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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 453 - (show annotations)
Fri Sep 18 19:12:35 2009 UTC (9 years, 11 months ago) by ph10
File MIME type: text/plain
File size: 328029 byte(s)
Error occurred while calculating annotation data.
Add more explanation about recursive subpatterns, and make it possible to 
process the documenation without building a whole release.
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. Neither has
6 the pcredemo program. There are separate text files for the pcregrep and
7 pcretest commands.
8 -----------------------------------------------------------------------------
9
10
11 PCRE(3) PCRE(3)
12
13
14 NAME
15 PCRE - Perl-compatible regular expressions
16
17
18 INTRODUCTION
19
20 The PCRE library is a set of functions that implement regular expres-
21 sion pattern matching using the same syntax and semantics as Perl, with
22 just a few differences. Certain features that appeared in Python and
23 PCRE before they appeared in Perl are also available using the Python
24 syntax. There is also some support for certain .NET and Oniguruma syn-
25 tax items, and there is an option for requesting some minor changes
26 that give better JavaScript compatibility.
27
28 The current implementation of PCRE (release 8.xx) corresponds approxi-
29 mately with Perl 5.10, including support for UTF-8 encoded strings and
30 Unicode general category properties. However, UTF-8 and Unicode support
31 has to be explicitly enabled; it is not the default. The Unicode tables
32 correspond to Unicode release 5.1.
33
34 In addition to the Perl-compatible matching function, PCRE contains an
35 alternative matching function that matches the same compiled patterns
36 in a different way. In certain circumstances, the alternative function
37 has some advantages. For a discussion of the two matching algorithms,
38 see the pcrematching page.
39
40 PCRE is written in C and released as a C library. A number of people
41 have written wrappers and interfaces of various kinds. In particular,
42 Google Inc. have provided a comprehensive C++ wrapper. This is now
43 included as part of the PCRE distribution. The pcrecpp page has details
44 of this interface. Other people's contributions can be found in the
45 Contrib directory at the primary FTP site, which is:
46
47 ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre
48
49 Details of exactly which Perl regular expression features are and are
50 not supported by PCRE are given in separate documents. See the pcrepat-
51 tern and pcrecompat pages. There is a syntax summary in the pcresyntax
52 page.
53
54 Some features of PCRE can be included, excluded, or changed when the
55 library is built. The pcre_config() function makes it possible for a
56 client to discover which features are available. The features them-
57 selves are described in the pcrebuild page. Documentation about build-
58 ing PCRE for various operating systems can be found in the README file
59 in the source distribution.
60
61 The library contains a number of undocumented internal functions and
62 data tables that are used by more than one of the exported external
63 functions, but which are not intended for use by external callers.
64 Their names all begin with "_pcre_", which hopefully will not provoke
65 any name clashes. In some environments, it is possible to control which
66 external symbols are exported when a shared library is built, and in
67 these cases the undocumented symbols are not exported.
68
69
70 USER DOCUMENTATION
71
72 The user documentation for PCRE comprises a number of different sec-
73 tions. In the "man" format, each of these is a separate "man page". In
74 the HTML format, each is a separate page, linked from the index page.
75 In the plain text format, all the sections, except the pcredemo sec-
76 tion, are concatenated, for ease of searching. The sections are as fol-
77 lows:
78
79 pcre this document
80 pcre-config show PCRE installation configuration information
81 pcreapi details of PCRE's native C API
82 pcrebuild options for building PCRE
83 pcrecallout details of the callout feature
84 pcrecompat discussion of Perl compatibility
85 pcrecpp details of the C++ wrapper
86 pcredemo a demonstration C program that uses PCRE
87 pcregrep description of the pcregrep command
88 pcrematching discussion of the two matching algorithms
89 pcrepartial details of the partial matching facility
90 pcrepattern syntax and semantics of supported
91 regular expressions
92 pcresyntax quick syntax reference
93 pcreperform discussion of performance issues
94 pcreposix the POSIX-compatible C API
95 pcreprecompile details of saving and re-using precompiled patterns
96 pcresample discussion of the pcredemo program
97 pcrestack discussion of stack usage
98 pcretest description of the pcretest testing command
99
100 In addition, in the "man" and HTML formats, there is a short page for
101 each C library function, listing its arguments and results.
102
103
104 LIMITATIONS
105
106 There are some size limitations in PCRE but it is hoped that they will
107 never in practice be relevant.
108
109 The maximum length of a compiled pattern is 65539 (sic) bytes if PCRE
110 is compiled with the default internal linkage size of 2. If you want to
111 process regular expressions that are truly enormous, you can compile
112 PCRE with an internal linkage size of 3 or 4 (see the README file in
113 the source distribution and the pcrebuild documentation for details).
114 In these cases the limit is substantially larger. However, the speed
115 of execution is slower.
