWEEN PCRE AND PERL
2717
2718 This document describes the differences in the ways that PCRE and Perl
2719 handle regular expressions. The differences described here are mainly
2720 with respect to Perl 5.8, though PCRE versions 7.0 and later contain
2721 some features that are expected to be in the forthcoming Perl 5.10.
2722
2723 1. PCRE has only a subset of Perl's UTF-8 and Unicode support. Details
2724 of what it does have are given in the section on UTF-8 support in the
2725 main pcre page.
2726
2727 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl
2728 permits them, but they do not mean what you might think. For example,
2729 (?!a){3} does not assert that the next three characters are not "a". It
2730 just asserts that the next character is not "a" three times.
2731
2732 3. Capturing subpatterns that occur inside negative lookahead asser-
2733 tions are counted, but their entries in the offsets vector are never
2734 set. Perl sets its numerical variables from any such patterns that are
2735 matched before the assertion fails to match something (thereby succeed-
2736 ing), but only if the negative lookahead assertion contains just one
2737 branch.
2738
2739 4. Though binary zero characters are supported in the subject string,
2740 they are not allowed in a pattern string because it is passed as a nor-
2741 mal C string, terminated by zero. The escape sequence \0 can be used in
2742 the pattern to represent a binary zero.
2743
2744 5. The following Perl escape sequences are not supported: \l, \u, \L,
2745 \U, and \N. In fact these are implemented by Perl's general string-han-
2746 dling and are not part of its pattern matching engine. If any of these
2747 are encountered by PCRE, an error is generated.
2748
2749 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
2750 is built with Unicode character property support. The properties that
2751 can be tested with \p and \P are limited to the general category prop-
2752 erties such as Lu and Nd, script names such as Greek or Han, and the
2753 derived properties Any and L&.
2754
2755 7. PCRE does support the \Q...\E escape for quoting substrings. Charac-
2756 ters in between are treated as literals. This is slightly different
2757 from Perl in that $ and @ are also handled as literals inside the
2758 quotes. In Perl, they cause variable interpolation (but of course PCRE
2759 does not have variables). Note the following examples:
2760
2761 Pattern PCRE matches Perl matches
2762
2763 \Qabc$xyz\E abc$xyz abc followed by the
2764 contents of $xyz
2765 \Qabc\$xyz\E abc\$xyz abc\$xyz
2766 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2767
2768 The \Q...\E sequence is recognized both inside and outside character
2769 classes.
2770
2771 8. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
2772 constructions. However, there is support for recursive patterns. This
2773 is not available in Perl 5.8, but will be in Perl 5.10. Also, the PCRE
2774 "callout" feature allows an external function to be called during pat-
2775 tern matching. See the pcrecallout documentation for details.
2776
2777 9. Subpatterns that are called recursively or as "subroutines" are
2778 always treated as atomic groups in PCRE. This is like Python, but
2779 unlike Perl.
2780
2781 10. There are some differences that are concerned with the settings of
2782 captured strings when part of a pattern is repeated. For example,
2783 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
2784 unset, but in PCRE it is set to "b".
2785
2786 11. PCRE does support Perl 5.10's backtracking verbs (*ACCEPT),
2787 (*FAIL), (*F), (*COMMIT), (*PRUNE), (*SKIP), and (*THEN), but only in
2788 the forms without an argument. PCRE does not support (*MARK). If
2789 (*ACCEPT) is within capturing parentheses, PCRE does not set that cap-
2790 ture group; this is different to Perl.
2791
2792 12. PCRE provides some extensions to the Perl regular expression facil-
2793 ities. Perl 5.10 will include new features that are not in earlier
2794 versions, some of which (such as named parentheses) have been in PCRE
2795 for some time. This list is with respect to Perl 5.10:
2796
2797 (a) Although lookbehind assertions must match fixed length strings,
2798 each alternative branch of a lookbehind assertion can match a different
2799 length of string. Perl requires them all to have the same length.
2800
2801 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
2802 meta-character matches only at the very end of the string.
2803
2804 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
2805 cial meaning is faulted. Otherwise, like Perl, the backslash is quietly
2806 ignored. (Perl can be made to issue a warning.)
2807
2808 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
2809 fiers is inverted, that is, by default they are not greedy, but if fol-
2810 lowed by a question mark they are.
2811
2812 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
2813 tried only at the first matching position in the subject string.
2814
2815 (f) The PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, and PCRE_NO_AUTO_CAP-
2816 TURE options for pcre_exec() have no Perl equivalents.
2817
2818 (g) The \R escape sequence can be restricted to match only CR, LF, or
2819 CRLF by the PCRE_BSR_ANYCRLF option.
2820
2821 (h) The callout facility is PCRE-specific.
2822
2823 (i) The partial matching facility is PCRE-specific.
2824
2825 (j) Patterns compiled by PCRE can be saved and re-used at a later time,
2826 even on different hosts that have the other endianness.
2827
2828 (k) The alternative matching function (pcre_dfa_exec()) matches in a
2829 different way and is not Perl-compatible.
2830
2831 (l) PCRE recognizes some special sequences such as (*CR) at the start
2832 of a pattern that set overall options that cannot be changed within the
2833 pattern.
2834
2835
2836 AUTHOR
2837
2838 Philip Hazel
2839 University Computing Service
2840 Cambridge CB2 3QH, England.
2841
2842
2843 REVISION
2844
2845 Last updated: 11 September 2007
2846 Copyright (c) 1997-2007 University of Cambridge.
2847 ------------------------------------------------------------------------------
2848
2849
2850 PCREPATTERN(3) PCREPATTERN(3)
2851
2852
2853 NAME
2854 PCRE - Perl-compatible regular expressions
2855
2856
2857 PCRE REGULAR EXPRESSION DETAILS
2858
2859 The syntax and semantics of the regular expressions that are supported
2860 by PCRE are described in detail below. There is a quick-reference syn-
2861 tax summary in the pcresyntax page. Perl's regular expressions are
2862 described in its own documentation, and regular expressions in general
2863 are covered in a number of books, some of which have copious examples.
2864 Jeffrey Friedl's "Mastering Regular Expressions", published by
2865 O'Reilly, covers regular expressions in great detail. This description
2866 of PCRE's regular expressions is intended as reference material.
2867
2868 The original operation of PCRE was on strings of one-byte characters.
2869 However, there is now also support for UTF-8 character strings. To use
2870 this, you must build PCRE to include UTF-8 support, and then call
2871 pcre_compile() with the PCRE_UTF8 option. How this affects pattern
2872 matching is mentioned in several places below. There is also a summary
2873 of UTF-8 features in the section on UTF-8 support in the main pcre
2874 page.
2875
2876 The remainder of this document discusses the patterns that are sup-
2877 ported by PCRE when its main matching function, pcre_exec(), is used.
2878 From release 6.0, PCRE offers a second matching function,
2879 pcre_dfa_exec(), which matches using a different algorithm that is not
2880 Perl-compatible. Some of the features discussed below are not available
2881 when pcre_dfa_exec() is used. The advantages and disadvantages of the
2882 alternative function, and how it differs from the normal function, are
2883 discussed in the pcrematching page.
2884
2885
2886 NEWLINE CONVENTIONS
2887
2888 PCRE supports five different conventions for indicating line breaks in
2889 strings: a single CR (carriage return) character, a single LF (line-
2890 feed) character, the two-character sequence CRLF, any of the three pre-
2891 ceding, or any Unicode newline sequence. The pcreapi page has further
2892 discussion about newlines, and shows how to set the newline convention
2893 in the options arguments for the compiling and matching functions.
2894
2895 It is also possible to specify a newline convention by starting a pat-
2896 tern string with one of the following five sequences:
2897
2898 (*CR) carriage return
2899 (*LF) linefeed
2900 (*CRLF) carriage return, followed by linefeed
2901 (*ANYCRLF) any of the three above
2902 (*ANY) all Unicode newline sequences
2903
2904 These override the default and the options given to pcre_compile(). For
2905 example, on a Unix system where LF is the default newline sequence, the
2906 pattern
2907
2908 (*CR)a.b
2909
2910 changes the convention to CR. That pattern matches "a\nb" because LF is
2911 no longer a newline. Note that these special settings, which are not
2912 Perl-compatible, are recognized only at the very start of a pattern,
2913 and that they must be in upper case. If more than one of them is
2914 present, the last one is used.
2915
2916 The newline convention does not affect what the \R escape sequence
2917 matches. By default, this is any Unicode newline sequence, for Perl
2918 compatibility. However, this can be changed; see the description of \R
2919 in the section entitled "Newline sequences" below. A change of \R set-
2920 ting can be combined with a change of newline convention.
2921
2922
2923 CHARACTERS AND METACHARACTERS
2924
2925 A regular expression is a pattern that is matched against a subject
2926 string from left to right. Most characters stand for themselves in a
2927 pattern, and match the corresponding characters in the subject. As a
2928 trivial example, the pattern
2929
2930 The quick brown fox
2931
2932 matches a portion of a subject string that is identical to itself. When
2933 caseless matching is specified (the PCRE_CASELESS option), letters are
2934 matched independently of case. In UTF-8 mode, PCRE always understands
2935 the concept of case for characters whose values are less than 128, so
2936 caseless matching is always possible. For characters with higher val-
2937 ues, the concept of case is supported if PCRE is compiled with Unicode
2938 property support, but not otherwise. If you want to use caseless
2939 matching for characters 128 and above, you must ensure that PCRE is
2940 compiled with Unicode property support as well as with UTF-8 support.
2941
2942 The power of regular expressions comes from the ability to include
2943 alternatives and repetitions in the pattern. These are encoded in the
2944 pattern by the use of metacharacters, which do not stand for themselves
2945 but instead are interpreted in some special way.
2946
2947 There are two different sets of metacharacters: those that are recog-
2948 nized anywhere in the pattern except within square brackets, and those
2949 that are recognized within square brackets. Outside square brackets,
2950 the metacharacters are as follows:
2951
2952 \ general escape character with several uses
2953 ^ assert start of string (or line, in multiline mode)
2954 $ assert end of string (or line, in multiline mode)
2955 . match any character except newline (by default)
2956 [ start character class definition
2957 | start of alternative branch
2958 ( start subpattern
2959 ) end subpattern
2960 ? extends the meaning of (
2961 also 0 or 1 quantifier
2962 also quantifier minimizer
2963 * 0 or more quantifier
2964 + 1 or more quantifier
2965 also "possessive quantifier"
2966 { start min/max quantifier
2967
2968 Part of a pattern that is in square brackets is called a "character
2969 class". In a character class the only metacharacters are:
2970
2971 \ general escape character
2972 ^ negate the class, but only if the first character
2973 - indicates character range
2974 [ POSIX character class (only if followed by POSIX
2975 syntax)
2976 ] terminates the character class
2977
2978 The following sections describe the use of each of the metacharacters.
2979
2980
2981 BACKSLASH
2982
2983 The backslash character has several uses. Firstly, if it is followed by
2984 a non-alphanumeric character, it takes away any special meaning that
2985 character may have. This use of backslash as an escape character
2986 applies both inside and outside character classes.
2987
2988 For example, if you want to match a * character, you write \* in the
2989 pattern. This escaping action applies whether or not the following
2990 character would otherwise be interpreted as a metacharacter, so it is
2991 always safe to precede a non-alphanumeric with backslash to specify
2992 that it stands for itself. In particular, if you want to match a back-
2993 slash, you write \\.
2994
2995 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
2996 the pattern (other than in a character class) and characters between a
2997 # outside a character class and the next newline are ignored. An escap-
2998 ing backslash can be used to include a whitespace or # character as
2999 part of the pattern.
3000
3001 If you want to remove the special meaning from a sequence of charac-
3002 ters, you can do so by putting them between \Q and \E. This is differ-
3003 ent from Perl in that $ and @ are handled as literals in \Q...\E
3004 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
3005 tion. Note the following examples:
3006
3007 Pattern PCRE matches Perl matches
3008
3009 \Qabc$xyz\E abc$xyz abc followed by the
3010 contents of $xyz
3011 \Qabc\$xyz\E abc\$xyz abc\$xyz
3012 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3013
3014 The \Q...\E sequence is recognized both inside and outside character
3015 classes.
3016
3017 Non-printing characters
3018
3019 A second use of backslash provides a way of encoding non-printing char-
3020 acters in patterns in a visible manner. There is no restriction on the
3021 appearance of non-printing characters, apart from the binary zero that
3022 terminates a pattern, but when a pattern is being prepared by text
3023 editing, it is usually easier to use one of the following escape
3024 sequences than the binary character it represents:
3025
3026 \a alarm, that is, the BEL character (hex 07)
3027 \cx "control-x", where x is any character
3028 \e escape (hex 1B)
3029 \f formfeed (hex 0C)
3030 \n linefeed (hex 0A)
3031 \r carriage return (hex 0D)
3032 \t tab (hex 09)
3033 \ddd character with octal code ddd, or backreference
3034 \xhh character with hex code hh
3035 \x{hhh..} character with hex code hhh..
3036
3037 The precise effect of \cx is as follows: if x is a lower case letter,
3038 it is converted to upper case. Then bit 6 of the character (hex 40) is
3039 inverted. Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c;
3040 becomes hex 7B.
3041
3042 After \x, from zero to two hexadecimal digits are read (letters can be
3043 in upper or lower case). Any number of hexadecimal digits may appear
3044 between \x{ and }, but the value of the character code must be less
3045 than 256 in non-UTF-8 mode, and less than 2**31 in UTF-8 mode. That is,
3046 the maximum value in hexadecimal is 7FFFFFFF. Note that this is bigger
3047 than the largest Unicode code point, which is 10FFFF.
3048
3049 If characters other than hexadecimal digits appear between \x{ and },
3050 or if there is no terminating }, this form of escape is not recognized.
3051 Instead, the initial \x will be interpreted as a basic hexadecimal
3052 escape, with no following digits, giving a character whose value is
3053 zero.
3054
3055 Characters whose value is less than 256 can be defined by either of the
3056 two syntaxes for \x. There is no difference in the way they are han-
3057 dled. For example, \xdc is exactly the same as \x{dc}.
3058
3059 After \0 up to two further octal digits are read. If there are fewer
3060 than two digits, just those that are present are used. Thus the
3061 sequence \0\x\07 specifies two binary zeros followed by a BEL character
3062 (code value 7). Make sure you supply two digits after the initial zero
3063 if the pattern character that follows is itself an octal digit.
3064
3065 The handling of a backslash followed by a digit other than 0 is compli-
3066 cated. Outside a character class, PCRE reads it and any following dig-
3067 its as a decimal number. If the number is less than 10, or if there
3068 have been at least that many previous capturing left parentheses in the
3069 expression, the entire sequence is taken as a back reference. A
3070 description of how this works is given later, following the discussion
3071 of parenthesized subpatterns.
3072
3073 Inside a character class, or if the decimal number is greater than 9
3074 and there have not been that many capturing subpatterns, PCRE re-reads
3075 up to three octal digits following the backslash, and uses them to gen-
3076 erate a data character. Any subsequent digits stand for themselves. In
3077 non-UTF-8 mode, the value of a character specified in octal must be
3078 less than \400. In UTF-8 mode, values up to \777 are permitted. For
3079 example:
3080
3081 \040 is another way of writing a space
3082 \40 is the same, provided there are fewer than 40
3083 previous capturing subpatterns
3084 \7 is always a back reference
3085 \11 might be a back reference, or another way of
3086 writing a tab
3087 \011 is always a tab
3088 \0113 is a tab followed by the character "3"
3089 \113 might be a back reference, otherwise the
3090 character with octal code 113
3091 \377 might be a back reference, otherwise
3092 the byte consisting entirely of 1 bits
3093 \81 is either a back reference, or a binary zero
3094 followed by the two characters "8" and "1"
3095
3096 Note that octal values of 100 or greater must not be introduced by a
3097 leading zero, because no more than three octal digits are ever read.
3098
3099 All the sequences that define a single character value can be used both
3100 inside and outside character classes. In addition, inside a character
3101 class, the sequence \b is interpreted as the backspace character (hex
3102 08), and the sequences \R and \X are interpreted as the characters "R"
3103 and "X", respectively. Outside a character class, these sequences have
3104 different meanings (see below).
3105
3106 Absolute and relative back references
3107
3108 The sequence \g followed by an unsigned or a negative number, option-
3109 ally enclosed in braces, is an absolute or relative back reference. A
3110 named back reference can be coded as \g{name}. Back references are dis-
3111 cussed later, following the discussion of parenthesized subpatterns.