116
117 All values in repeating quantifiers must be less than 65536.
118
119 There is no limit to the number of parenthesized subpatterns, but there
120 can be no more than 65535 capturing subpatterns.
121
122 The maximum length of name for a named subpattern is 32 characters, and
123 the maximum number of named subpatterns is 10000.
124
125 The maximum length of a subject string is the largest positive number
126 that an integer variable can hold. However, when using the traditional
127 matching function, PCRE uses recursion to handle subpatterns and indef-
128 inite repetition. This means that the available stack space may limit
129 the size of a subject string that can be processed by certain patterns.
130 For a discussion of stack issues, see the pcrestack documentation.
131
132
133 UTF-8 AND UNICODE PROPERTY SUPPORT
134
135 From release 3.3, PCRE has had some support for character strings
136 encoded in the UTF-8 format. For release 4.0 this was greatly extended
137 to cover most common requirements, and in release 5.0 additional sup-
138 port for Unicode general category properties was added.
139
140 In order process UTF-8 strings, you must build PCRE to include UTF-8
141 support in the code, and, in addition, you must call pcre_compile()
142 with the PCRE_UTF8 option flag, or the pattern must start with the
143 sequence (*UTF8). When either of these is the case, both the pattern
144 and any subject strings that are matched against it are treated as
145 UTF-8 strings instead of just strings of bytes.
146
147 If you compile PCRE with UTF-8 support, but do not use it at run time,
148 the library will be a bit bigger, but the additional run time overhead
149 is limited to testing the PCRE_UTF8 flag occasionally, so should not be
150 very big.
151
152 If PCRE is built with Unicode character property support (which implies
153 UTF-8 support), the escape sequences \p{..}, \P{..}, and \X are sup-
154 ported. The available properties that can be tested are limited to the
155 general category properties such as Lu for an upper case letter or Nd
156 for a decimal number, the Unicode script names such as Arabic or Han,
157 and the derived properties Any and L&. A full list is given in the
158 pcrepattern documentation. Only the short names for properties are sup-
159 ported. For example, \p{L} matches a letter. Its Perl synonym, \p{Let-
160 ter}, is not supported. Furthermore, in Perl, many properties may
161 optionally be prefixed by "Is", for compatibility with Perl 5.6. PCRE
162 does not support this.
163
164 Validity of UTF-8 strings
165
166 When you set the PCRE_UTF8 flag, the strings passed as patterns and
167 subjects are (by default) checked for validity on entry to the relevant
168 functions. From release 7.3 of PCRE, the check is according the rules
169 of RFC 3629, which are themselves derived from the Unicode specifica-
170 tion. Earlier releases of PCRE followed the rules of RFC 2279, which
171 allows the full range of 31-bit values (0 to 0x7FFFFFFF). The current
172 check allows only values in the range U+0 to U+10FFFF, excluding U+D800
173 to U+DFFF.
174
175 The excluded code points are the "Low Surrogate Area" of Unicode, of
176 which the Unicode Standard says this: "The Low Surrogate Area does not
177 contain any character assignments, consequently no character code
178 charts or namelists are provided for this area. Surrogates are reserved
179 for use with UTF-16 and then must be used in pairs." The code points
180 that are encoded by UTF-16 pairs are available as independent code
181 points in the UTF-8 encoding. (In other words, the whole surrogate
182 thing is a fudge for UTF-16 which unfortunately messes up UTF-8.)
183
184 If an invalid UTF-8 string is passed to PCRE, an error return
185 (PCRE_ERROR_BADUTF8) is given. In some situations, you may already know
186 that your strings are valid, and therefore want to skip these checks in
187 order to improve performance. If you set the PCRE_NO_UTF8_CHECK flag at
188 compile time or at run time, PCRE assumes that the pattern or subject
189 it is given (respectively) contains only valid UTF-8 codes. In this
190 case, it does not diagnose an invalid UTF-8 string.
191
192 If you pass an invalid UTF-8 string when PCRE_NO_UTF8_CHECK is set,
193 what happens depends on why the string is invalid. If the string con-
194 forms to the "old" definition of UTF-8 (RFC 2279), it is processed as a
195 string of characters in the range 0 to 0x7FFFFFFF. In other words,
196 apart from the initial validity test, PCRE (when in UTF-8 mode) handles
197 strings according to the more liberal rules of RFC 2279. However, if
198 the string does not even conform to RFC 2279, the result is undefined.
199 Your program may crash.
200
201 If you want to process strings of values in the full range 0 to
202 0x7FFFFFFF, encoded in a UTF-8-like manner as per the old RFC, you can
203 set PCRE_NO_UTF8_CHECK to bypass the more restrictive test. However, in
204 this situation, you will have to apply your own validity check.
205
206 General comments about UTF-8 mode
207
208 1. An unbraced hexadecimal escape sequence (such as \xb3) matches a
209 two-byte UTF-8 character if the value is greater than 127.
210
211 2. Octal numbers up to \777 are recognized, and match two-byte UTF-8
212 characters for values greater than \177.
213
214 3. Repeat quantifiers apply to complete UTF-8 characters, not to indi-
215 vidual bytes, for example: \x{100}{3}.
216
217 4. The dot metacharacter matches one UTF-8 character instead of a sin-
218 gle byte.
219
220 5. The escape sequence \C can be used to match a single byte in UTF-8
221 mode, but its use can lead to some strange effects. This facility is
222 not available in the alternative matching function, pcre_dfa_exec().
223
224 6. The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly
225 test characters of any code value, but the characters that PCRE recog-
226 nizes as digits, spaces, or word characters remain the same set as
227 before, all with values less than 256. This remains true even when PCRE
228 includes Unicode property support, because to do otherwise would slow
229 down PCRE in many common cases. If you really want to test for a wider
230 sense of, say, "digit", you must use Unicode property tests such as
231 \p{Nd}. Note that this also applies to \b, because it is defined in
232 terms of \w and \W.
233
234 7. Similarly, characters that match the POSIX named character classes
235 are all low-valued characters.
236
237 8. However, the Perl 5.10 horizontal and vertical whitespace matching
238 escapes (\h, \H, \v, and \V) do match all the appropriate Unicode char-
239 acters.
240
241 9. Case-insensitive matching applies only to characters whose values
242 are less than 128, unless PCRE is built with Unicode property support.
243 Even when Unicode property support is available, PCRE still uses its
244 own character tables when checking the case of low-valued characters,
245 so as not to degrade performance. The Unicode property information is
246 used only for characters with higher values. Even when Unicode property
247 support is available, PCRE supports case-insensitive matching only when
248 there is a one-to-one mapping between a letter's cases. There are a
249 small number of many-to-one mappings in Unicode; these are not sup-
250 ported by PCRE.
251
252
253 AUTHOR
254
255 Philip Hazel
256 University Computing Service
257 Cambridge CB2 3QH, England.
258
259 Putting an actual email address here seems to have been a spam magnet,
260 so I've taken it away. If you want to email me, use my two initials,
261 followed by the two digits 10, at the domain cam.ac.uk.
262
263
264 REVISION
265
266 Last updated: 01 September 2009
267 Copyright (c) 1997-2009 University of Cambridge.
268 ------------------------------------------------------------------------------
269
270
271 PCREBUILD(3) PCREBUILD(3)
272
273
274 NAME
275 PCRE - Perl-compatible regular expressions
276
277
278 PCRE BUILD-TIME OPTIONS
279
280 This document describes the optional features of PCRE that can be
281 selected when the library is compiled. It assumes use of the configure
282 script, where the optional features are selected or deselected by pro-
283 viding options to configure before running the make command. However,
284 the same options can be selected in both Unix-like and non-Unix-like
285 environments using the GUI facility of cmake-gui if you are using CMake
286 instead of configure to build PCRE.
287
288 There is a lot more information about building PCRE in non-Unix-like
289 environments in the file called NON_UNIX_USE, which is part of the PCRE
290 distribution. You should consult this file as well as the README file
291 if you are building in a non-Unix-like environment.
292
293 The complete list of options for configure (which includes the standard
294 ones such as the selection of the installation directory) can be
295 obtained by running
296
297 ./configure --help
298
299 The following sections include descriptions of options whose names
300 begin with --enable or --disable. These settings specify changes to the
301 defaults for the configure command. Because of the way that configure
302 works, --enable and --disable always come in pairs, so the complemen-
303 tary option always exists as well, but as it specifies the default, it
304 is not described.
305
306
307 C++ SUPPORT
308
309 By default, the configure script will search for a C++ compiler and C++
310 header files. If it finds them, it automatically builds the C++ wrapper
311 library for PCRE. You can disable this by adding
312
313 --disable-cpp
314
315 to the configure command.
316
317
318 UTF-8 SUPPORT
319
320 To build PCRE with support for UTF-8 Unicode character strings, add
321
322 --enable-utf8
323
324 to the configure command. Of itself, this does not make PCRE treat
325 strings as UTF-8. As well as compiling PCRE with this option, you also
326 have have to set the PCRE_UTF8 option when you call the pcre_compile()
327 function.
328
329 If you set --enable-utf8 when compiling in an EBCDIC environment, PCRE
330 expects its input to be either ASCII or UTF-8 (depending on the runtime
331 option). It is not possible to support both EBCDIC and UTF-8 codes in
332 the same version of the library. Consequently, --enable-utf8 and
333 --enable-ebcdic are mutually exclusive.