3112
3113 Generic character types
3114
3115 Another use of backslash is for specifying generic character types. The
3116 following are always recognized:
3117
3118 \d any decimal digit
3119 \D any character that is not a decimal digit
3120 \h any horizontal whitespace character
3121 \H any character that is not a horizontal whitespace character
3122 \s any whitespace character
3123 \S any character that is not a whitespace character
3124 \v any vertical whitespace character
3125 \V any character that is not a vertical whitespace character
3126 \w any "word" character
3127 \W any "non-word" character
3128
3129 Each pair of escape sequences partitions the complete set of characters
3130 into two disjoint sets. Any given character matches one, and only one,
3131 of each pair.
3132
3133 These character type sequences can appear both inside and outside char-
3134 acter classes. They each match one character of the appropriate type.
3135 If the current matching point is at the end of the subject string, all
3136 of them fail, since there is no character to match.
3137
3138 For compatibility with Perl, \s does not match the VT character (code
3139 11). This makes it different from the the POSIX "space" class. The \s
3140 characters are HT (9), LF (10), FF (12), CR (13), and space (32). If
3141 "use locale;" is included in a Perl script, \s may match the VT charac-
3142 ter. In PCRE, it never does.
3143
3144 In UTF-8 mode, characters with values greater than 128 never match \d,
3145 \s, or \w, and always match \D, \S, and \W. This is true even when Uni-
3146 code character property support is available. These sequences retain
3147 their original meanings from before UTF-8 support was available, mainly
3148 for efficiency reasons.
3149
3150 The sequences \h, \H, \v, and \V are Perl 5.10 features. In contrast to
3151 the other sequences, these do match certain high-valued codepoints in
3152 UTF-8 mode. The horizontal space characters are:
3153
3154 U+0009 Horizontal tab
3155 U+0020 Space
3156 U+00A0 Non-break space
3157 U+1680 Ogham space mark
3158 U+180E Mongolian vowel separator
3159 U+2000 En quad
3160 U+2001 Em quad
3161 U+2002 En space
3162 U+2003 Em space
3163 U+2004 Three-per-em space
3164 U+2005 Four-per-em space
3165 U+2006 Six-per-em space
3166 U+2007 Figure space
3167 U+2008 Punctuation space
3168 U+2009 Thin space
3169 U+200A Hair space
3170 U+202F Narrow no-break space
3171 U+205F Medium mathematical space
3172 U+3000 Ideographic space
3173
3174 The vertical space characters are:
3175
3176 U+000A Linefeed
3177 U+000B Vertical tab
3178 U+000C Formfeed
3179 U+000D Carriage return
3180 U+0085 Next line
3181 U+2028 Line separator
3182 U+2029 Paragraph separator
3183
3184 A "word" character is an underscore or any character less than 256 that
3185 is a letter or digit. The definition of letters and digits is con-
3186 trolled by PCRE's low-valued character tables, and may vary if locale-
3187 specific matching is taking place (see "Locale support" in the pcreapi
3188 page). For example, in a French locale such as "fr_FR" in Unix-like
3189 systems, or "french" in Windows, some character codes greater than 128
3190 are used for accented letters, and these are matched by \w. The use of
3191 locales with Unicode is discouraged.
3192
3193 Newline sequences
3194
3195 Outside a character class, by default, the escape sequence \R matches
3196 any Unicode newline sequence. This is a Perl 5.10 feature. In non-UTF-8
3197 mode \R is equivalent to the following:
3198
3199 (?>\r\n|\n|\x0b|\f|\r|\x85)
3200
3201 This is an example of an "atomic group", details of which are given
3202 below. This particular group matches either the two-character sequence
3203 CR followed by LF, or one of the single characters LF (linefeed,
3204 U+000A), VT (vertical tab, U+000B), FF (formfeed, U+000C), CR (carriage
3205 return, U+000D), or NEL (next line, U+0085). The two-character sequence
3206 is treated as a single unit that cannot be split.
3207
3208 In UTF-8 mode, two additional characters whose codepoints are greater
3209 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
3210 rator, U+2029). Unicode character property support is not needed for
3211 these characters to be recognized.
3212
3213 It is possible to restrict \R to match only CR, LF, or CRLF (instead of
3214 the complete set of Unicode line endings) by setting the option
3215 PCRE_BSR_ANYCRLF either at compile time or when the pattern is matched.
3216 (BSR is an abbrevation for "backslash R".) This can be made the default
3217 when PCRE is built; if this is the case, the other behaviour can be
3218 requested via the PCRE_BSR_UNICODE option. It is also possible to
3219 specify these settings by starting a pattern string with one of the
3220 following sequences:
3221
3222 (*BSR_ANYCRLF) CR, LF, or CRLF only
3223 (*BSR_UNICODE) any Unicode newline sequence
3224
3225 These override the default and the options given to pcre_compile(), but
3226 they can be overridden by options given to pcre_exec(). Note that these
3227 special settings, which are not Perl-compatible, are recognized only at
3228 the very start of a pattern, and that they must be in upper case. If
3229 more than one of them is present, the last one is used. They can be
3230 combined with a change of newline convention, for example, a pattern
3231 can start with:
3232
3233 (*ANY)(*BSR_ANYCRLF)
3234
3235 Inside a character class, \R matches the letter "R".
3236
3237 Unicode character properties
3238
3239 When PCRE is built with Unicode character property support, three addi-
3240 tional escape sequences that match characters with specific properties
3241 are available. When not in UTF-8 mode, these sequences are of course
3242 limited to testing characters whose codepoints are less than 256, but
3243 they do work in this mode. The extra escape sequences are:
3244
3245 \p{xx} a character with the xx property
3246 \P{xx} a character without the xx property
3247 \X an extended Unicode sequence
3248
3249 The property names represented by xx above are limited to the Unicode
3250 script names, the general category properties, and "Any", which matches
3251 any character (including newline). Other properties such as "InMusical-
3252 Symbols" are not currently supported by PCRE. Note that \P{Any} does
3253 not match any characters, so always causes a match failure.
3254
3255 Sets of Unicode characters are defined as belonging to certain scripts.
3256 A character from one of these sets can be matched using a script name.
3257 For example:
3258
3259 \p{Greek}
3260 \P{Han}
3261
3262 Those that are not part of an identified script are lumped together as
3263 "Common". The current list of scripts is:
3264
3265 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
3266 Buhid, Canadian_Aboriginal, Cherokee, Common, Coptic, Cuneiform,
3267 Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian, Glagolitic,
3268 Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
3269 gana, Inherited, Kannada, Katakana, Kharoshthi, Khmer, Lao, Latin,
3270 Limbu, Linear_B, Malayalam, Mongolian, Myanmar, New_Tai_Lue, Nko,
3271 Ogham, Old_Italic, Old_Persian, Oriya, Osmanya, Phags_Pa, Phoenician,
3272 Runic, Shavian, Sinhala, Syloti_Nagri, Syriac, Tagalog, Tagbanwa,
3273 Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Yi.
3274
3275 Each character has exactly one general category property, specified by
3276 a two-letter abbreviation. For compatibility with Perl, negation can be
3277 specified by including a circumflex between the opening brace and the
3278 property name. For example, \p{^Lu} is the same as \P{Lu}.
3279
3280 If only one letter is specified with \p or \P, it includes all the gen-
3281 eral category properties that start with that letter. In this case, in
3282 the absence of negation, the curly brackets in the escape sequence are
3283 optional; these two examples have the same effect:
3284
3285 \p{L}
3286 \pL
3287
3288 The following general category property codes are supported:
3289
3290 C Other
3291 Cc Control
3292 Cf Format
3293 Cn Unassigned
3294 Co Private use
3295 Cs Surrogate
3296
3297 L Letter
3298 Ll Lower case letter
3299 Lm Modifier letter
3300 Lo Other letter
3301 Lt Title case letter
3302 Lu Upper case letter
3303
3304 M Mark
3305 Mc Spacing mark
3306 Me Enclosing mark
3307 Mn Non-spacing mark
3308
3309 N Number
3310 Nd Decimal number
3311 Nl Letter number
3312 No Other number
3313
3314 P Punctuation
3315 Pc Connector punctuation
3316 Pd Dash punctuation
3317 Pe Close punctuation
3318 Pf Final punctuation
3319 Pi Initial punctuation
3320 Po Other punctuation
3321 Ps Open punctuation
3322
3323 S Symbol
3324 Sc Currency symbol
3325 Sk Modifier symbol
3326 Sm Mathematical symbol
3327 So Other symbol
3328
3329 Z Separator
3330 Zl Line separator
3331 Zp Paragraph separator
3332 Zs Space separator
3333
3334 The special property L& is also supported: it matches a character that
3335 has the Lu, Ll, or Lt property, in other words, a letter that is not
3336 classified as a modifier or "other".
3337
3338 The Cs (Surrogate) property applies only to characters in the range
3339 U+D800 to U+DFFF. Such characters are not valid in UTF-8 strings (see
3340 RFC 3629) and so cannot be tested by PCRE, unless UTF-8 validity check-
3341 ing has been turned off (see the discussion of PCRE_NO_UTF8_CHECK in
3342 the pcreapi page).
3343
3344 The long synonyms for these properties that Perl supports (such as
3345 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
3346 any of these properties with "Is".
3347
3348 No character that is in the Unicode table has the Cn (unassigned) prop-
3349 erty. Instead, this property is assumed for any code point that is not
3350 in the Unicode table.
3351
3352 Specifying caseless matching does not affect these escape sequences.
3353 For example, \p{Lu} always matches only upper case letters.
3354
3355 The \X escape matches any number of Unicode characters that form an
3356 extended Unicode sequence. \X is equivalent to
3357
3358 (?>\PM\pM*)
3359
3360 That is, it matches a character without the "mark" property, followed
3361 by zero or more characters with the "mark" property, and treats the
3362 sequence as an atomic group (see below). Characters with the "mark"
3363 property are typically accents that affect the preceding character.
3364 None of them have codepoints less than 256, so in non-UTF-8 mode \X
3365 matches any one character.
3366
3367 Matching characters by Unicode property is not fast, because PCRE has
3368 to search a structure that contains data for over fifteen thousand
3369 characters. That is why the traditional escape sequences such as \d and
3370 \w do not use Unicode properties in PCRE.
3371
3372 Resetting the match start
3373
3374 The escape sequence \K, which is a Perl 5.10 feature, causes any previ-
3375 ously matched characters not to be included in the final matched
3376 sequence. For example, the pattern:
3377
3378 foo\Kbar
3379
3380 matches "foobar", but reports that it has matched "bar". This feature
3381 is similar to a lookbehind assertion (described below). However, in
3382 this case, the part of the subject before the real match does not have
3383 to be of fixed length, as lookbehind assertions do. The use of \K does
3384 not interfere with the setting of captured substrings. For example,
3385 when the pattern
3386
3387 (foo)\Kbar
3388
3389 matches "foobar", the first substring is still set to "foo".
3390
3391 Simple assertions
3392
3393 The final use of backslash is for certain simple assertions. An asser-
3394 tion specifies a condition that has to be met at a particular point in
3395 a match, without consuming any characters from the subject string. The
3396 use of subpatterns for more complicated assertions is described below.
3397 The backslashed assertions are:
3398
3399 \b matches at a word boundary
3400 \B matches when not at a word boundary
3401 \A matches at the start of the subject
3402 \Z matches at the end of the subject
3403 also matches before a newline at the end of the subject
3404 \z matches only at the end of the subject
3405 \G matches at the first matching position in the subject
3406
3407 These assertions may not appear in character classes (but note that \b
3408 has a different meaning, namely the backspace character, inside a char-
3409 acter class).
3410
3411 A word boundary is a position in the subject string where the current
3412 character and the previous character do not both match \w or \W (i.e.
3413 one matches \w and the other matches \W), or the start or end of the
3414 string if the first or last character matches \w, respectively.
3415
3416 The \A, \Z, and \z assertions differ from the traditional circumflex
3417 and dollar (described in the next section) in that they only ever match
3418 at the very start and end of the subject string, whatever options are
3419 set. Thus, they are independent of multiline mode. These three asser-
3420 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
3421 affect only the behaviour of the circumflex and dollar metacharacters.
3422 However, if the startoffset argument of pcre_exec() is non-zero, indi-
3423 cating that matching is to start at a point other than the beginning of
3424 the subject, \A can never match. The difference between \Z and \z is
3425 that \Z matches before a newline at the end of the string as well as at
3426 the very end, whereas \z matches only at the end.
3427
3428 The \G assertion is true only when the current matching position is at
3429 the start point of the match, as specified by the startoffset argument
3430 of pcre_exec(). It differs from \A when the value of startoffset is
3431 non-zero. By calling pcre_exec() multiple times with appropriate argu-
3432 ments, you can mimic Perl's /g option, and it is in this kind of imple-
3433 mentation where \G can be useful.
3434
3435 Note, however, that PCRE's interpretation of \G, as the start of the
3436 current match, is subtly different from Perl's, which defines it as the
3437 end of the previous match. In Perl, these can be different when the
3438 previously matched string was empty. Because PCRE does just one match
3439 at a time, it cannot reproduce this behaviour.
3440
3441 If all the alternatives of a pattern begin with \G, the expression is
3442 anchored to the starting match position, and the "anchored" flag is set
3443 in the compiled regular expression.
3444
3445
3446 CIRCUMFLEX AND DOLLAR
3447
3448 Outside a character class, in the default matching mode, the circumflex
3449 character is an assertion that is true only if the current matching
3450 point is at the start of the subject string. If the startoffset argu-
3451 ment of pcre_exec() is non-zero, circumflex can never match if the
3452 PCRE_MULTILINE option is unset. Inside a character class, circumflex
3453 has an entirely different meaning (see below).
3454
3455 Circumflex need not be the first character of the pattern if a number
3456 of alternatives are involved, but it should be the first thing in each
3457 alternative in which it appears if the pattern is ever to match that
3458 branch. If all possible alternatives start with a circumflex, that is,
3459 if the pattern is constrained to match only at the start of the sub-
3460 ject, it is said to be an "anchored" pattern. (There are also other
3461 constructs that can cause a pattern to be anchored.)
3462
3463 A dollar character is an assertion that is true only if the current
3464 matching point is at the end of the subject string, or immediately
3465 before a newline at the end of the string (by default). Dollar need not
3466 be the last character of the pattern if a number of alternatives are
3467 involved, but it should be the last item in any branch in which it
3468 appears. Dollar has no special meaning in a character class.
3469
3470 The meaning of dollar can be changed so that it matches only at the
3471 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
3472 compile time. This does not affect the \Z assertion.
3473
3474 The meanings of the circumflex and dollar characters are changed if the
3475 PCRE_MULTILINE option is set. When this is the case, a circumflex
3476 matches immediately after internal newlines as well as at the start of
3477 the subject string. It does not match after a newline that ends the
3478 string. A dollar matches before any newlines in the string, as well as
3479 at the very end, when PCRE_MULTILINE is set. When newline is specified
3480 as the two-character sequence CRLF, isolated CR and LF characters do
3481 not indicate newlines.
3482
3483 For example, the pattern /^abc$/ matches the subject string "def\nabc"
3484 (where \n represents a newline) in multiline mode, but not otherwise.
3485 Consequently, patterns that are anchored in single line mode because
3486 all branches start with ^ are not anchored in multiline mode, and a
3487 match for circumflex is possible when the startoffset argument of
3488 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
3489 PCRE_MULTILINE is set.
3490
3491 Note that the sequences \A, \Z, and \z can be used to match the start
3492 and end of the subject in both modes, and if all branches of a pattern
3493 start with \A it is always anchored, whether or not PCRE_MULTILINE is
3494 set.
3495
3496
3497 FULL STOP (PERIOD, DOT)
3498
3499 Outside a character class, a dot in the pattern matches any one charac-
3500 ter in the subject string except (by default) a character that signi-
3501 fies the end of a line. In UTF-8 mode, the matched character may be
3502 more than one byte long.
3503
3504 When a line ending is defined as a single character, dot never matches
3505 that character; when the two-character sequence CRLF is used, dot does
3506 not match CR if it is immediately followed by LF, but otherwise it
3507 matches all characters (including isolated CRs and LFs). When any Uni-
3508 code line endings are being recognized, dot does not match CR or LF or
3509 any of the other line ending characters.
3510
3511 The behaviour of dot with regard to newlines can be changed. If the
3512 PCRE_DOTALL option is set, a dot matches any one character, without
3513 exception. If the two-character sequence CRLF is present in the subject
3514 string, it takes two dots to match it.
3515
3516 The handling of dot is entirely independent of the handling of circum-
3517 flex and dollar, the only relationship being that they both involve
3518 newlines. Dot has no special meaning in a character class.
3519
3520
3521 MATCHING A SINGLE BYTE
3522
3523 Outside a character class, the escape sequence \C matches any one byte,
3524 both in and out of UTF-8 mode. Unlike a dot, it always matches any
3525 line-ending characters. The feature is provided in Perl in order to
3526 match individual bytes in UTF-8 mode. Because it breaks up UTF-8 char-
3527 acters into individual bytes, what remains in the string may be a mal-
3528 formed UTF-8 string. For this reason, the \C escape sequence is best
3529 avoided.
3530
3531 PCRE does not allow \C to appear in lookbehind assertions (described
3532 below), because in UTF-8 mode this would make it impossible to calcu-
3533 late the length of the lookbehind.
3534
3535
3536 SQUARE BRACKETS AND CHARACTER CLASSES
3537
3538 An opening square bracket introduces a character class, terminated by a
3539 closing square bracket. A closing square bracket on its own is not spe-
3540 cial. If a closing square bracket is required as a member of the class,
3541 it should be the first data character in the class (after an initial
3542 circumflex, if present) or escaped with a backslash.
3543
3544 A character class matches a single character in the subject. In UTF-8
3545 mode, the character may occupy more than one byte. A matched character
3546 must be in the set of characters defined by the class, unless the first
3547 character in the class definition is a circumflex, in which case the
3548 subject character must not be in the set defined by the class. If a
3549 circumflex is actually required as a member of the class, ensure it is
3550 not the first character, or escape it with a backslash.
3551
3552 For example, the character class [aeiou] matches any lower case vowel,
3553 while [^aeiou] matches any character that is not a lower case vowel.
3554 Note that a circumflex is just a convenient notation for specifying the
3555 characters that are in the class by enumerating those that are not. A
3556 class that starts with a circumflex is not an assertion: it still con-
3557 sumes a character from the subject string, and therefore it fails if
3558 the current pointer is at the end of the string.
3559
3560 In UTF-8 mode, characters with values greater than 255 can be included
3561 in a class as a literal string of bytes, or by using the \x{ escaping
3562 mechanism.
3563
3564 When caseless matching is set, any letters in a class represent both
3565 their upper case and lower case versions, so for example, a caseless
3566 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
3567 match "A", whereas a caseful version would. In UTF-8 mode, PCRE always
3568 understands the concept of case for characters whose values are less
3569 than 128, so caseless matching is always possible. For characters with
3570 higher values, the concept of case is supported if PCRE is compiled
3571 with Unicode property support, but not otherwise. If you want to use
3572 caseless matching for characters 128 and above, you must ensure that
3573 PCRE is compiled with Unicode property support as well as with UTF-8
3574 support.
3575
3576 Characters that might indicate line breaks are never treated in any
3577 special way when matching character classes, whatever line-ending
3578 sequence is in use, and whatever setting of the PCRE_DOTALL and
3579 PCRE_MULTILINE options is used. A class such as [^a] always matches one
3580 of these characters.
3581
3582 The minus (hyphen) character can be used to specify a range of charac-
3583 ters in a character class. For example, [d-m] matches any letter
3584 between d and m, inclusive. If a minus character is required in a
3585 class, it must be escaped with a backslash or appear in a position
3586 where it cannot be interpreted as indicating a range, typically as the
3587 first or last character in the class.
3588
3589 It is not possible to have the literal character "]" as the end charac-
3590 ter of a range. A pattern such as [W-]46] is interpreted as a class of
3591 two characters ("W" and "-") followed by a literal string "46]", so it
3592 would match "W46]" or "-46]". However, if the "]" is escaped with a
3593 backslash it is interpreted as the end of range, so [W-\]46] is inter-
3594 preted as a class containing a range followed by two other characters.
3595 The octal or hexadecimal representation of "]" can also be used to end
3596 a range.
3597
3598 Ranges operate in the collating sequence of character values. They can
3599 also be used for characters specified numerically, for example
3600 [\000-\037]. In UTF-8 mode, ranges can include characters whose values
3601 are greater than 255, for example [\x{100}-\x{2ff}].
3602
3603 If a range that includes letters is used when caseless matching is set,
3604 it matches the letters in either case. For example, [W-c] is equivalent
3605 to [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if
3606 character tables for a French locale are in use, [\xc8-\xcb] matches
3607 accented E characters in both cases. In UTF-8 mode, PCRE supports the
3608 concept of case for characters with values greater than 128 only when
3609 it is compiled with Unicode property support.
3610
3611 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
3612 in a character class, and add the characters that they match to the
3613 class. For example, [\dABCDEF] matches any hexadecimal digit. A circum-
3614 flex can conveniently be used with the upper case character types to
3615 specify a more restricted set of characters than the matching lower
3616 case type. For example, the class [^\W_] matches any letter or digit,
3617 but not underscore.
3618
3619 The only metacharacters that are recognized in character classes are
3620 backslash, hyphen (only where it can be interpreted as specifying a
3621 range), circumflex (only at the start), opening square bracket (only
3622 when it can be interpreted as introducing a POSIX class name - see the
3623 next section), and the terminating closing square bracket. However,
3624 escaping other non-alphanumeric characters does no harm.
3625
3626
3627 POSIX CHARACTER CLASSES
3628
3629 Perl supports the POSIX notation for character classes. This uses names
3630 enclosed by [: and :] within the enclosing square brackets. PCRE also
3631 supports this notation. For example,
3632
3633 [01[:alpha:]%]
3634
3635 matches "0", "1", any alphabetic character, or "%". The supported class
3636 names are
3637
3638 alnum letters and digits
3639 alpha letters
3640 ascii character codes 0 - 127
3641 blank space or tab only
3642 cntrl control characters
3643 digit decimal digits (same as \d)
3644 graph printing characters, excluding space
3645 lower lower case letters
3646 print printing characters, including space
3647 punct printing characters, excluding letters and digits
3648 space white space (not quite the same as \s)
3649 upper upper case letters
3650 word "word" characters (same as \w)
3651 xdigit hexadecimal digits
3652
3653 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
3654 and space (32). Notice that this list includes the VT character (code
3655 11). This makes "space" different to \s, which does not include VT (for
3656 Perl compatibility).
3657
3658 The name "word" is a Perl extension, and "blank" is a GNU extension
3659 from Perl 5.8. Another Perl extension is negation, which is indicated
3660 by a ^ character after the colon. For example,
3661
3662 [12[:^digit:]]
3663
3664 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
3665 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
3666 these are not supported, and an error is given if they are encountered.
3667
3668 In UTF-8 mode, characters with values greater than 128 do not match any
3669 of the POSIX character classes.
3670
3671
3672 VERTICAL BAR
3673
3674 Vertical bar characters are used to separate alternative patterns. For
3675 example, the pattern
3676
3677 gilbert|sullivan
3678
3679 matches either "gilbert" or "sullivan". Any number of alternatives may
3680 appear, and an empty alternative is permitted (matching the empty
3681 string). The matching process tries each alternative in turn, from left
3682 to right, and the first one that succeeds is used. If the alternatives
3683 are within a subpattern (defined below), "succeeds" means matching the
3684 rest of the main pattern as well as the alternative in the subpattern.
3685
3686
3687 INTERNAL OPTION SETTING
3688
3689 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
3690 PCRE_EXTENDED options (which are Perl-compatible) can be changed from
3691 within the pattern by a sequence of Perl option letters enclosed
3692 between "(?" and ")". The option letters are
3693
3694 i for PCRE_CASELESS
3695 m for PCRE_MULTILINE
3696 s for PCRE_DOTALL
3697 x for PCRE_EXTENDED
3698
3699 For example, (?im) sets caseless, multiline matching. It is also possi-
3700 ble to unset these options by preceding the letter with a hyphen, and a
3701 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
3702 LESS and PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED,
3703 is also permitted. If a letter appears both before and after the
3704 hyphen, the option is unset.
3705
3706 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA
3707 can be changed in the same way as the Perl-compatible options by using
3708 the characters J, U and X respectively.
3709
3710 When an option change occurs at top level (that is, not inside subpat-
3711 tern parentheses), the change applies to the remainder of the pattern
3712 that follows. If the change is placed right at the start of a pattern,
3713 PCRE extracts it into the global options (and it will therefore show up
3714 in data extracted by the pcre_fullinfo() function).
3715
3716 An option change within a subpattern (see below for a description of
3717 subpatterns) affects only that part of the current pattern that follows
3718 it, so
3719
3720 (a(?i)b)c
3721
3722 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
3723 used). By this means, options can be made to have different settings
3724 in different parts of the pattern. Any changes made in one alternative
3725 do carry on into subsequent branches within the same subpattern. For
3726 example,
3727
3728 (a(?i)b|c)
3729
3730 matches "ab", "aB", "c", and "C", even though when matching "C" the
3731 first branch is abandoned before the option setting. This is because
3732 the effects of option settings happen at compile time. There would be
3733 some very weird behaviour otherwise.
3734
3735 Note: There are other PCRE-specific options that can be set by the
3736 application when the compile or match functions are called. In some
3737 cases the pattern can contain special leading sequences to override
3738 what the application has set or what has been defaulted. Details are
3739 given in the section entitled "Newline sequences" above.
3740
3741
3742 SUBPATTERNS
3743
3744 Subpatterns are delimited by parentheses (round brackets), which can be
3745 nested. Turning part of a pattern into a subpattern does two things:
3746
3747 1. It localizes a set of alternatives. For example, the pattern
3748
3749 cat(aract|erpillar|)
3750
3751 matches one of the words "cat", "cataract", or "caterpillar". Without
3752 the parentheses, it would match "cataract", "erpillar" or an empty
3753 string.
3754
3755 2. It sets up the subpattern as a capturing subpattern. This means
3756 that, when the whole pattern matches, that portion of the subject
3757 string that matched the subpattern is passed back to the caller via the
3758 ovector argument of pcre_exec(). Opening parentheses are counted from
3759 left to right (starting from 1) to obtain numbers for the capturing
3760 subpatterns.
3761
3762 For example, if the string "the red king" is matched against the pat-
3763 tern
3764
3765 the ((red|white) (king|queen))
3766
3767 the captured substrings are "red king", "red", and "king", and are num-
3768 bered 1, 2, and 3, respectively.
3769
3770 The fact that plain parentheses fulfil two functions is not always
3771 helpful. There are often times when a grouping subpattern is required
3772 without a capturing requirement. If an opening parenthesis is followed
3773 by a question mark and a colon, the subpattern does not do any captur-
3774 ing, and is not counted when computing the number of any subsequent
3775 capturing subpatterns. For example, if the string "the white queen" is
3776 matched against the pattern
3777
3778 the ((?:red|white) (king|queen))
3779
3780 the captured substrings are "white queen" and "queen", and are numbered
3781 1 and 2. The maximum number of capturing subpatterns is 65535.
3782
3783 As a convenient shorthand, if any option settings are required at the
3784 start of a non-capturing subpattern, the option letters may appear
3785 between the "?" and the ":". Thus the two patterns
3786
3787 (?i:saturday|sunday)
3788 (?:(?i)saturday|sunday)
3789
3790 match exactly the same set of strings. Because alternative branches are
3791 tried from left to right, and options are not reset until the end of
3792 the subpattern is reached, an option setting in one branch does affect
3793 subsequent branches, so the above patterns match "SUNDAY" as well as
3794 "Saturday".
3795
3796
3797 DUPLICATE SUBPATTERN NUMBERS
3798
3799 Perl 5.10 introduced a feature whereby each alternative in a subpattern
3800 uses the same numbers for its capturing parentheses. Such a subpattern
3801 starts with (?| and is itself a non-capturing subpattern. For example,
3802 consider this pattern:
3803
3804 (?|(Sat)ur|(Sun))day
3805
3806 Because the two alternatives are inside a (?| group, both sets of cap-
3807 turing parentheses are numbered one. Thus, when the pattern matches,
3808 you can look at captured substring number one, whichever alternative
3809 matched. This construct is useful when you want to capture part, but
3810 not all, of one of a number of alternatives. Inside a (?| group, paren-
3811 theses are numbered as usual, but the number is reset at the start of
3812 each branch. The numbers of any capturing buffers that follow the sub-
3813 pattern start after the highest number used in any branch. The follow-
3814 ing example is taken from the Perl documentation. The numbers under-
3815 neath show in which buffer the captured content will be stored.
3816
3817 # before ---------------branch-reset----------- after
3818 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
3819 # 1 2 2 3 2 3 4
3820
3821 A backreference or a recursive call to a numbered subpattern always
3822 refers to the first one in the pattern with the given number.
3823
3824 An alternative approach to using this "branch reset" feature is to use
3825 duplicate named subpatterns, as described in the next section.
3826
3827
3828 NAMED SUBPATTERNS
3829
3830 Identifying capturing parentheses by number is simple, but it can be
3831 very hard to keep track of the numbers in complicated regular expres-
3832 sions. Furthermore, if an expression is modified, the numbers may
3833 change. To help with this difficulty, PCRE supports the naming of sub-
3834 patterns. This feature was not added to Perl until release 5.10. Python
3835 had the feature earlier, and PCRE introduced it at release 4.0, using
3836 the Python syntax. PCRE now supports both the Perl and the Python syn-
3837 tax.
3838
3839 In PCRE, a subpattern can be named in one of three ways: (?<name>...)
3840 or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
3841 to capturing parentheses from other parts of the pattern, such as back-
3842 references, recursion, and conditions, can be made by name as well as
3843 by number.
3844
3845 Names consist of up to 32 alphanumeric characters and underscores.
3846 Named capturing parentheses are still allocated numbers as well as
3847 names, exactly as if the names were not present. The PCRE API provides
3848 function calls for extracting the name-to-number translation table from
3849 a compiled pattern. There is also a convenience function for extracting
3850 a captured substring by name.
3851
3852 By default, a name must be unique within a pattern, but it is possible
3853 to relax this constraint by setting the PCRE_DUPNAMES option at compile
3854 time. This can be useful for patterns where only one instance of the
3855 named parentheses can match. Suppose you want to match the name of a
3856 weekday, either as a 3-letter abbreviation or as the full name, and in
3857 both cases you want to extract the abbreviation. This pattern (ignoring
3858 the line breaks) does the job:
3859
3860 (?<DN>Mon|Fri|Sun)(?:day)?|
3861 (?<DN>Tue)(?:sday)?|
3862 (?<DN>Wed)(?:nesday)?|
3863 (?<DN>Thu)(?:rsday)?|
3864 (?<DN>Sat)(?:urday)?
3865
3866 There are five capturing substrings, but only one is ever set after a
3867 match. (An alternative way of solving this problem is to use a "branch
3868 reset" subpattern, as described in the previous section.)
3869
3870 The convenience function for extracting the data by name returns the
3871 substring for the first (and in this example, the only) subpattern of
3872 that name that matched. This saves searching to find which numbered
3873 subpattern it was. If you make a reference to a non-unique named sub-
3874 pattern from elsewhere in the pattern, the one that corresponds to the
3875 lowest number is used. For further details of the interfaces for han-
3876 dling named subpatterns, see the pcreapi documentation.
3877
3878
3879 REPETITION
3880
3881 Repetition is specified by quantifiers, which can follow any of the
3882 following items:
3883
3884 a literal data character
3885 the dot metacharacter
3886 the \C escape sequence
3887 the \X escape sequence (in UTF-8 mode with Unicode properties)
3888 the \R escape sequence
3889 an escape such as \d that matches a single character
3890 a character class
3891 a back reference (see next section)
3892 a parenthesized subpattern (unless it is an assertion)
3893
3894 The general repetition quantifier specifies a minimum and maximum num-
3895 ber of permitted matches, by giving the two numbers in curly brackets
3896 (braces), separated by a comma. The numbers must be less than 65536,
3897 and the first must be less than or equal to the second. For example:
3898
3899 z{2,4}
3900
3901 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
3902 special character. If the second number is omitted, but the comma is
3903 present, there is no upper limit; if the second number and the comma
3904 are both omitted, the quantifier specifies an exact number of required
3905 matches. Thus
3906
3907 [aeiou]{3,}
3908
3909 matches at least 3 successive vowels, but may match many more, while
3910
3911 \d{8}
3912
3913 matches exactly 8 digits. An opening curly bracket that appears in a
3914 position where a quantifier is not allowed, or one that does not match
3915 the syntax of a quantifier, is taken as a literal character. For exam-
3916 ple, {,6} is not a quantifier, but a literal string of four characters.
3917
3918 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to
3919 individual bytes. Thus, for example, \x{100}{2} matches two UTF-8 char-
3920 acters, each of which is represented by a two-byte sequence. Similarly,
3921 when Unicode property support is available, \X{3} matches three Unicode
3922 extended sequences, each of which may be several bytes long (and they
3923 may be of different lengths).
3924
3925 The quantifier {0} is permitted, causing the expression to behave as if
3926 the previous item and the quantifier were not present.