334
335
336 UNICODE CHARACTER PROPERTY SUPPORT
337
338 UTF-8 support allows PCRE to process character values greater than 255
339 in the strings that it handles. On its own, however, it does not pro-
340 vide any facilities for accessing the properties of such characters. If
341 you want to be able to use the pattern escapes \P, \p, and \X, which
342 refer to Unicode character properties, you must add
343
344 --enable-unicode-properties
345
346 to the configure command. This implies UTF-8 support, even if you have
347 not explicitly requested it.
348
349 Including Unicode property support adds around 30K of tables to the
350 PCRE library. Only the general category properties such as Lu and Nd
351 are supported. Details are given in the pcrepattern documentation.
352
353
354 CODE VALUE OF NEWLINE
355
356 By default, PCRE interprets the linefeed (LF) character as indicating
357 the end of a line. This is the normal newline character on Unix-like
358 systems. You can compile PCRE to use carriage return (CR) instead, by
359 adding
360
361 --enable-newline-is-cr
362
363 to the configure command. There is also a --enable-newline-is-lf
364 option, which explicitly specifies linefeed as the newline character.
365
366 Alternatively, you can specify that line endings are to be indicated by
367 the two character sequence CRLF. If you want this, add
368
369 --enable-newline-is-crlf
370
371 to the configure command. There is a fourth option, specified by
372
373 --enable-newline-is-anycrlf
374
375 which causes PCRE to recognize any of the three sequences CR, LF, or
376 CRLF as indicating a line ending. Finally, a fifth option, specified by
377
378 --enable-newline-is-any
379
380 causes PCRE to recognize any Unicode newline sequence.
381
382 Whatever line ending convention is selected when PCRE is built can be
383 overridden when the library functions are called. At build time it is
384 conventional to use the standard for your operating system.
385
386
387 WHAT \R MATCHES
388
389 By default, the sequence \R in a pattern matches any Unicode newline
390 sequence, whatever has been selected as the line ending sequence. If
391 you specify
392
393 --enable-bsr-anycrlf
394
395 the default is changed so that \R matches only CR, LF, or CRLF. What-
396 ever is selected when PCRE is built can be overridden when the library
397 functions are called.
398
399
400 BUILDING SHARED AND STATIC LIBRARIES
401
402 The PCRE building process uses libtool to build both shared and static
403 Unix libraries by default. You can suppress one of these by adding one
404 of
405
406 --disable-shared
407 --disable-static
408
409 to the configure command, as required.
410
411
412 POSIX MALLOC USAGE
413
414 When PCRE is called through the POSIX interface (see the pcreposix doc-
415 umentation), additional working storage is required for holding the
416 pointers to capturing substrings, because PCRE requires three integers
417 per substring, whereas the POSIX interface provides only two. If the
418 number of expected substrings is small, the wrapper function uses space
419 on the stack, because this is faster than using malloc() for each call.
420 The default threshold above which the stack is no longer used is 10; it
421 can be changed by adding a setting such as
422
423 --with-posix-malloc-threshold=20
424
425 to the configure command.
426
427
428 HANDLING VERY LARGE PATTERNS
429
430 Within a compiled pattern, offset values are used to point from one
431 part to another (for example, from an opening parenthesis to an alter-
432 nation metacharacter). By default, two-byte values are used for these
433 offsets, leading to a maximum size for a compiled pattern of around
434 64K. This is sufficient to handle all but the most gigantic patterns.
435 Nevertheless, some people do want to process enormous patterns, so it
436 is possible to compile PCRE to use three-byte or four-byte offsets by
437 adding a setting such as
438
439 --with-link-size=3
440
441 to the configure command. The value given must be 2, 3, or 4. Using
442 longer offsets slows down the operation of PCRE because it has to load
443 additional bytes when handling them.
444
445
446 AVOIDING EXCESSIVE STACK USAGE
447
448 When matching with the pcre_exec() function, PCRE implements backtrack-
449 ing by making recursive calls to an internal function called match().
450 In environments where the size of the stack is limited, this can se-
451 verely limit PCRE's operation. (The Unix environment does not usually
452 suffer from this problem, but it may sometimes be necessary to increase
453 the maximum stack size. There is a discussion in the pcrestack docu-
454 mentation.) An alternative approach to recursion that uses memory from
455 the heap to remember data, instead of using recursive function calls,
456 has been implemented to work round the problem of limited stack size.
457 If you want to build a version of PCRE that works this way, add
458
459 --disable-stack-for-recursion
460
461 to the configure command. With this configuration, PCRE will use the
462 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
463 ment functions. By default these point to malloc() and free(), but you
464 can replace the pointers so that your own functions are used.