3927
3928 For convenience, the three most common quantifiers have single-charac-
3929 ter abbreviations:
3930
3931 * is equivalent to {0,}
3932 + is equivalent to {1,}
3933 ? is equivalent to {0,1}
3934
3935 It is possible to construct infinite loops by following a subpattern
3936 that can match no characters with a quantifier that has no upper limit,
3937 for example:
3938
3939 (a?)*
3940
3941 Earlier versions of Perl and PCRE used to give an error at compile time
3942 for such patterns. However, because there are cases where this can be
3943 useful, such patterns are now accepted, but if any repetition of the
3944 subpattern does in fact match no characters, the loop is forcibly bro-
3945 ken.
3946
3947 By default, the quantifiers are "greedy", that is, they match as much
3948 as possible (up to the maximum number of permitted times), without
3949 causing the rest of the pattern to fail. The classic example of where
3950 this gives problems is in trying to match comments in C programs. These
3951 appear between /* and */ and within the comment, individual * and /
3952 characters may appear. An attempt to match C comments by applying the
3953 pattern
3954
3955 /\*.*\*/
3956
3957 to the string
3958
3959 /* first comment */ not comment /* second comment */
3960
3961 fails, because it matches the entire string owing to the greediness of
3962 the .* item.
3963
3964 However, if a quantifier is followed by a question mark, it ceases to
3965 be greedy, and instead matches the minimum number of times possible, so
3966 the pattern
3967
3968 /\*.*?\*/
3969
3970 does the right thing with the C comments. The meaning of the various
3971 quantifiers is not otherwise changed, just the preferred number of
3972 matches. Do not confuse this use of question mark with its use as a
3973 quantifier in its own right. Because it has two uses, it can sometimes
3974 appear doubled, as in
3975
3976 \d??\d
3977
3978 which matches one digit by preference, but can match two if that is the
3979 only way the rest of the pattern matches.
3980
3981 If the PCRE_UNGREEDY option is set (an option that is not available in
3982 Perl), the quantifiers are not greedy by default, but individual ones
3983 can be made greedy by following them with a question mark. In other
3984 words, it inverts the default behaviour.
3985
3986 When a parenthesized subpattern is quantified with a minimum repeat
3987 count that is greater than 1 or with a limited maximum, more memory is
3988 required for the compiled pattern, in proportion to the size of the
3989 minimum or maximum.
3990
3991 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
3992 alent to Perl's /s) is set, thus allowing the dot to match newlines,
3993 the pattern is implicitly anchored, because whatever follows will be
3994 tried against every character position in the subject string, so there
3995 is no point in retrying the overall match at any position after the
3996 first. PCRE normally treats such a pattern as though it were preceded
3997 by \A.
3998
3999 In cases where it is known that the subject string contains no new-
4000 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
4001 mization, or alternatively using ^ to indicate anchoring explicitly.
4002
4003 However, there is one situation where the optimization cannot be used.
4004 When .* is inside capturing parentheses that are the subject of a
4005 backreference elsewhere in the pattern, a match at the start may fail
4006 where a later one succeeds. Consider, for example:
4007
4008 (.*)abc\1
4009
4010 If the subject is "xyz123abc123" the match point is the fourth charac-
4011 ter. For this reason, such a pattern is not implicitly anchored.
4012
4013 When a capturing subpattern is repeated, the value captured is the sub-
4014 string that matched the final iteration. For example, after
4015
4016 (tweedle[dume]{3}\s*)+
4017
4018 has matched "tweedledum tweedledee" the value of the captured substring
4019 is "tweedledee". However, if there are nested capturing subpatterns,
4020 the corresponding captured values may have been set in previous itera-
4021 tions. For example, after
4022
4023 /(a|(b))+/
4024
4025 matches "aba" the value of the second captured substring is "b".
4026
4027
4028 ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS
4029
4030 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
4031 repetition, failure of what follows normally causes the repeated item
4032 to be re-evaluated to see if a different number of repeats allows the
4033 rest of the pattern to match. Sometimes it is useful to prevent this,
4034 either to change the nature of the match, or to cause it fail earlier
4035 than it otherwise might, when the author of the pattern knows there is
4036 no point in carrying on.
4037
4038 Consider, for example, the pattern \d+foo when applied to the subject
4039 line
4040
4041 123456bar
4042
4043 After matching all 6 digits and then failing to match "foo", the normal
4044 action of the matcher is to try again with only 5 digits matching the
4045 \d+ item, and then with 4, and so on, before ultimately failing.
4046 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
4047 the means for specifying that once a subpattern has matched, it is not
4048 to be re-evaluated in this way.
4049
4050 If we use atomic grouping for the previous example, the matcher gives
4051 up immediately on failing to match "foo" the first time. The notation
4052 is a kind of special parenthesis, starting with (?> as in this example:
4053
4054 (?>\d+)foo
4055
4056 This kind of parenthesis "locks up" the part of the pattern it con-
4057 tains once it has matched, and a failure further into the pattern is
4058 prevented from backtracking into it. Backtracking past it to previous
4059 items, however, works as normal.
4060
4061 An alternative description is that a subpattern of this type matches
4062 the string of characters that an identical standalone pattern would
4063 match, if anchored at the current point in the subject string.
4064
4065 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
4066 such as the above example can be thought of as a maximizing repeat that
4067 must swallow everything it can. So, while both \d+ and \d+? are pre-
4068 pared to adjust the number of digits they match in order to make the
4069 rest of the pattern match, (?>\d+) can only match an entire sequence of
4070 digits.
4071
4072 Atomic groups in general can of course contain arbitrarily complicated
4073 subpatterns, and can be nested. However, when the subpattern for an
4074 atomic group is just a single repeated item, as in the example above, a
4075 simpler notation, called a "possessive quantifier" can be used. This
4076 consists of an additional + character following a quantifier. Using
4077 this notation, the previous example can be rewritten as
4078
4079 \d++foo
4080
4081 Note that a possessive quantifier can be used with an entire group, for
4082 example:
4083
4084 (abc|xyz){2,3}+
4085
4086 Possessive quantifiers are always greedy; the setting of the
4087 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
4088 simpler forms of atomic group. However, there is no difference in the
4089 meaning of a possessive quantifier and the equivalent atomic group,
4090 though there may be a performance difference; possessive quantifiers
4091 should be slightly faster.
4092
4093 The possessive quantifier syntax is an extension to the Perl 5.8 syn-
4094 tax. Jeffrey Friedl originated the idea (and the name) in the first
4095 edition of his book. Mike McCloskey liked it, so implemented it when he
4096 built Sun's Java package, and PCRE copied it from there. It ultimately
4097 found its way into Perl at release 5.10.
4098
4099 PCRE has an optimization that automatically "possessifies" certain sim-
4100 ple pattern constructs. For example, the sequence A+B is treated as
4101 A++B because there is no point in backtracking into a sequence of A's
4102 when B must follow.
4103
4104 When a pattern contains an unlimited repeat inside a subpattern that
4105 can itself be repeated an unlimited number of times, the use of an
4106 atomic group is the only way to avoid some failing matches taking a
4107 very long time indeed. The pattern
4108
4109 (\D+|<\d+>)*[!?]
4110
4111 matches an unlimited number of substrings that either consist of non-
4112 digits, or digits enclosed in <>, followed by either ! or ?. When it
4113 matches, it runs quickly. However, if it is applied to
4114
4115 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
4116
4117 it takes a long time before reporting failure. This is because the
4118 string can be divided between the internal \D+ repeat and the external
4119 * repeat in a large number of ways, and all have to be tried. (The
4120 example uses [!?] rather than a single character at the end, because
4121 both PCRE and Perl have an optimization that allows for fast failure
4122 when a single character is used. They remember the last single charac-
4123 ter that is required for a match, and fail early if it is not present
4124 in the string.) If the pattern is changed so that it uses an atomic
4125 group, like this:
4126
4127 ((?>\D+)|<\d+>)*[!?]
4128
4129 sequences of non-digits cannot be broken, and failure happens quickly.
4130
4131
4132 BACK REFERENCES
4133
4134 Outside a character class, a backslash followed by a digit greater than
4135 0 (and possibly further digits) is a back reference to a capturing sub-
4136 pattern earlier (that is, to its left) in the pattern, provided there
4137 have been that many previous capturing left parentheses.
4138
4139 However, if the decimal number following the backslash is less than 10,
4140 it is always taken as a back reference, and causes an error only if
4141 there are not that many capturing left parentheses in the entire pat-
4142 tern. In other words, the parentheses that are referenced need not be
4143 to the left of the reference for numbers less than 10. A "forward back
4144 reference" of this type can make sense when a repetition is involved
4145 and the subpattern to the right has participated in an earlier itera-
4146 tion.
4147
4148 It is not possible to have a numerical "forward back reference" to a
4149 subpattern whose number is 10 or more using this syntax because a
4150 sequence such as \50 is interpreted as a character defined in octal.
4151 See the subsection entitled "Non-printing characters" above for further
4152 details of the handling of digits following a backslash. There is no
4153 such problem when named parentheses are used. A back reference to any
4154 subpattern is possible using named parentheses (see below).
4155
4156 Another way of avoiding the ambiguity inherent in the use of digits
4157 following a backslash is to use the \g escape sequence, which is a fea-
4158 ture introduced in Perl 5.10. This escape must be followed by an
4159 unsigned number or a negative number, optionally enclosed in braces.
4160 These examples are all identical:
4161
4162 (ring), \1
4163 (ring), \g1
4164 (ring), \g{1}
4165
4166 An unsigned number specifies an absolute reference without the ambigu-
4167 ity that is present in the older syntax. It is also useful when literal
4168 digits follow the reference. A negative number is a relative reference.
4169 Consider this example:
4170
4171 (abc(def)ghi)\g{-1}
4172
4173 The sequence \g{-1} is a reference to the most recently started captur-
4174 ing subpattern before \g, that is, is it equivalent to \2. Similarly,
4175 \g{-2} would be equivalent to \1. The use of relative references can be
4176 helpful in long patterns, and also in patterns that are created by
4177 joining together fragments that contain references within themselves.
4178
4179 A back reference matches whatever actually matched the capturing sub-
4180 pattern in the current subject string, rather than anything matching
4181 the subpattern itself (see "Subpatterns as subroutines" below for a way
4182 of doing that). So the pattern
4183
4184 (sens|respons)e and \1ibility
4185
4186 matches "sense and sensibility" and "response and responsibility", but
4187 not "sense and responsibility". If caseful matching is in force at the
4188 time of the back reference, the case of letters is relevant. For exam-
4189 ple,
4190
4191 ((?i)rah)\s+\1
4192
4193 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
4194 original capturing subpattern is matched caselessly.
4195
4196 There are several different ways of writing back references to named
4197 subpatterns. The .NET syntax \k{name} and the Perl syntax \k<name> or
4198 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's
4199 unified back reference syntax, in which \g can be used for both numeric
4200 and named references, is also supported. We could rewrite the above
4201 example in any of the following ways:
4202
4203 (?<p1>(?i)rah)\s+\k<p1>
4204 (?'p1'(?i)rah)\s+\k{p1}
4205 (?P<p1>(?i)rah)\s+(?P=p1)
4206 (?<p1>(?i)rah)\s+\g{p1}
4207
4208 A subpattern that is referenced by name may appear in the pattern
4209 before or after the reference.
4210
4211 There may be more than one back reference to the same subpattern. If a
4212 subpattern has not actually been used in a particular match, any back
4213 references to it always fail. For example, the pattern
4214
4215 (a|(bc))\2
4216
4217 always fails if it starts to match "a" rather than "bc". Because there
4218 may be many capturing parentheses in a pattern, all digits following
4219 the backslash are taken as part of a potential back reference number.
4220 If the pattern continues with a digit character, some delimiter must be
4221 used to terminate the back reference. If the PCRE_EXTENDED option is
4222 set, this can be whitespace. Otherwise an empty comment (see "Com-
4223 ments" below) can be used.
4224
4225 A back reference that occurs inside the parentheses to which it refers
4226 fails when the subpattern is first used, so, for example, (a\1) never
4227 matches. However, such references can be useful inside repeated sub-
4228 patterns. For example, the pattern
4229
4230 (a|b\1)+
4231
4232 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
4233 ation of the subpattern, the back reference matches the character
4234 string corresponding to the previous iteration. In order for this to
4235 work, the pattern must be such that the first iteration does not need
4236 to match the back reference. This can be done using alternation, as in
4237 the example above, or by a quantifier with a minimum of zero.
4238
4239
4240 ASSERTIONS
4241
4242 An assertion is a test on the characters following or preceding the
4243 current matching point that does not actually consume any characters.
4244 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
4245 described above.
4246
4247 More complicated assertions are coded as subpatterns. There are two
4248 kinds: those that look ahead of the current position in the subject
4249 string, and those that look behind it. An assertion subpattern is
4250 matched in the normal way, except that it does not cause the current
4251 matching position to be changed.
4252
4253 Assertion subpatterns are not capturing subpatterns, and may not be
4254 repeated, because it makes no sense to assert the same thing several
4255 times. If any kind of assertion contains capturing subpatterns within
4256 it, these are counted for the purposes of numbering the capturing sub-
4257 patterns in the whole pattern. However, substring capturing is carried
4258 out only for positive assertions, because it does not make sense for
4259 negative assertions.
4260
4261 Lookahead assertions
4262
4263 Lookahead assertions start with (?= for positive assertions and (?! for
4264 negative assertions. For example,
4265
4266 \w+(?=;)
4267
4268 matches a word followed by a semicolon, but does not include the semi-
4269 colon in the match, and
4270
4271 foo(?!bar)
4272
4273 matches any occurrence of "foo" that is not followed by "bar". Note
4274 that the apparently similar pattern
4275
4276 (?!foo)bar
4277
4278 does not find an occurrence of "bar" that is preceded by something
4279 other than "foo"; it finds any occurrence of "bar" whatsoever, because
4280 the assertion (?!foo) is always true when the next three characters are
4281 "bar". A lookbehind assertion is needed to achieve the other effect.
4282
4283 If you want to force a matching failure at some point in a pattern, the
4284 most convenient way to do it is with (?!) because an empty string
4285 always matches, so an assertion that requires there not to be an empty
4286 string must always fail.
4287
4288 Lookbehind assertions
4289
4290 Lookbehind assertions start with (?<= for positive assertions and (?<!
4291 for negative assertions. For example,
4292
4293 (?<!foo)bar
4294
4295 does find an occurrence of "bar" that is not preceded by "foo". The
4296 contents of a lookbehind assertion are restricted such that all the
4297 strings it matches must have a fixed length. However, if there are sev-
4298 eral top-level alternatives, they do not all have to have the same
4299 fixed length. Thus
4300
4301 (?<=bullock|donkey)
4302
4303 is permitted, but
4304
4305 (?<!dogs?|cats?)
4306
4307 causes an error at compile time. Branches that match different length
4308 strings are permitted only at the top level of a lookbehind assertion.
4309 This is an extension compared with Perl (at least for 5.8), which
4310 requires all branches to match the same length of string. An assertion
4311 such as
4312
4313 (?<=ab(c|de))
4314
4315 is not permitted, because its single top-level branch can match two
4316 different lengths, but it is acceptable if rewritten to use two top-
4317 level branches:
4318
4319 (?<=abc|abde)
4320
4321 In some cases, the Perl 5.10 escape sequence \K (see above) can be used
4322 instead of a lookbehind assertion; this is not restricted to a fixed-
4323 length.
4324
4325 The implementation of lookbehind assertions is, for each alternative,
4326 to temporarily move the current position back by the fixed length and
4327 then try to match. If there are insufficient characters before the cur-
4328 rent position, the assertion fails.
4329
4330 PCRE does not allow the \C escape (which matches a single byte in UTF-8
4331 mode) to appear in lookbehind assertions, because it makes it impossi-
4332 ble to calculate the length of the lookbehind. The \X and \R escapes,
4333 which can match different numbers of bytes, are also not permitted.
4334
4335 Possessive quantifiers can be used in conjunction with lookbehind
4336 assertions to specify efficient matching at the end of the subject
4337 string. Consider a simple pattern such as
4338
4339 abcd$
4340
4341 when applied to a long string that does not match. Because matching
4342 proceeds from left to right, PCRE will look for each "a" in the subject
4343 and then see if what follows matches the rest of the pattern. If the
4344 pattern is specified as
4345
4346 ^.*abcd$
4347
4348 the initial .* matches the entire string at first, but when this fails
4349 (because there is no following "a"), it backtracks to match all but the
4350 last character, then all but the last two characters, and so on. Once
4351 again the search for "a" covers the entire string, from right to left,
4352 so we are no better off. However, if the pattern is written as
4353
4354 ^.*+(?<=abcd)
4355
4356 there can be no backtracking for the .*+ item; it can match only the
4357 entire string. The subsequent lookbehind assertion does a single test
4358 on the last four characters. If it fails, the match fails immediately.