465
466 Separate functions are provided rather than using pcre_malloc and
467 pcre_free because the usage is very predictable: the block sizes
468 requested are always the same, and the blocks are always freed in
469 reverse order. A calling program might be able to implement optimized
470 functions that perform better than malloc() and free(). PCRE runs
471 noticeably more slowly when built in this way. This option affects only
472 the pcre_exec() function; it is not relevant for the the
473 pcre_dfa_exec() function.
474
475
476 LIMITING PCRE RESOURCE USAGE
477
478 Internally, PCRE has a function called match(), which it calls repeat-
479 edly (sometimes recursively) when matching a pattern with the
480 pcre_exec() function. By controlling the maximum number of times this
481 function may be called during a single matching operation, a limit can
482 be placed on the resources used by a single call to pcre_exec(). The
483 limit can be changed at run time, as described in the pcreapi documen-
484 tation. The default is 10 million, but this can be changed by adding a
485 setting such as
486
487 --with-match-limit=500000
488
489 to the configure command. This setting has no effect on the
490 pcre_dfa_exec() matching function.
491
492 In some environments it is desirable to limit the depth of recursive
493 calls of match() more strictly than the total number of calls, in order
494 to restrict the maximum amount of stack (or heap, if --disable-stack-
495 for-recursion is specified) that is used. A second limit controls this;
496 it defaults to the value that is set for --with-match-limit, which
497 imposes no additional constraints. However, you can set a lower limit
498 by adding, for example,
499
500 --with-match-limit-recursion=10000
501
502 to the configure command. This value can also be overridden at run
503 time.
504
505
506 CREATING CHARACTER TABLES AT BUILD TIME
507
508 PCRE uses fixed tables for processing characters whose code values are
509 less than 256. By default, PCRE is built with a set of tables that are
510 distributed in the file pcre_chartables.c.dist. These tables are for
511 ASCII codes only. If you add
512
513 --enable-rebuild-chartables
514
515 to the configure command, the distributed tables are no longer used.
516 Instead, a program called dftables is compiled and run. This outputs
517 the source for new set of tables, created in the default locale of your
518 C runtime system. (This method of replacing the tables does not work if
519 you are cross compiling, because dftables is run on the local host. If
520 you need to create alternative tables when cross compiling, you will
521 have to do so "by hand".)
522
523
524 USING EBCDIC CODE
525
526 PCRE assumes by default that it will run in an environment where the
527 character code is ASCII (or Unicode, which is a superset of ASCII).
528 This is the case for most computer operating systems. PCRE can, how-
529 ever, be compiled to run in an EBCDIC environment by adding
530
531 --enable-ebcdic
532
533 to the configure command. This setting implies --enable-rebuild-charta-
534 bles. You should only use it if you know that you are in an EBCDIC
535 environment (for example, an IBM mainframe operating system). The
536 --enable-ebcdic option is incompatible with --enable-utf8.
537
538
539 PCREGREP OPTIONS FOR COMPRESSED FILE SUPPORT
540
541 By default, pcregrep reads all files as plain text. You can build it so
542 that it recognizes files whose names end in .gz or .bz2, and reads them
543 with libz or libbz2, respectively, by adding one or both of
544
545 --enable-pcregrep-libz
546 --enable-pcregrep-libbz2
547
548 to the configure command. These options naturally require that the rel-
549 evant libraries are installed on your system. Configuration will fail
550 if they are not.
551
552
553 PCRETEST OPTION FOR LIBREADLINE SUPPORT
554
555 If you add
556
557 --enable-pcretest-libreadline
558
559 to the configure command, pcretest is linked with the libreadline
560 library, and when its input is from a terminal, it reads it using the
561 readline() function. This provides line-editing and history facilities.
562 Note that libreadline is GPL-licenced, so if you distribute a binary of
563 pcretest linked in this way, there may be licensing issues.
564
565 Setting this option causes the -lreadline option to be added to the
566 pcretest build. In many operating environments with a sytem-installed
567 libreadline this is sufficient. However, in some environments (e.g. if
568 an unmodified distribution version of readline is in use), some extra
569 configuration may be necessary. The INSTALL file for libreadline says
570 this:
571
572 "Readline uses the termcap functions, but does not link with the
573 termcap or curses library itself, allowing applications which link
574 with readline the to choose an appropriate library."
575
576 If your environment has not been set up so that an appropriate library
577 is automatically included, you may need to add something like
578
579 LIBS="-ncurses"
580
581 immediately before the configure command.
582
583
584 SEE ALSO
585
586 pcreapi(3), pcre_config(3).
587
588
589 AUTHOR
590
591 Philip Hazel
592 University Computing Service
593 Cambridge CB2 3QH, England.
594
595
596 REVISION
597
598 Last updated: 06 September 2009
599 Copyright (c) 1997-2009 University of Cambridge.
600 ------------------------------------------------------------------------------
601
602
603 PCREMATCHING(3) PCREMATCHING(3)
604
605
606 NAME
607 PCRE - Perl-compatible regular expressions
608
609
610 PCRE MATCHING ALGORITHMS
611
612 This document describes the two different algorithms that are available
613 in PCRE for matching a compiled regular expression against a given sub-
614 ject string. The "standard" algorithm is the one provided by the
615 pcre_exec() function. This works in the same was as Perl's matching
616 function, and provides a Perl-compatible matching operation.
617
618 An alternative algorithm is provided by the pcre_dfa_exec() function;
619 this operates in a different way, and is not Perl-compatible. It has
620 advantages and disadvantages compared with the standard algorithm, and
621 these are described below.
622
623 When there is only one possible way in which a given subject string can
624 match a pattern, the two algorithms give the same answer. A difference
625 arises, however, when there are multiple possibilities. For example, if
626 the pattern
627
628 ^<.*>
629
630 is matched against the string
631
632 <something> <something else> <something further>
633
634 there are three possible answers. The standard algorithm finds only one
635 of them, whereas the alternative algorithm finds all three.
636
637
638 REGULAR EXPRESSIONS AS TREES
639
640 The set of strings that are matched by a regular expression can be rep-
641 resented as a tree structure. An unlimited repetition in the pattern
642 makes the tree of infinite size, but it is still a tree. Matching the
643 pattern to a given subject string (from a given starting point) can be
644 thought of as a search of the tree. There are two ways to search a
645 tree: depth-first and breadth-first, and these correspond to the two
646 matching algorithms provided by PCRE.
647
648
649 THE STANDARD MATCHING ALGORITHM
650
651 In the terminology of Jeffrey Friedl's book "Mastering Regular Expres-
652 sions", the standard algorithm is an "NFA algorithm". It conducts a
653 depth-first search of the pattern tree. That is, it proceeds along a
654 single path through the tree, checking that the subject matches what is
655 required. When there is a mismatch, the algorithm tries any alterna-
656 tives at the current point, and if they all fail, it backs up to the
657 previous branch point in the tree, and tries the next alternative
658 branch at that level. This often involves backing up (moving to the
659 left) in the subject string as well. The order in which repetition
660 branches are tried is controlled by the greedy or ungreedy nature of
661 the quantifier.
662
663 If a leaf node is reached, a matching string has been found, and at
664 that point the algorithm stops. Thus, if there is more than one possi-
665 ble match, this algorithm returns the first one that it finds. Whether
666 this is the shortest, the longest, or some intermediate length depends
667 on the way the greedy and ungreedy repetition quantifiers are specified
668 in the pattern.
669
670 Because it ends up with a single path through the tree, it is rela-
671 tively straightforward for this algorithm to keep track of the sub-
672 strings that are matched by portions of the pattern in parentheses.
673 This provides support for capturing parentheses and back references.
674
675
676 THE ALTERNATIVE MATCHING ALGORITHM
677
678 This algorithm conducts a breadth-first search of the tree. Starting
679 from the first matching point in the subject, it scans the subject
680 string from left to right, once, character by character, and as it does
681 this, it remembers all the paths through the tree that represent valid
682 matches. In Friedl's terminology, this is a kind of "DFA algorithm",
683 though it is not implemented as a traditional finite state machine (it
684 keeps multiple states active simultaneously).
685
686 The scan continues until either the end of the subject is reached, or
687 there are no more unterminated paths. At this point, terminated paths
688 represent the different matching possibilities (if there are none, the
689 match has failed). Thus, if there is more than one possible match,
690 this algorithm finds all of them, and in particular, it finds the long-
691 est. In PCRE, there is an option to stop the algorithm after the first
692 match (which is necessarily the shortest) has been found.
693
694 Note that all the matches that are found start at the same point in the
695 subject. If the pattern
696
697 cat(er(pillar)?)
698
699 is matched against the string "the caterpillar catchment", the result
700 will be the three strings "cat", "cater", and "caterpillar" that start
701 at the fourth character of the subject. The algorithm does not automat-
702 ically move on to find matches that start at later positions.
703
704 Although the general principle of this matching algorithm is that it
705 scans the subject string only once, without backtracking, there is one
706 exception: when a lookbehind assertion is encountered, the preceding
707 characters have to be re-inspected.
708
709 There are a number of features of PCRE regular expressions that are not
710 supported by the alternative matching algorithm. They are as follows:
711
712 1. Because the algorithm finds all possible matches, the greedy or
713 ungreedy nature of repetition quantifiers is not relevant. Greedy and
714 ungreedy quantifiers are treated in exactly the same way. However, pos-
715 sessive quantifiers can make a difference when what follows could also
716 match what is quantified, for example in a pattern like this:
717
718 ^a++\w!