4359 For long strings, this approach makes a significant difference to the
4360 processing time.
4361
4362 Using multiple assertions
4363
4364 Several assertions (of any sort) may occur in succession. For example,
4365
4366 (?<=\d{3})(?<!999)foo
4367
4368 matches "foo" preceded by three digits that are not "999". Notice that
4369 each of the assertions is applied independently at the same point in
4370 the subject string. First there is a check that the previous three
4371 characters are all digits, and then there is a check that the same
4372 three characters are not "999". This pattern does not match "foo" pre-
4373 ceded by six characters, the first of which are digits and the last
4374 three of which are not "999". For example, it doesn't match "123abc-
4375 foo". A pattern to do that is
4376
4377 (?<=\d{3}...)(?<!999)foo
4378
4379 This time the first assertion looks at the preceding six characters,
4380 checking that the first three are digits, and then the second assertion
4381 checks that the preceding three characters are not "999".
4382
4383 Assertions can be nested in any combination. For example,
4384
4385 (?<=(?<!foo)bar)baz
4386
4387 matches an occurrence of "baz" that is preceded by "bar" which in turn
4388 is not preceded by "foo", while
4389
4390 (?<=\d{3}(?!999)...)foo
4391
4392 is another pattern that matches "foo" preceded by three digits and any
4393 three characters that are not "999".
4394
4395
4396 CONDITIONAL SUBPATTERNS
4397
4398 It is possible to cause the matching process to obey a subpattern con-
4399 ditionally or to choose between two alternative subpatterns, depending
4400 on the result of an assertion, or whether a previous capturing subpat-
4401 tern matched or not. The two possible forms of conditional subpattern
4402 are
4403
4404 (?(condition)yes-pattern)
4405 (?(condition)yes-pattern|no-pattern)
4406
4407 If the condition is satisfied, the yes-pattern is used; otherwise the
4408 no-pattern (if present) is used. If there are more than two alterna-
4409 tives in the subpattern, a compile-time error occurs.
4410
4411 There are four kinds of condition: references to subpatterns, refer-
4412 ences to recursion, a pseudo-condition called DEFINE, and assertions.
4413
4414 Checking for a used subpattern by number
4415
4416 If the text between the parentheses consists of a sequence of digits,
4417 the condition is true if the capturing subpattern of that number has
4418 previously matched. An alternative notation is to precede the digits
4419 with a plus or minus sign. In this case, the subpattern number is rela-
4420 tive rather than absolute. The most recently opened parentheses can be
4421 referenced by (?(-1), the next most recent by (?(-2), and so on. In
4422 looping constructs it can also make sense to refer to subsequent groups
4423 with constructs such as (?(+2).
4424
4425 Consider the following pattern, which contains non-significant white
4426 space to make it more readable (assume the PCRE_EXTENDED option) and to
4427 divide it into three parts for ease of discussion:
4428
4429 ( \( )? [^()]+ (?(1) \) )
4430
4431 The first part matches an optional opening parenthesis, and if that
4432 character is present, sets it as the first captured substring. The sec-
4433 ond part matches one or more characters that are not parentheses. The
4434 third part is a conditional subpattern that tests whether the first set
4435 of parentheses matched or not. If they did, that is, if subject started
4436 with an opening parenthesis, the condition is true, and so the yes-pat-
4437 tern is executed and a closing parenthesis is required. Otherwise,
4438 since no-pattern is not present, the subpattern matches nothing. In
4439 other words, this pattern matches a sequence of non-parentheses,
4440 optionally enclosed in parentheses.
4441
4442 If you were embedding this pattern in a larger one, you could use a
4443 relative reference:
4444
4445 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
4446
4447 This makes the fragment independent of the parentheses in the larger
4448 pattern.
4449
4450 Checking for a used subpattern by name
4451
4452 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
4453 used subpattern by name. For compatibility with earlier versions of
4454 PCRE, which had this facility before Perl, the syntax (?(name)...) is
4455 also recognized. However, there is a possible ambiguity with this syn-
4456 tax, because subpattern names may consist entirely of digits. PCRE
4457 looks first for a named subpattern; if it cannot find one and the name
4458 consists entirely of digits, PCRE looks for a subpattern of that num-
4459 ber, which must be greater than zero. Using subpattern names that con-
4460 sist entirely of digits is not recommended.
4461
4462 Rewriting the above example to use a named subpattern gives this:
4463
4464 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
4465
4466
4467 Checking for pattern recursion
4468
4469 If the condition is the string (R), and there is no subpattern with the
4470 name R, the condition is true if a recursive call to the whole pattern
4471 or any subpattern has been made. If digits or a name preceded by amper-
4472 sand follow the letter R, for example:
4473
4474 (?(R3)...) or (?(R&name)...)
4475
4476 the condition is true if the most recent recursion is into the subpat-
4477 tern whose number or name is given. This condition does not check the
4478 entire recursion stack.
4479
4480 At "top level", all these recursion test conditions are false. Recur-
4481 sive patterns are described below.
4482
4483 Defining subpatterns for use by reference only
4484
4485 If the condition is the string (DEFINE), and there is no subpattern
4486 with the name DEFINE, the condition is always false. In this case,
4487 there may be only one alternative in the subpattern. It is always
4488 skipped if control reaches this point in the pattern; the idea of
4489 DEFINE is that it can be used to define "subroutines" that can be ref-
4490 erenced from elsewhere. (The use of "subroutines" is described below.)
4491 For example, a pattern to match an IPv4 address could be written like
4492 this (ignore whitespace and line breaks):
4493
4494 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
4495 \b (?&byte) (\.(?&byte)){3} \b
4496
4497 The first part of the pattern is a DEFINE group inside which a another
4498 group named "byte" is defined. This matches an individual component of
4499 an IPv4 address (a number less than 256). When matching takes place,
4500 this part of the pattern is skipped because DEFINE acts like a false
4501 condition.
4502
4503 The rest of the pattern uses references to the named group to match the
4504 four dot-separated components of an IPv4 address, insisting on a word
4505 boundary at each end.
4506
4507 Assertion conditions
4508
4509 If the condition is not in any of the above formats, it must be an
4510 assertion. This may be a positive or negative lookahead or lookbehind
4511 assertion. Consider this pattern, again containing non-significant
4512 white space, and with the two alternatives on the second line:
4513
4514 (?(?=[^a-z]*[a-z])
4515 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
4516
4517 The condition is a positive lookahead assertion that matches an
4518 optional sequence of non-letters followed by a letter. In other words,
4519 it tests for the presence of at least one letter in the subject. If a
4520 letter is found, the subject is matched against the first alternative;
4521 otherwise it is matched against the second. This pattern matches
4522 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
4523 letters and dd are digits.
4524
4525
4526 COMMENTS
4527
4528 The sequence (?# marks the start of a comment that continues up to the
4529 next closing parenthesis. Nested parentheses are not permitted. The
4530 characters that make up a comment play no part in the pattern matching
4531 at all.
4532
4533 If the PCRE_EXTENDED option is set, an unescaped # character outside a
4534 character class introduces a comment that continues to immediately
4535 after the next newline in the pattern.
4536
4537
4538 RECURSIVE PATTERNS
4539
4540 Consider the problem of matching a string in parentheses, allowing for
4541 unlimited nested parentheses. Without the use of recursion, the best
4542 that can be done is to use a pattern that matches up to some fixed
4543 depth of nesting. It is not possible to handle an arbitrary nesting
4544 depth.
4545
4546 For some time, Perl has provided a facility that allows regular expres-
4547 sions to recurse (amongst other things). It does this by interpolating
4548 Perl code in the expression at run time, and the code can refer to the
4549 expression itself. A Perl pattern using code interpolation to solve the
4550 parentheses problem can be created like this:
4551
4552 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
4553
4554 The (?p{...}) item interpolates Perl code at run time, and in this case
4555 refers recursively to the pattern in which it appears.
4556
4557 Obviously, PCRE cannot support the interpolation of Perl code. Instead,
4558 it supports special syntax for recursion of the entire pattern, and
4559 also for individual subpattern recursion. After its introduction in
4560 PCRE and Python, this kind of recursion was introduced into Perl at
4561 release 5.10.
4562
4563 A special item that consists of (? followed by a number greater than
4564 zero and a closing parenthesis is a recursive call of the subpattern of
4565 the given number, provided that it occurs inside that subpattern. (If
4566 not, it is a "subroutine" call, which is described in the next sec-
4567 tion.) The special item (?R) or (?0) is a recursive call of the entire
4568 regular expression.
4569
4570 In PCRE (like Python, but unlike Perl), a recursive subpattern call is
4571 always treated as an atomic group. That is, once it has matched some of
4572 the subject string, it is never re-entered, even if it contains untried
4573 alternatives and there is a subsequent matching failure.
4574
4575 This PCRE pattern solves the nested parentheses problem (assume the
4576 PCRE_EXTENDED option is set so that white space is ignored):
4577
4578 \( ( (?>[^()]+) | (?R) )* \)
4579
4580 First it matches an opening parenthesis. Then it matches any number of
4581 substrings which can either be a sequence of non-parentheses, or a
4582 recursive match of the pattern itself (that is, a correctly parenthe-
4583 sized substring). Finally there is a closing parenthesis.
4584
4585 If this were part of a larger pattern, you would not want to recurse
4586 the entire pattern, so instead you could use this:
4587
4588 ( \( ( (?>[^()]+) | (?1) )* \) )
4589
4590 We have put the pattern into parentheses, and caused the recursion to
4591 refer to them instead of the whole pattern.
4592
4593 In a larger pattern, keeping track of parenthesis numbers can be
4594 tricky. This is made easier by the use of relative references. (A Perl
4595 5.10 feature.) Instead of (?1) in the pattern above you can write
4596 (?-2) to refer to the second most recently opened parentheses preceding
4597 the recursion. In other words, a negative number counts capturing
4598 parentheses leftwards from the point at which it is encountered.
4599
4600 It is also possible to refer to subsequently opened parentheses, by
4601 writing references such as (?+2). However, these cannot be recursive
4602 because the reference is not inside the parentheses that are refer-
4603 enced. They are always "subroutine" calls, as described in the next
4604 section.
4605
4606 An alternative approach is to use named parentheses instead. The Perl
4607 syntax for this is (?&name); PCRE's earlier syntax (?P>name) is also
4608 supported. We could rewrite the above example as follows:
4609
4610 (?<pn> \( ( (?>[^()]+) | (?&pn) )* \) )
4611
4612 If there is more than one subpattern with the same name, the earliest
4613 one is used.
4614
4615 This particular example pattern that we have been looking at contains
4616 nested unlimited repeats, and so the use of atomic grouping for match-
4617 ing strings of non-parentheses is important when applying the pattern
4618 to strings that do not match. For example, when this pattern is applied
4619 to
4620
4621 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
4622
4623 it yields "no match" quickly. However, if atomic grouping is not used,
4624 the match runs for a very long time indeed because there are so many
4625 different ways the + and * repeats can carve up the subject, and all
4626 have to be tested before failure can be reported.
4627
4628 At the end of a match, the values set for any capturing subpatterns are
4629 those from the outermost level of the recursion at which the subpattern
4630 value is set. If you want to obtain intermediate values, a callout
4631 function can be used (see below and the pcrecallout documentation). If
4632 the pattern above is matched against
4633
4634 (ab(cd)ef)
4635
4636 the value for the capturing parentheses is "ef", which is the last
4637 value taken on at the top level. If additional parentheses are added,
4638 giving
4639
4640 \( ( ( (?>[^()]+) | (?R) )* ) \)
4641 ^ ^
4642 ^ ^
4643
4644 the string they capture is "ab(cd)ef", the contents of the top level
4645 parentheses. If there are more than 15 capturing parentheses in a pat-
4646 tern, PCRE has to obtain extra memory to store data during a recursion,
4647 which it does by using pcre_malloc, freeing it via pcre_free after-
4648 wards. If no memory can be obtained, the match fails with the
4649 PCRE_ERROR_NOMEMORY error.
4650
4651 Do not confuse the (?R) item with the condition (R), which tests for
4652 recursion. Consider this pattern, which matches text in angle brack-
4653 ets, allowing for arbitrary nesting. Only digits are allowed in nested
4654 brackets (that is, when recursing), whereas any characters are permit-
4655 ted at the outer level.
4656
4657 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
4658
4659 In this pattern, (?(R) is the start of a conditional subpattern, with
4660 two different alternatives for the recursive and non-recursive cases.
4661 The (?R) item is the actual recursive call.
4662
4663
4664 SUBPATTERNS AS SUBROUTINES
4665
4666 If the syntax for a recursive subpattern reference (either by number or
4667 by name) is used outside the parentheses to which it refers, it oper-
4668 ates like a subroutine in a programming language. The "called" subpat-
4669 tern may be defined before or after the reference. A numbered reference
4670 can be absolute or relative, as in these examples:
4671
4672 (...(absolute)...)...(?2)...
4673 (...(relative)...)...(?-1)...
4674 (...(?+1)...(relative)...
4675
4676 An earlier example pointed out that the pattern
4677
4678 (sens|respons)e and \1ibility
4679
4680 matches "sense and sensibility" and "response and responsibility", but
4681 not "sense and responsibility". If instead the pattern
4682
4683 (sens|respons)e and (?1)ibility
4684
4685 is used, it does match "sense and responsibility" as well as the other
4686 two strings. Another example is given in the discussion of DEFINE
4687 above.
4688
4689 Like recursive subpatterns, a "subroutine" call is always treated as an
4690 atomic group. That is, once it has matched some of the subject string,
4691 it is never re-entered, even if it contains untried alternatives and
4692 there is a subsequent matching failure.
4693
4694 When a subpattern is used as a subroutine, processing options such as
4695 case-independence are fixed when the subpattern is defined. They cannot
4696 be changed for different calls. For example, consider this pattern:
4697
4698 (abc)(?i:(?-1))
4699
4700 It matches "abcabc". It does not match "abcABC" because the change of
4701 processing option does not affect the called subpattern.
4702
4703
4704 CALLOUTS
4705
4706 Perl has a feature whereby using the sequence (?{...}) causes arbitrary
4707 Perl code to be obeyed in the middle of matching a regular expression.
4708 This makes it possible, amongst other things, to extract different sub-
4709 strings that match the same pair of parentheses when there is a repeti-
4710 tion.
4711
4712 PCRE provides a similar feature, but of course it cannot obey arbitrary
4713 Perl code. The feature is called "callout". The caller of PCRE provides
4714 an external function by putting its entry point in the global variable
4715 pcre_callout. By default, this variable contains NULL, which disables
4716 all calling out.
4717
4718 Within a regular expression, (?C) indicates the points at which the
4719 external function is to be called. If you want to identify different
4720 callout points, you can put a number less than 256 after the letter C.
4721 The default value is zero. For example, this pattern has two callout
4722 points:
4723
4724 (?C1)abc(?C2)def
4725
4726 If the PCRE_AUTO_CALLOUT flag is passed to pcre_compile(), callouts are
4727 automatically installed before each item in the pattern. They are all
4728 numbered 255.
4729
4730 During matching, when PCRE reaches a callout point (and pcre_callout is
4731 set), the external function is called. It is provided with the number
4732 of the callout, the position in the pattern, and, optionally, one item
4733 of data originally supplied by the caller of pcre_exec(). The callout
4734 function may cause matching to proceed, to backtrack, or to fail alto-
4735 gether. A complete description of the interface to the callout function
4736 is given in the pcrecallout documentation.
4737
4738
4739 BACKTRACKING CONTROL
4740
4741 Perl 5.10 introduced a number of "Special Backtracking Control Verbs",
4742 which are described in the Perl documentation as "experimental and sub-
4743 ject to change or removal in a future version of Perl". It goes on to
4744 say: "Their usage in production code should be noted to avoid problems
4745 during upgrades." The same remarks apply to the PCRE features described
4746 in this section.
4747
4748 Since these verbs are specifically related to backtracking, they can be
4749 used only when the pattern is to be matched using pcre_exec(), which
4750 uses a backtracking algorithm. They cause an error if encountered by
4751 pcre_dfa_exec().
4752
4753 The new verbs make use of what was previously invalid syntax: an open-
4754 ing parenthesis followed by an asterisk. In Perl, they are generally of
4755 the form (*VERB:ARG) but PCRE does not support the use of arguments, so
4756 its general form is just (*VERB). Any number of these verbs may occur
4757 in a pattern. There are two kinds:
4758
4759 Verbs that act immediately
4760
4761 The following verbs act as soon as they are encountered:
4762
4763 (*ACCEPT)
4764
4765 This verb causes the match to end successfully, skipping the remainder
4766 of the pattern. When inside a recursion, only the innermost pattern is
4767 ended immediately. PCRE differs from Perl in what happens if the
4768 (*ACCEPT) is inside capturing parentheses. In Perl, the data so far is
4769 captured: in PCRE no data is captured. For example:
4770
4771 A(A|B(*ACCEPT)|C)D
4772
4773 This matches "AB", "AAD", or "ACD", but when it matches "AB", no data
4774 is captured.
4775
4776 (*FAIL) or (*F)
4777
4778 This verb causes the match to fail, forcing backtracking to occur. It
4779 is equivalent to (?!) but easier to read. The Perl documentation notes
4780 that it is probably useful only when combined with (?{}) or (??{}).
4781 Those are, of course, Perl features that are not present in PCRE. The
4782 nearest equivalent is the callout feature, as for example in this pat-
4783 tern:
4784
4785 a+(?C)(*FAIL)
4786
4787 A match with the string "aaaa" always fails, but the callout is taken
4788 before each backtrack happens (in this example, 10 times).
4789
4790 Verbs that act after backtracking
4791
4792 The following verbs do nothing when they are encountered. Matching con-
4793 tinues with what follows, but if there is no subsequent match, a fail-
4794 ure is forced. The verbs differ in exactly what kind of failure
4795 occurs.
4796
4797 (*COMMIT)
4798
4799 This verb causes the whole match to fail outright if the rest of the
4800 pattern does not match. Even if the pattern is unanchored, no further
4801 attempts to find a match by advancing the start point take place. Once
4802 (*COMMIT) has been passed, pcre_exec() is committed to finding a match
4803 at the current starting point, or not at all. For example:
4804
4805 a+(*COMMIT)b
4806
4807 This matches "xxaab" but not "aacaab". It can be thought of as a kind
4808 of dynamic anchor, or "I've started, so I must finish."
4809
4810 (*PRUNE)
4811
4812 This verb causes the match to fail at the current position if the rest
4813 of the pattern does not match. If the pattern is unanchored, the normal
4814 "bumpalong" advance to the next starting character then happens. Back-
4815 tracking can occur as usual to the left of (*PRUNE), or when matching
4816 to the right of (*PRUNE), but if there is no match to the right, back-
4817 tracking cannot cross (*PRUNE). In simple cases, the use of (*PRUNE)
4818 is just an alternative to an atomic group or possessive quantifier, but
4819 there are some uses of (*PRUNE) that cannot be expressed in any other
4820 way.
4821
4822 (*SKIP)
4823
4824 This verb is like (*PRUNE), except that if the pattern is unanchored,
4825 the "bumpalong" advance is not to the next character, but to the posi-
4826 tion in the subject where (*SKIP) was encountered. (*SKIP) signifies
4827 that whatever text was matched leading up to it cannot be part of a
4828 successful match. Consider:
4829
4830 a+(*SKIP)b
4831
4832 If the subject is "aaaac...", after the first match attempt fails
4833 (starting at the first character in the string), the starting point
4834 skips on to start the next attempt at "c". Note that a possessive quan-
4835 tifer does not have the same effect in this example; although it would
4836 suppress backtracking during the first match attempt, the second
4837 attempt would start at the second character instead of skipping on to
4838 "c".
4839
4840 (*THEN)
4841
4842 This verb causes a skip to the next alternation if the rest of the pat-
4843 tern does not match. That is, it cancels pending backtracking, but only
4844 within the current alternation. Its name comes from the observation
4845 that it can be used for a pattern-based if-then-else block:
4846
4847 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
4848
4849 If the COND1 pattern matches, FOO is tried (and possibly further items
4850 after the end of the group if FOO succeeds); on failure the matcher
4851 skips to the second alternative and tries COND2, without backtracking
4852 into COND1. If (*THEN) is used outside of any alternation, it acts
4853 exactly like (*PRUNE).
4854
4855
4856 SEE ALSO
4857
4858 pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
4859
4860
4861 AUTHOR
4862
4863 Philip Hazel
4864 University Computing Service
4865 Cambridge CB2 3QH, England.
4866
4867
4868 REVISION
4869
4870 Last updated: 17 September 2007
4871 Copyright (c) 1997-2007 University of Cambridge.
4872 ------------------------------------------------------------------------------
4873
4874
4875 PCRESYNTAX(3) PCRESYNTAX(3)
4876
4877
4878 NAME
4879 PCRE - Perl-compatible regular expressions
4880
4881
4882 PCRE REGULAR EXPRESSION SYNTAX SUMMARY
4883
4884 The full syntax and semantics of the regular expressions that are sup-
4885 ported by PCRE are described in the pcrepattern documentation. This
4886 document contains just a quick-reference summary of the syntax.
4887
4888
4889 QUOTING
4890
4891 \x where x is non-alphanumeric is a literal x
4892 \Q...\E treat enclosed characters as literal
4893
4894
4895 CHARACTERS
4896
4897 \a alarm, that is, the BEL character (hex 07)
4898 \cx "control-x", where x is any character
4899 \e escape (hex 1B)
4900 \f formfeed (hex 0C)
4901 \n newline (hex 0A)
4902 \r carriage return (hex 0D)
4903 \t tab (hex 09)
4904 \ddd character with octal code ddd, or backreference
4905 \xhh character with hex code hh
4906 \x{hhh..} character with hex code hhh..
4907
4908
4909 CHARACTER TYPES
4910
4911 . any character except newline;
4912 in dotall mode, any character whatsoever
4913 \C one byte, even in UTF-8 mode (best avoided)
4914 \d a decimal digit
4915 \D a character that is not a decimal digit
4916 \h a horizontal whitespace character
4917 \H a character that is not a horizontal whitespace character
4918 \p{xx} a character with the xx property
4919 \P{xx} a character without the xx property
4920 \R a newline sequence
4921 \s a whitespace character
4922 \S a character that is not a whitespace character
4923 \v a vertical whitespace character
4924 \V a character that is not a vertical whitespace character
4925 \w a "word" character
4926 \W a "non-word" character
4927 \X an extended Unicode sequence
4928
4929 In PCRE, \d, \D, \s, \S, \w, and \W recognize only ASCII characters.
4930
4931
4932 GENERAL CATEGORY PROPERTY CODES FOR \p and \P
4933
4934 C Other
4935 Cc Control
4936 Cf Format
4937 Cn Unassigned
4938 Co Private use
4939 Cs Surrogate
4940
4941 L Letter
4942 Ll Lower case letter
4943 Lm Modifier letter
4944 Lo Other letter
4945 Lt Title case letter
4946 Lu Upper case letter
4947 L& Ll, Lu, or Lt
4948
4949 M Mark
4950 Mc Spacing mark
4951 Me Enclosing mark
4952 Mn Non-spacing mark
4953
4954 N Number
4955 Nd Decimal number
4956 Nl Letter number
4957 No Other number
4958
4959 P Punctuation
4960 Pc Connector punctuation
4961 Pd Dash punctuation
4962 Pe Close punctuation
4963 Pf Final punctuation
4964 Pi Initial punctuation
4965 Po Other punctuation
4966 Ps Open punctuation
4967
4968 S Symbol
4969 Sc Currency symbol
4970 Sk Modifier symbol
4971 Sm Mathematical symbol
4972 So Other symbol
4973
4974 Z Separator
4975 Zl Line separator
4976 Zp Paragraph separator
4977 Zs Space separator
4978
4979
4980 SCRIPT NAMES FOR \p AND \P
4981
4982 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
4983 Buhid, Canadian_Aboriginal, Cherokee, Common, Coptic, Cuneiform,
4984 Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian, Glagolitic,
4985 Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
4986 gana, Inherited, Kannada, Katakana, Kharoshthi, Khmer, Lao, Latin,
4987 Limbu, Linear_B, Malayalam, Mongolian, Myanmar, New_Tai_Lue, Nko,
4988 Ogham, Old_Italic, Old_Persian, Oriya, Osmanya, Phags_Pa, Phoenician,
4989 Runic, Shavian, Sinhala, Syloti_Nagri, Syriac, Tagalog, Tagbanwa,
4990 Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Yi.
4991
4992
4993 CHARACTER CLASSES
4994
4995 [...] positive character class
4996 [^...] negative character class
4997 [x-y] range (can be used for hex characters)
4998 [[:xxx:]] positive POSIX named set
4999 [[^:xxx:]] negative POSIX named set
5000
5001 alnum alphanumeric
5002 alpha alphabetic
5003 ascii 0-127
5004 blank space or tab
5005 cntrl control character
5006 digit decimal digit
5007 graph printing, excluding space
5008 lower lower case letter
5009 print printing, including space
5010 punct printing, excluding alphanumeric
5011 space whitespace
5012 upper upper case letter
5013 word same as \w
5014 xdigit hexadecimal digit
5015
5016 In PCRE, POSIX character set names recognize only ASCII characters. You
5017 can use \Q...\E inside a character class.
5018
5019
5020 QUANTIFIERS
5021
5022 ? 0 or 1, greedy
5023 ?+ 0 or 1, possessive
5024 ?? 0 or 1, lazy
5025 * 0 or more, greedy
5026 *+ 0 or more, possessive
5027 *? 0 or more, lazy
5028 + 1 or more, greedy
5029 ++ 1 or more, possessive
5030 +? 1 or more, lazy
5031 {n} exactly n
5032 {n,m} at least n, no more than m, greedy
5033 {n,m}+ at least n, no more than m, possessive
5034 {n,m}? at least n, no more than m, lazy
5035 {n,} n or more, greedy
5036 {n,}+ n or more, possessive
5037 {n,}? n or more, lazy
5038
5039
5040 ANCHORS AND SIMPLE ASSERTIONS
5041
5042 \b word boundary
5043 \B not a word boundary
5044 ^ start of subject
5045 also after internal newline in multiline mode
5046 \A start of subject
5047 $ end of subject
5048 also before newline at end of subject
5049 also before internal newline in multiline mode
5050 \Z end of subject
5051 also before newline at end of subject
5052 \z end of subject
5053 \G first matching position in subject
5054
5055
5056 MATCH POINT RESET
5057
5058 \K reset start of match
5059
5060
5061 ALTERNATION
5062
5063 expr|expr|expr...
5064
5065
5066 CAPTURING
5067
5068 (...) capturing group
5069 (?<name>...) named capturing group (Perl)
5070 (?'name'...) named capturing group (Perl)
5071 (?P<name>...) named capturing group (Python)
5072 (?:...) non-capturing group
5073 (?|...) non-capturing group; reset group numbers for
5074 capturing groups in each alternative
5075
5076
5077 ATOMIC GROUPS
5078
5079 (?>...) atomic, non-capturing group
5080
5081
5082 COMMENT
5083
5084 (?#....) comment (not nestable)
5085
5086
5087 OPTION SETTING
5088
5089 (?i) caseless
5090 (?J) allow duplicate names
5091 (?m) multiline
5092 (?s) single line (dotall)
5093 (?U) default ungreedy (lazy)
5094 (?x) extended (ignore white space)
5095 (?-...) unset option(s)
5096
5097
5098 LOOKAHEAD AND LOOKBEHIND ASSERTIONS
5099
5100 (?=...) positive look ahead
5101 (?!...) negative look ahead
5102 (?<=...) positive look behind
5103 (?<!...) negative look behind
5104
5105 Each top-level branch of a look behind must be of a fixed length.
5106
5107
5108 BACKREFERENCES
5109
5110 \n reference by number (can be ambiguous)
5111 \gn reference by number
5112 \g{n} reference by number
5113 \g{-n} relative reference by number
5114 \k<name> reference by name (Perl)
5115 \k'name' reference by name (Perl)
5116 \g{name} reference by name (Perl)
5117 \k{name} reference by name (.NET)
5118 (?P=name) reference by name (Python)
5119
5120
5121 SUBROUTINE REFERENCES (POSSIBLY RECURSIVE)
5122
5123 (?R) recurse whole pattern
5124 (?n) call subpattern by absolute number
5125 (?+n) call subpattern by relative number
5126 (?-n) call subpattern by relative number
5127 (?&name) call subpattern by name (Perl)
5128 (?P>name) call subpattern by name (Python)
5129
5130
5131 CONDITIONAL PATTERNS
5132
5133 (?(condition)yes-pattern)
5134 (?(condition)yes-pattern|no-pattern)
5135
5136 (?(n)... absolute reference condition
5137 (?(+n)... relative reference condition
5138 (?(-n)... relative reference condition
5139 (?(<name>)... named reference condition (Perl)
5140 (?('name')... named reference condition (Perl)
5141 (?(name)... named reference condition (PCRE)
5142 (?(R)... overall recursion condition
5143 (?(Rn)... specific group recursion condition
5144 (?(R&name)... specific recursion condition
5145 (?(DEFINE)... define subpattern for reference
5146 (?(assert)... assertion condition
5147
5148
5149 BACKTRACKING CONTROL
5150
5151 The following act immediately they are reached:
5152
5153 (*ACCEPT) force successful match
5154 (*FAIL) force backtrack; synonym (*F)
5155
5156 The following act only when a subsequent match failure causes a back-
5157 track to reach them. They all force a match failure, but they differ in
5158 what happens afterwards. Those that advance the start-of-match point do
5159 so only if the pattern is not anchored.
5160
5161 (*COMMIT) overall failure, no advance of starting point
5162 (*PRUNE) advance to next starting character
5163 (*SKIP) advance start to current matching position
5164 (*THEN) local failure, backtrack to next alternation
5165
5166
5167 NEWLINE CONVENTIONS
5168
5169 These are recognized only at the very start of the pattern or after a
5170 (*BSR_...) option.
5171
5172 (*CR)
5173 (*LF)
5174 (*CRLF)
5175 (*ANYCRLF)
5176 (*ANY)
5177
5178
5179 WHAT \R MATCHES
5180
5181 These are recognized only at the very start of the pattern or after a
5182 (*...) option that sets the newline convention.
5183
5184 (*BSR_ANYCRLF)
5185 (*BSR_UNICODE)
5186
5187
5188 CALLOUTS
5189
5190 (?C) callout
5191 (?Cn) callout with data n
5192
5193
5194 SEE ALSO
5195
5196 pcrepattern(3), pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
5197
5198
5199 AUTHOR
5200
5201 Philip Hazel
5202 University Computing Service
5203 Cambridge CB2 3QH, England.
5204
5205
5206 REVISION
5207
5208 Last updated: 21 September 2007
5209 Copyright (c) 1997-2007 University of Cambridge.
5210 ------------------------------------------------------------------------------
5211
5212
5213 PCREPARTIAL(3) PCREPARTIAL(3)
5214
5215
5216 NAME
5217 PCRE - Perl-compatible regular expressions
5218
5219
5220 PARTIAL MATCHING IN PCRE
5221
5222 In normal use of PCRE, if the subject string that is passed to
5223 pcre_exec() or pcre_dfa_exec() matches as far as it goes, but is too
5224 short to match the entire pattern, PCRE_ERROR_NOMATCH is returned.
5225 There are circumstances where it might be helpful to distinguish this
5226 case from other cases in which there is no match.
5227
5228 Consider, for example, an application where a human is required to type
5229 in data for a field with specific formatting requirements. An example
5230 might be a date in the form ddmmmyy, defined by this pattern:
5231
5232 ^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$
5233
5234 If the application sees the user's keystrokes one by one, and can check
5235 that what has been typed so far is potentially valid, it is able to
5236 raise an error as soon as a mistake is made, possibly beeping and not
5237 reflecting the character that has been typed. This immediate feedback
5238 is likely to be a better user interface than a check that is delayed
5239 until the entire string has been entered.
5240
5241 PCRE supports the concept of partial matching by means of the PCRE_PAR-
5242 TIAL option, which can be set when calling pcre_exec() or
5243 pcre_dfa_exec(). When this flag is set for pcre_exec(), the return code
5244 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if at any time
5245 during the matching process the last part of the subject string matched
5246 part of the pattern. Unfortunately, for non-anchored matching, it is
5247 not possible to obtain the position of the start of the partial match.
5248 No captured data is set when PCRE_ERROR_PARTIAL is returned.
5249
5250 When PCRE_PARTIAL is set for pcre_dfa_exec(), the return code
5251 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end of
5252 the subject is reached, there have been no complete matches, but there
5253 is still at least one matching possibility. The portion of the string
5254 that provided the partial match is set as the first matching string.
5255
5256 Using PCRE_PARTIAL disables one of PCRE's optimizations. PCRE remembers
5257 the last literal byte in a pattern, and abandons matching immediately
5258 if such a byte is not present in the subject string. This optimization
5259 cannot be used for a subject string that might match only partially.