719
720 This pattern matches "aaab!" but not "aaa!", which would be matched by
721 a non-possessive quantifier. Similarly, if an atomic group is present,
722 it is matched as if it were a standalone pattern at the current point,
723 and the longest match is then "locked in" for the rest of the overall
724 pattern.
725
726 2. When dealing with multiple paths through the tree simultaneously, it
727 is not straightforward to keep track of captured substrings for the
728 different matching possibilities, and PCRE's implementation of this
729 algorithm does not attempt to do this. This means that no captured sub-
730 strings are available.
731
732 3. Because no substrings are captured, back references within the pat-
733 tern are not supported, and cause errors if encountered.
734
735 4. For the same reason, conditional expressions that use a backrefer-
736 ence as the condition or test for a specific group recursion are not
737 supported.
738
739 5. Because many paths through the tree may be active, the \K escape
740 sequence, which resets the start of the match when encountered (but may
741 be on some paths and not on others), is not supported. It causes an
742 error if encountered.
743
744 6. Callouts are supported, but the value of the capture_top field is
745 always 1, and the value of the capture_last field is always -1.
746
747 7. The \C escape sequence, which (in the standard algorithm) matches a
748 single byte, even in UTF-8 mode, is not supported because the alterna-
749 tive algorithm moves through the subject string one character at a
750 time, for all active paths through the tree.
751
752 8. Except for (*FAIL), the backtracking control verbs such as (*PRUNE)
753 are not supported. (*FAIL) is supported, and behaves like a failing
754 negative assertion.
755
756
757 ADVANTAGES OF THE ALTERNATIVE ALGORITHM
758
759 Using the alternative matching algorithm provides the following advan-
760 tages:
761
762 1. All possible matches (at a single point in the subject) are automat-
763 ically found, and in particular, the longest match is found. To find
764 more than one match using the standard algorithm, you have to do kludgy
765 things with callouts.
766
767 2. Because the alternative algorithm scans the subject string just
768 once, and never needs to backtrack, it is possible to pass very long
769 subject strings to the matching function in several pieces, checking
770 for partial matching each time.
771
772
773 DISADVANTAGES OF THE ALTERNATIVE ALGORITHM
774
775 The alternative algorithm suffers from a number of disadvantages:
776
777 1. It is substantially slower than the standard algorithm. This is
778 partly because it has to search for all possible matches, but is also
779 because it is less susceptible to optimization.
780
781 2. Capturing parentheses and back references are not supported.
782
783 3. Although atomic groups are supported, their use does not provide the
784 performance advantage that it does for the standard algorithm.
785
786
787 AUTHOR
788
789 Philip Hazel
790 University Computing Service
791 Cambridge CB2 3QH, England.
792
793
794 REVISION
795
796 Last updated: 05 September 2009
797 Copyright (c) 1997-2009 University of Cambridge.
798 ------------------------------------------------------------------------------
799
800
801 PCREAPI(3) PCREAPI(3)
802
803
804 NAME
805 PCRE - Perl-compatible regular expressions
806
807
808 PCRE NATIVE API
809
810 #include <pcre.h>
811
812 pcre *pcre_compile(const char *pattern, int options,
813 const char **errptr, int *erroffset,
814 const unsigned char *tableptr);
815
816 pcre *pcre_compile2(const char *pattern, int options,
817 int *errorcodeptr,
818 const char **errptr, int *erroffset,
819 const unsigned char *tableptr);
820
821 pcre_extra *pcre_study(const pcre *code, int options,
822 const char **errptr);
823
824 int pcre_exec(const pcre *code, const pcre_extra *extra,
825 const char *subject, int length, int startoffset,
826 int options, int *ovector, int ovecsize);
827
828 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
829 const char *subject, int length, int startoffset,
830 int options, int *ovector, int ovecsize,
831 int *workspace, int wscount);
832
833 int pcre_copy_named_substring(const pcre *code,
834 const char *subject, int *ovector,
835 int stringcount, const char *stringname,
836 char *buffer, int buffersize);
837
838 int pcre_copy_substring(const char *subject, int *ovector,
839 int stringcount, int stringnumber, char *buffer,
840 int buffersize);
841
842 int pcre_get_named_substring(const pcre *code,
843 const char *subject, int *ovector,
844 int stringcount, const char *stringname,
845 const char **stringptr);
846
847 int pcre_get_stringnumber(const pcre *code,
848 const char *name);
849
850 int pcre_get_stringtable_entries(const pcre *code,
851 const char *name, char **first, char **last);
852
853 int pcre_get_substring(const char *subject, int *ovector,
854 int stringcount, int stringnumber,
855 const char **stringptr);
856
857 int pcre_get_substring_list(const char *subject,
858 int *ovector, int stringcount, const char ***listptr);
859
860 void pcre_free_substring(const char *stringptr);
861
862 void pcre_free_substring_list(const char **stringptr);
863
864 const unsigned char *pcre_maketables(void);
865
866 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
867 int what, void *where);
868
869 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
870
871 int pcre_refcount(pcre *code, int adjust);
872
873 int pcre_config(int what, void *where);
874
875 char *pcre_version(void);
876
877 void *(*pcre_malloc)(size_t);
878
879 void (*pcre_free)(void *);
880
881 void *(*pcre_stack_malloc)(size_t);
882
883 void (*pcre_stack_free)(void *);
884
885 int (*pcre_callout)(pcre_callout_block *);
886
887
888 PCRE API OVERVIEW
889
890 PCRE has its own native API, which is described in this document. There
891 are also some wrapper functions that correspond to the POSIX regular
892 expression API. These are described in the pcreposix documentation.