5260
5261
5262 RESTRICTED PATTERNS FOR PCRE_PARTIAL
5263
5264 Because of the way certain internal optimizations are implemented in
5265 the pcre_exec() function, the PCRE_PARTIAL option cannot be used with
5266 all patterns. These restrictions do not apply when pcre_dfa_exec() is
5267 used. For pcre_exec(), repeated single characters such as
5268
5269 a{2,4}
5270
5271 and repeated single metasequences such as
5272
5273 \d+
5274
5275 are not permitted if the maximum number of occurrences is greater than
5276 one. Optional items such as \d? (where the maximum is one) are permit-
5277 ted. Quantifiers with any values are permitted after parentheses, so
5278 the invalid examples above can be coded thus:
5279
5280 (a){2,4}
5281 (\d)+
5282
5283 These constructions run more slowly, but for the kinds of application
5284 that are envisaged for this facility, this is not felt to be a major
5285 restriction.
5286
5287 If PCRE_PARTIAL is set for a pattern that does not conform to the
5288 restrictions, pcre_exec() returns the error code PCRE_ERROR_BADPARTIAL
5289 (-13). You can use the PCRE_INFO_OKPARTIAL call to pcre_fullinfo() to
5290 find out if a compiled pattern can be used for partial matching.
5291
5292
5293 EXAMPLE OF PARTIAL MATCHING USING PCRETEST
5294
5295 If the escape sequence \P is present in a pcretest data line, the
5296 PCRE_PARTIAL flag is used for the match. Here is a run of pcretest that
5297 uses the date example quoted above:
5298
5299 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5300 data> 25jun04\P
5301 0: 25jun04
5302 1: jun
5303 data> 25dec3\P
5304 Partial match
5305 data> 3ju\P
5306 Partial match
5307 data> 3juj\P
5308 No match
5309 data> j\P
5310 No match
5311
5312 The first data string is matched completely, so pcretest shows the
5313 matched substrings. The remaining four strings do not match the com-
5314 plete pattern, but the first two are partial matches. The same test,
5315 using pcre_dfa_exec() matching (by means of the \D escape sequence),
5316 produces the following output:
5317
5318 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5319 data> 25jun04\P\D
5320 0: 25jun04
5321 data> 23dec3\P\D
5322 Partial match: 23dec3
5323 data> 3ju\P\D
5324 Partial match: 3ju
5325 data> 3juj\P\D
5326 No match
5327 data> j\P\D
5328 No match
5329
5330 Notice that in this case the portion of the string that was matched is
5331 made available.
5332
5333
5334 MULTI-SEGMENT MATCHING WITH pcre_dfa_exec()
5335
5336 When a partial match has been found using pcre_dfa_exec(), it is possi-
5337 ble to continue the match by providing additional subject data and
5338 calling pcre_dfa_exec() again with the same compiled regular expres-
5339 sion, this time setting the PCRE_DFA_RESTART option. You must also pass
5340 the same working space as before, because this is where details of the
5341 previous partial match are stored. Here is an example using pcretest,
5342 using the \R escape sequence to set the PCRE_DFA_RESTART option (\P and
5343 \D are as above):
5344
5345 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5346 data> 23ja\P\D
5347 Partial match: 23ja
5348 data> n05\R\D
5349 0: n05
5350
5351 The first call has "23ja" as the subject, and requests partial match-
5352 ing; the second call has "n05" as the subject for the continued
5353 (restarted) match. Notice that when the match is complete, only the
5354 last part is shown; PCRE does not retain the previously partially-
5355 matched string. It is up to the calling program to do that if it needs
5356 to.
5357
5358 You can set PCRE_PARTIAL with PCRE_DFA_RESTART to continue partial
5359 matching over multiple segments. This facility can be used to pass very
5360 long subject strings to pcre_dfa_exec(). However, some care is needed
5361 for certain types of pattern.
5362
5363 1. If the pattern contains tests for the beginning or end of a line,
5364 you need to pass the PCRE_NOTBOL or PCRE_NOTEOL options, as appropri-
5365 ate, when the subject string for any call does not contain the begin-
5366 ning or end of a line.
5367
5368 2. If the pattern contains backward assertions (including \b or \B),
5369 you need to arrange for some overlap in the subject strings to allow
5370 for this. For example, you could pass the subject in chunks that are
5371 500 bytes long, but in a buffer of 700 bytes, with the starting offset
5372 set to 200 and the previous 200 bytes at the start of the buffer.
5373
5374 3. Matching a subject string that is split into multiple segments does
5375 not always produce exactly the same result as matching over one single
5376 long string. The difference arises when there are multiple matching
5377 possibilities, because a partial match result is given only when there
5378 are no completed matches in a call to pcre_dfa_exec(). This means that
5379 as soon as the shortest match has been found, continuation to a new
5380 subject segment is no longer possible. Consider this pcretest example:
5381
5382 re> /dog(sbody)?/
5383 data> do\P\D
5384 Partial match: do
5385 data> gsb\R\P\D
5386 0: g
5387 data> dogsbody\D
5388 0: dogsbody
5389 1: dog
5390
5391 The pattern matches the words "dog" or "dogsbody". When the subject is
5392 presented in several parts ("do" and "gsb" being the first two) the
5393 match stops when "dog" has been found, and it is not possible to con-
5394 tinue. On the other hand, if "dogsbody" is presented as a single
5395 string, both matches are found.
5396
5397 Because of this phenomenon, it does not usually make sense to end a
5398 pattern that is going to be matched in this way with a variable repeat.
5399
5400 4. Patterns that contain alternatives at the top level which do not all
5401 start with the same pattern item may not work as expected. For example,
5402 consider this pattern:
5403
5404 1234|3789
5405
5406 If the first part of the subject is "ABC123", a partial match of the
5407 first alternative is found at offset 3. There is no partial match for
5408 the second alternative, because such a match does not start at the same
5409 point in the subject string. Attempting to continue with the string
5410 "789" does not yield a match because only those alternatives that match
5411 at one point in the subject are remembered. The problem arises because
5412 the start of the second alternative matches within the first alterna-
5413 tive. There is no problem with anchored patterns or patterns such as:
5414
5415 1234|ABCD
5416
5417 where no string can be a partial match for both alternatives.
5418
5419
5420 AUTHOR
5421
5422 Philip Hazel
5423 University Computing Service
5424 Cambridge CB2 3QH, England.
5425
5426
5427 REVISION
5428
5429 Last updated: 04 June 2007
5430 Copyright (c) 1997-2007 University of Cambridge.
5431 ------------------------------------------------------------------------------
5432
5433
5434 PCREPRECOMPILE(3) PCREPRECOMPILE(3)
5435
5436
5437 NAME
5438 PCRE - Perl-compatible regular expressions
5439
5440
5441 SAVING AND RE-USING PRECOMPILED PCRE PATTERNS
5442
5443 If you are running an application that uses a large number of regular
5444 expression patterns, it may be useful to store them in a precompiled
5445 form instead of having to compile them every time the application is
5446 run. If you are not using any private character tables (see the
5447 pcre_maketables() documentation), this is relatively straightforward.
5448 If you are using private tables, it is a little bit more complicated.
5449
5450 If you save compiled patterns to a file, you can copy them to a differ-
5451 ent host and run them there. This works even if the new host has the
5452 opposite endianness to the one on which the patterns were compiled.
5453 There may be a small performance penalty, but it should be insignifi-
5454 cant. However, compiling regular expressions with one version of PCRE
5455 for use with a different version is not guaranteed to work and may
5456 cause crashes.
5457
5458
5459 SAVING A COMPILED PATTERN
5460 The value returned by pcre_compile() points to a single block of memory
5461 that holds the compiled pattern and associated data. You can find the
5462 length of this block in bytes by calling pcre_fullinfo() with an argu-
5463 ment of PCRE_INFO_SIZE. You can then save the data in any appropriate
5464 manner. Here is sample code that compiles a pattern and writes it to a
5465 file. It assumes that the variable fd refers to a file that is open for
5466 output:
5467
5468 int erroroffset, rc, size;
5469 char *error;
5470 pcre *re;
5471
5472 re = pcre_compile("my pattern", 0, &error, &erroroffset, NULL);
5473 if (re == NULL) { ... handle errors ... }
5474 rc = pcre_fullinfo(re, NULL, PCRE_INFO_SIZE, &size);
5475 if (rc < 0) { ... handle errors ... }
5476 rc = fwrite(re, 1, size, fd);
5477 if (rc != size) { ... handle errors ... }
5478
5479 In this example, the bytes that comprise the compiled pattern are
5480 copied exactly. Note that this is binary data that may contain any of
5481 the 256 possible byte values. On systems that make a distinction
5482 between binary and non-binary data, be sure that the file is opened for
5483 binary output.
5484
5485 If you want to write more than one pattern to a file, you will have to
5486 devise a way of separating them. For binary data, preceding each pat-
5487 tern with its length is probably the most straightforward approach.
5488 Another possibility is to write out the data in hexadecimal instead of
5489 binary, one pattern to a line.
5490
5491 Saving compiled patterns in a file is only one possible way of storing
5492 them for later use. They could equally well be saved in a database, or
5493 in the memory of some daemon process that passes them via sockets to
5494 the processes that want them.
5495
5496 If the pattern has been studied, it is also possible to save the study
5497 data in a similar way to the compiled pattern itself. When studying
5498 generates additional information, pcre_study() returns a pointer to a
5499 pcre_extra data block. Its format is defined in the section on matching
5500 a pattern in the pcreapi documentation. The study_data field points to
5501 the binary study data, and this is what you must save (not the
5502 pcre_extra block itself). The length of the study data can be obtained
5503 by calling pcre_fullinfo() with an argument of PCRE_INFO_STUDYSIZE.
5504 Remember to check that pcre_study() did return a non-NULL value before
5505 trying to save the study data.
5506
5507
5508 RE-USING A PRECOMPILED PATTERN
5509
5510 Re-using a precompiled pattern is straightforward. Having reloaded it
5511 into main memory, you pass its pointer to pcre_exec() or
5512 pcre_dfa_exec() in the usual way. This should work even on another
5513 host, and even if that host has the opposite endianness to the one
5514 where the pattern was compiled.
5515
5516 However, if you passed a pointer to custom character tables when the
5517 pattern was compiled (the tableptr argument of pcre_compile()), you
5518 must now pass a similar pointer to pcre_exec() or pcre_dfa_exec(),
5519 because the value saved with the compiled pattern will obviously be
5520 nonsense. A field in a pcre_extra() block is used to pass this data, as
5521 described in the section on matching a pattern in the pcreapi documen-
5522 tation.
5523
5524 If you did not provide custom character tables when the pattern was
5525 compiled, the pointer in the compiled pattern is NULL, which causes
5526 pcre_exec() to use PCRE's internal tables. Thus, you do not need to
5527 take any special action at run time in this case.
5528
5529 If you saved study data with the compiled pattern, you need to create
5530 your own pcre_extra data block and set the study_data field to point to
5531 the reloaded study data. You must also set the PCRE_EXTRA_STUDY_DATA
5532 bit in the flags field to indicate that study data is present. Then
5533 pass the pcre_extra block to pcre_exec() or pcre_dfa_exec() in the
5534 usual way.
5535
5536
5537 COMPATIBILITY WITH DIFFERENT PCRE RELEASES
5538
5539 In general, it is safest to recompile all saved patterns when you
5540 update to a new PCRE release, though not all updates actually require
5541 this. Recompiling is definitely needed for release 7.2.
5542
5543
5544 AUTHOR
5545
5546 Philip Hazel
5547 University Computing Service
5548 Cambridge CB2 3QH, England.
5549
5550
5551 REVISION
5552
5553 Last updated: 13 June 2007
5554 Copyright (c) 1997-2007 University of Cambridge.
5555 ------------------------------------------------------------------------------
5556
5557
5558 PCREPERFORM(3) PCREPERFORM(3)
5559
5560
5561 NAME
5562 PCRE - Perl-compatible regular expressions
5563
5564
5565 PCRE PERFORMANCE
5566
5567 Two aspects of performance are discussed below: memory usage and pro-
5568 cessing time. The way you express your pattern as a regular expression
5569 can affect both of them.
5570
5571
5572 MEMORY USAGE
5573
5574 Patterns are compiled by PCRE into a reasonably efficient byte code, so
5575 that most simple patterns do not use much memory. However, there is one
5576 case where memory usage can be unexpectedly large. When a parenthesized
5577 subpattern has a quantifier with a minimum greater than 1 and/or a lim-
5578 ited maximum, the whole subpattern is repeated in the compiled code.
5579 For example, the pattern
5580
5581 (abc|def){2,4}
5582
5583 is compiled as if it were
5584
5585 (abc|def)(abc|def)((abc|def)(abc|def)?)?
5586
5587 (Technical aside: It is done this way so that backtrack points within
5588 each of the repetitions can be independently maintained.)
5589
5590 For regular expressions whose quantifiers use only small numbers, this
5591 is not usually a problem. However, if the numbers are large, and par-
5592 ticularly if such repetitions are nested, the memory usage can become
5593 an embarrassment. For example, the very simple pattern
5594
5595 ((ab){1,1000}c){1,3}
5596
5597 uses 51K bytes when compiled. When PCRE is compiled with its default
5598 internal pointer size of two bytes, the size limit on a compiled pat-
5599 tern is 64K, and this is reached with the above pattern if the outer
5600 repetition is increased from 3 to 4. PCRE can be compiled to use larger
5601 internal pointers and thus handle larger compiled patterns, but it is
5602 better to try to rewrite your pattern to use less memory if you can.
5603
5604 One way of reducing the memory usage for such patterns is to make use
5605 of PCRE's "subroutine" facility. Re-writing the above pattern as
5606
5607 ((ab)(?2){0,999}c)(?1){0,2}
5608
5609 reduces the memory requirements to 18K, and indeed it remains under 20K
5610 even with the outer repetition increased to 100. However, this pattern
5611 is not exactly equivalent, because the "subroutine" calls are treated
5612 as atomic groups into which there can be no backtracking if there is a
5613 subsequent matching failure. Therefore, PCRE cannot do this kind of
5614 rewriting automatically. Furthermore, there is a noticeable loss of
5615 speed when executing the modified pattern. Nevertheless, if the atomic
5616 grouping is not a problem and the loss of speed is acceptable, this
5617 kind of rewriting will allow you to process patterns that PCRE cannot
5618 otherwise handle.
5619
5620
5621 PROCESSING TIME
5622
5623 Certain items in regular expression patterns are processed more effi-
5624 ciently than others. It is more efficient to use a character class like
5625 [aeiou] than a set of single-character alternatives such as
5626 (a|e|i|o|u). In general, the simplest construction that provides the
5627 required behaviour is usually the most efficient. Jeffrey Friedl's book
5628 contains a lot of useful general discussion about optimizing regular
5629 expressions for efficient performance. This document contains a few
5630 observations about PCRE.
5631
5632 Using Unicode character properties (the \p, \P, and \X escapes) is
5633 slow, because PCRE has to scan a structure that contains data for over
5634 fifteen thousand characters whenever it needs a character's property.
5635 If you can find an alternative pattern that does not use character
5636 properties, it will probably be faster.
5637
5638 When a pattern begins with .* not in parentheses, or in parentheses
5639 that are not the subject of a backreference, and the PCRE_DOTALL option
5640 is set, the pattern is implicitly anchored by PCRE, since it can match
5641 only at the start of a subject string. However, if PCRE_DOTALL is not
5642 set, PCRE cannot make this optimization, because the . metacharacter
5643 does not then match a newline, and if the subject string contains new-
5644 lines, the pattern may match from the character immediately following
5645 one of them instead of from the very start. For example, the pattern
5646
5647 .*second
5648
5649 matches the subject "first\nand second" (where \n stands for a newline
5650 character), with the match starting at the seventh character. In order
5651 to do this, PCRE has to retry the match starting after every newline in
5652 the subject.
5653
5654 If you are using such a pattern with subject strings that do not con-
5655 tain newlines, the best performance is obtained by setting PCRE_DOTALL,
5656 or starting the pattern with ^.* or ^.*? to indicate explicit anchor-
5657 ing. That saves PCRE from having to scan along the subject looking for
5658 a newline to restart at.
5659
5660 Beware of patterns that contain nested indefinite repeats. These can
5661 take a long time to run when applied to a string that does not match.
5662 Consider the pattern fragment
5663
5664 ^(a+)*
5665
5666 This can match "aaaa" in 16 different ways, and this number increases
5667 very rapidly as the string gets longer. (The * repeat can match 0, 1,
5668 2, 3, or 4 times, and for each of those cases other than 0 or 4, the +
5669 repeats can match different numbers of times.) When the remainder of
5670 the pattern is such that the entire match is going to fail, PCRE has in
5671 principle to try every possible variation, and this can take an
5672 extremely long time, even for relatively short strings.
5673
5674 An optimization catches some of the more simple cases such as
5675
5676 (a+)*b
5677
5678 where a literal character follows. Before embarking on the standard
5679 matching procedure, PCRE checks that there is a "b" later in the sub-
5680 ject string, and if there is not, it fails the match immediately. How-
5681 ever, when there is no following literal this optimization cannot be
5682 used. You can see the difference by comparing the behaviour of
5683
5684 (a+)*\d
5685
5686 with the pattern above. The former gives a failure almost instantly
5687 when applied to a whole line of "a" characters, whereas the latter
5688 takes an appreciable time with strings longer than about 20 characters.
5689
5690 In many cases, the solution to this kind of performance issue is to use
5691 an atomic group or a possessive quantifier.
5692
5693
5694 AUTHOR
5695
5696 Philip Hazel
5697 University Computing Service
5698 Cambridge CB2 3QH, England.
5699
5700
5701 REVISION
5702
5703 Last updated: 06 March 2007
5704 Copyright (c) 1997-2007 University of Cambridge.
5705 ------------------------------------------------------------------------------
5706
5707
5708 PCREPOSIX(3) PCREPOSIX(3)
5709
5710
5711 NAME
5712 PCRE - Perl-compatible regular expressions.
5713
5714
5715 SYNOPSIS OF POSIX API
5716
5717 #include <pcreposix.h>
5718
5719 int regcomp(regex_t *preg, const char *pattern,
5720 int cflags);
5721
5722 int regexec(regex_t *preg, const char *string,
5723 size_t nmatch, regmatch_t pmatch[], int eflags);
5724
5725 size_t regerror(int errcode, const regex_t *preg,
5726 char *errbuf, size_t errbuf_size);
5727
5728 void regfree(regex_t *preg);
5729
5730
5731 DESCRIPTION
5732
5733 This set of functions provides a POSIX-style API to the PCRE regular
5734 expression package. See the pcreapi documentation for a description of
5735 PCRE's native API, which contains much additional functionality.
5736
5737 The functions described here are just wrapper functions that ultimately
5738 call the PCRE native API. Their prototypes are defined in the
5739 pcreposix.h header file, and on Unix systems the library itself is
5740 called pcreposix.a, so can be accessed by adding -lpcreposix to the
5741 command for linking an application that uses them. Because the POSIX
5742 functions call the native ones, it is also necessary to add -lpcre.
5743
5744 I have implemented only those option bits that can be reasonably mapped
5745 to PCRE native options. In addition, the option REG_EXTENDED is defined
5746 with the value zero. This has no effect, but since programs that are
5747 written to the POSIX interface often use it, this makes it easier to
5748 slot in PCRE as a replacement library. Other POSIX options are not even
5749 defined.
5750
5751 When PCRE is called via these functions, it is only the API that is
5752 POSIX-like in style. The syntax and semantics of the regular expres-
5753 sions themselves are still those of Perl, subject to the setting of
5754 various PCRE options, as described below. "POSIX-like in style" means
5755 that the API approximates to the POSIX definition; it is not fully
5756 POSIX-compatible, and in multi-byte encoding domains it is probably
5757 even less compatible.
5758
5759 The header for these functions is supplied as pcreposix.h to avoid any
5760 potential clash with other POSIX libraries. It can, of course, be
5761 renamed or aliased as regex.h, which is the "correct" name. It provides
5762 two structure types, regex_t for compiled internal forms, and reg-
5763 match_t for returning captured substrings. It also defines some con-
5764 stants whose names start with "REG_"; these are used for setting
5765 options and identifying error codes.
5766
5767
5768 COMPILING A PATTERN
5769
5770 The function regcomp() is called to compile a pattern into an internal
5771 form. The pattern is a C string terminated by a binary zero, and is
5772 passed in the argument pattern. The preg argument is a pointer to a
5773 regex_t structure that is used as a base for storing information about
5774 the compiled regular expression.
5775
5776 The argument cflags is either zero, or contains one or more of the bits
5777 defined by the following macros:
5778
5779 REG_DOTALL
5780
5781 The PCRE_DOTALL option is set when the regular expression is passed for
5782 compilation to the native function. Note that REG_DOTALL is not part of
5783 the POSIX standard.
5784
5785 REG_ICASE
5786
5787 The PCRE_CASELESS option is set when the regular expression is passed
5788 for compilation to the native function.
5789
5790 REG_NEWLINE
5791
5792 The PCRE_MULTILINE option is set when the regular expression is passed
5793 for compilation to the native function. Note that this does not mimic
5794 the defined POSIX behaviour for REG_NEWLINE (see the following sec-
5795 tion).
5796
5797 REG_NOSUB
5798
5799 The PCRE_NO_AUTO_CAPTURE option is set when the regular expression is
5800 passed for compilation to the native function. In addition, when a pat-
5801 tern that is compiled with this flag is passed to regexec() for match-
5802 ing, the nmatch and pmatch arguments are ignored, and no captured
5803 strings are returned.
5804
5805 REG_UTF8
5806
5807 The PCRE_UTF8 option is set when the regular expression is passed for
5808 compilation to the native function. This causes the pattern itself and
5809 all data strings used for matching it to be treated as UTF-8 strings.
5810 Note that REG_UTF8 is not part of the POSIX standard.
5811
5812 In the absence of these flags, no options are passed to the native
5813 function. This means the the regex is compiled with PCRE default
5814 semantics. In particular, the way it handles newline characters in the
5815 subject string is the Perl way, not the POSIX way. Note that setting
5816 PCRE_MULTILINE has only some of the effects specified for REG_NEWLINE.
5817 It does not affect the way newlines are matched by . (they aren't) or
5818 by a negative class such as [^a] (they are).
5819
5820 The yield of regcomp() is zero on success, and non-zero otherwise. The
5821 preg structure is filled in on success, and one member of the structure
5822 is public: re_nsub contains the number of capturing subpatterns in the
5823 regular expression. Various error codes are defined in the header file.
5824
5825
5826 MATCHING NEWLINE CHARACTERS
5827
5828 This area is not simple, because POSIX and Perl take different views of
5829 things. It is not possible to get PCRE to obey POSIX semantics, but
5830 then PCRE was never intended to be a POSIX engine. The following table
5831 lists the different possibilities for matching newline characters in
5832 PCRE:
5833
5834 Default Change with
5835
5836 . matches newline no PCRE_DOTALL
5837 newline matches [^a] yes not changeable
5838 $ matches \n at end yes PCRE_DOLLARENDONLY
5839 $ matches \n in middle no PCRE_MULTILINE
5840 ^ matches \n in middle no PCRE_MULTILINE
5841
5842 This is the equivalent table for POSIX:
5843
5844 Default Change with
5845
5846 . matches newline yes REG_NEWLINE
5847 newline matches [^a] yes REG_NEWLINE
5848 $ matches \n at end no REG_NEWLINE
5849 $ matches \n in middle no REG_NEWLINE
5850 ^ matches \n in middle no REG_NEWLINE
5851
5852 PCRE's behaviour is the same as Perl's, except that there is no equiva-
5853 lent for PCRE_DOLLAR_ENDONLY in Perl. In both PCRE and Perl, there is
5854 no way to stop newline from matching [^a].
5855
5856 The default POSIX newline handling can be obtained by setting
5857 PCRE_DOTALL and PCRE_DOLLAR_ENDONLY, but there is no way to make PCRE
5858 behave exactly as for the REG_NEWLINE action.
5859
5860
5861 MATCHING A PATTERN
5862
5863 The function regexec() is called to match a compiled pattern preg
5864 against a given string, which is terminated by a zero byte, subject to
5865 the options in eflags. These can be:
5866
5867 REG_NOTBOL
5868
5869 The PCRE_NOTBOL option is set when calling the underlying PCRE matching
5870 function.
5871
5872 REG_NOTEOL
5873
5874 The PCRE_NOTEOL option is set when calling the underlying PCRE matching
5875 function.
5876
5877 If the pattern was compiled with the REG_NOSUB flag, no data about any
5878 matched strings is returned. The nmatch and pmatch arguments of
5879 regexec() are ignored.
5880
5881 Otherwise,the portion of the string that was matched, and also any cap-
5882 tured substrings, are returned via the pmatch argument, which points to
5883 an array of nmatch structures of type regmatch_t, containing the mem-
5884 bers rm_so and rm_eo. These contain the offset to the first character
5885 of each substring and the offset to the first character after the end
5886 of each substring, respectively. The 0th element of the vector relates
5887 to the entire portion of string that was matched; subsequent elements
5888 relate to the capturing subpatterns of the regular expression. Unused
5889 entries in the array have both structure members set to -1.
5890
5891 A successful match yields a zero return; various error codes are
5892 defined in the header file, of which REG_NOMATCH is the "expected"
5893 failure code.
5894
5895
5896 ERROR MESSAGES
5897
5898 The regerror() function maps a non-zero errorcode from either regcomp()
5899 or regexec() to a printable message. If preg is not NULL, the error
5900 should have arisen from the use of that structure. A message terminated
5901 by a binary zero is placed in errbuf. The length of the message,
5902 including the zero, is limited to errbuf_size. The yield of the func-
5903 tion is the size of buffer needed to hold the whole message.
5904
5905
5906 MEMORY USAGE
5907
5908 Compiling a regular expression causes memory to be allocated and asso-
5909 ciated with the preg structure. The function regfree() frees all such
5910 memory, after which preg may no longer be used as a compiled expres-
5911 sion.
5912
5913
5914 AUTHOR
5915
5916 Philip Hazel
5917 University Computing Service
5918 Cambridge CB2 3QH, England.
5919
5920
5921 REVISION
5922
5923 Last updated: 06 March 2007
5924 Copyright (c) 1997-2007 University of Cambridge.
5925 ------------------------------------------------------------------------------
5926
5927
5928 PCRECPP(3) PCRECPP(3)
5929
5930
5931 NAME
5932 PCRE - Perl-compatible regular expressions.
5933
5934
5935 SYNOPSIS OF C++ WRAPPER
5936
5937 #include <pcrecpp.h>
5938
5939
5940 DESCRIPTION
5941
5942 The C++ wrapper for PCRE was provided by Google Inc. Some additional
5943 functionality was added by Giuseppe Maxia. This brief man page was con-
5944 structed from the notes in the pcrecpp.h file, which should be con-
5945 sulted for further details.
5946
5947
5948 MATCHING INTERFACE
5949
5950 The "FullMatch" operation checks that supplied text matches a supplied
5951 pattern exactly. If pointer arguments are supplied, it copies matched
5952 sub-strings that match sub-patterns into them.
5953
5954 Example: successful match
5955 pcrecpp::RE re("h.*o");
5956 re.FullMatch("hello");
5957
5958 Example: unsuccessful match (requires full match):
5959 pcrecpp::RE re("e");
5960 !re.FullMatch("hello");
5961
5962 Example: creating a temporary RE object:
5963 pcrecpp::RE("h.*o").FullMatch("hello");
5964
5965 You can pass in a "const char*" or a "string" for "text". The examples
5966 below tend to use a const char*. You can, as in the different examples
5967 above, store the RE object explicitly in a variable or use a temporary
5968 RE object. The examples below use one mode or the other arbitrarily.
5969 Either could correctly be used for any of these examples.
5970
5971 You must supply extra pointer arguments to extract matched subpieces.
5972
5973 Example: extracts "ruby" into "s" and 1234 into "i"
5974 int i;
5975 string s;
5976 pcrecpp::RE re("(\\w+):(\\d+)");
5977 re.FullMatch("ruby:1234", &s, &i);
5978
5979 Example: does not try to extract any extra sub-patterns
5980 re.FullMatch("ruby:1234", &s);
5981
5982 Example: does not try to extract into NULL
5983 re.FullMatch("ruby:1234", NULL, &i);
5984
5985 Example: integer overflow causes failure
5986 !re.FullMatch("ruby:1234567891234", NULL, &i);
5987
5988 Example: fails because there aren't enough sub-patterns:
5989 !pcrecpp::RE("\\w+:\\d+").FullMatch("ruby:1234", &s);
5990
5991 Example: fails because string cannot be stored in integer
5992 !pcrecpp::RE("(.*)").FullMatch("ruby", &i);
5993
5994 The provided pointer arguments can be pointers to any scalar numeric
5995 type, or one of:
5996
5997 string (matched piece is copied to string)
5998 StringPiece (StringPiece is mutated to point to matched piece)
5999 T (where "bool T::ParseFrom(const char*, int)" exists)
6000 NULL (the corresponding matched sub-pattern is not copied)
6001
6002 The function returns true iff all of the following conditions are sat-
6003 isfied:
6004
6005 a. "text" matches "pattern" exactly;
6006
6007 b. The number of matched sub-patterns is >= number of supplied
6008 pointers;
6009
6010 c. The "i"th argument has a suitable type for holding the
6011 string captured as the "i"th sub-pattern. If you pass in
6012 NULL for the "i"th argument, or pass fewer arguments than
6013 number of sub-patterns, "i"th captured sub-pattern is
6014 ignored.
6015
6016 CAVEAT: An optional sub-pattern that does not exist in the matched
6017 string is assigned the empty string. Therefore, the following will
6018 return false (because the empty string is not a valid number):
6019
6020 int number;
6021 pcrecpp::RE::FullMatch("abc", "[a-z]+(\\d+)?", &number);
6022
6023 The matching interface supports at most 16 arguments per call. If you
6024 need more, consider using the more general interface
6025 pcrecpp::RE::DoMatch. See pcrecpp.h for the signature for DoMatch.
6026
6027
6028 QUOTING METACHARACTERS
6029
6030 You can use the "QuoteMeta" operation to insert backslashes before all
6031 potentially meaningful characters in a string. The returned string,
6032 used as a regular expression, will exactly match the original string.
6033
6034 Example:
6035 string quoted = RE::QuoteMeta(unquoted);
6036
6037 Note that it's legal to escape a character even if it has no special
6038 meaning in a regular expression -- so this function does that. (This
6039 also makes it identical to the perl function of the same name; see
6040 "perldoc -f quotemeta".) For example, "1.5-2.0?" becomes
6041 "1\.5\-2\.0\?".
6042
6043
6044 PARTIAL MATCHES
6045
6046 You can use the "PartialMatch" operation when you want the pattern to
6047 match any substring of the text.
6048
6049 Example: simple search for a string:
6050 pcrecpp::RE("ell").PartialMatch("hello");
6051
6052 Example: find first number in a string:
6053 int number;
6054 pcrecpp::RE re("(\\d+)");
6055 re.PartialMatch("x*100 + 20", &number);
6056 assert(number == 100);
6057
6058
6059 UTF-8 AND THE MATCHING INTERFACE
6060
6061 By default, pattern and text are plain text, one byte per character.
6062 The UTF8 flag, passed to the constructor, causes both pattern and
6063 string to be treated as UTF-8 text, still a byte stream but potentially
6064 multiple bytes per character. In practice, the text is likelier to be
6065 UTF-8 than the pattern, but the match returned may depend on the UTF8
6066 flag, so always use it when matching UTF8 text. For example, "." will
6067 match one byte normally but with UTF8 set may match up to three bytes
6068 of a multi-byte character.
6069
6070 Example:
6071 pcrecpp::RE_Options options;
6072 options.set_utf8();
6073 pcrecpp::RE re(utf8_pattern, options);
6074 re.FullMatch(utf8_string);
6075
6076 Example: using the convenience function UTF8():
6077 pcrecpp::RE re(utf8_pattern, pcrecpp::UTF8());
6078 re.FullMatch(utf8_string);
6079
6080 NOTE: The UTF8 flag is ignored if pcre was not configured with the
6081 --enable-utf8 flag.
6082
6083
6084 PASSING MODIFIERS TO THE REGULAR EXPRESSION ENGINE
6085
6086 PCRE defines some modifiers to change the behavior of the regular
6087 expression engine. The C++ wrapper defines an auxiliary class,
6088 RE_Options, as a vehicle to pass such modifiers to a RE class. Cur-
6089 rently, the following modifiers are supported:
6090
6091 modifier description Perl corresponding
6092
6093 PCRE_CASELESS case insensitive match /i
6094 PCRE_MULTILINE multiple lines match /m
6095 PCRE_DOTALL dot matches newlines /s
6096 PCRE_DOLLAR_ENDONLY $ matches only at end N/A
6097 PCRE_EXTRA strict escape parsing N/A
6098 PCRE_EXTENDED ignore whitespaces /x
6099 PCRE_UTF8 handles UTF8 chars built-in
6100 PCRE_UNGREEDY reverses * and *? N/A
6101 PCRE_NO_AUTO_CAPTURE disables capturing parens N/A (*)
6102
6103 (*) Both Perl and PCRE allow non capturing parentheses by means of the
6104 "?:" modifier within the pattern itself. e.g. (?:ab|cd) does not cap-
6105 ture, while (ab|cd) does.
6106
6107 For a full account on how each modifier works, please check the PCRE
6108 API reference page.
6109
6110 For each modifier, there are two member functions whose name is made