893 Both of these APIs define a set of C function calls. A C++ wrapper is
894 distributed with PCRE. It is documented in the pcrecpp page.
895
896 The native API C function prototypes are defined in the header file
897 pcre.h, and on Unix systems the library itself is called libpcre. It
898 can normally be accessed by adding -lpcre to the command for linking an
899 application that uses PCRE. The header file defines the macros
900 PCRE_MAJOR and PCRE_MINOR to contain the major and minor release num-
901 bers for the library. Applications can use these to include support
902 for different releases of PCRE.
903
904 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
905 pcre_exec() are used for compiling and matching regular expressions in
906 a Perl-compatible manner. A sample program that demonstrates the sim-
907 plest way of using them is provided in the file called pcredemo.c in
908 the PCRE source distribution. A listing of this program is given in the
909 pcredemo documentation, and the pcresample documentation describes how
910 to compile and run it.
911
912 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
913 ble, is also provided. This uses a different algorithm for the match-
914 ing. The alternative algorithm finds all possible matches (at a given
915 point in the subject), and scans the subject just once (unless there
916 are lookbehind assertions). However, this algorithm does not return
917 captured substrings. A description of the two matching algorithms and
918 their advantages and disadvantages is given in the pcrematching docu-
919 mentation.
920
921 In addition to the main compiling and matching functions, there are
922 convenience functions for extracting captured substrings from a subject
923 string that is matched by pcre_exec(). They are:
924
925 pcre_copy_substring()
926 pcre_copy_named_substring()
927 pcre_get_substring()
928 pcre_get_named_substring()
929 pcre_get_substring_list()
930 pcre_get_stringnumber()
931 pcre_get_stringtable_entries()
932
933 pcre_free_substring() and pcre_free_substring_list() are also provided,
934 to free the memory used for extracted strings.
935
936 The function pcre_maketables() is used to build a set of character
937 tables in the current locale for passing to pcre_compile(),
938 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
939 provided for specialist use. Most commonly, no special tables are
940 passed, in which case internal tables that are generated when PCRE is
941 built are used.
942
943 The function pcre_fullinfo() is used to find out information about a
944 compiled pattern; pcre_info() is an obsolete version that returns only
945 some of the available information, but is retained for backwards com-
946 patibility. The function pcre_version() returns a pointer to a string
947 containing the version of PCRE and its date of release.
948
949 The function pcre_refcount() maintains a reference count in a data
950 block containing a compiled pattern. This is provided for the benefit
951 of object-oriented applications.
952
953 The global variables pcre_malloc and pcre_free initially contain the
954 entry points of the standard malloc() and free() functions, respec-
955 tively. PCRE calls the memory management functions via these variables,
956 so a calling program can replace them if it wishes to intercept the
957 calls. This should be done before calling any PCRE functions.
958
959 The global variables pcre_stack_malloc and pcre_stack_free are also
960 indirections to memory management functions. These special functions
961 are used only when PCRE is compiled to use the heap for remembering
962 data, instead of recursive function calls, when running the pcre_exec()
963 function. See the pcrebuild documentation for details of how to do
964 this. It is a non-standard way of building PCRE, for use in environ-
965 ments that have limited stacks. Because of the greater use of memory
966 management, it runs more slowly. Separate functions are provided so
967 that special-purpose external code can be used for this case. When
968 used, these functions are always called in a stack-like manner (last
969 obtained, first freed), and always for memory blocks of the same size.
970 There is a discussion about PCRE's stack usage in the pcrestack docu-
971 mentation.
972
973 The global variable pcre_callout initially contains NULL. It can be set
974 by the caller to a "callout" function, which PCRE will then call at
975 specified points during a matching operation. Details are given in the
976 pcrecallout documentation.
977
978
979 NEWLINES
980
981 PCRE supports five different conventions for indicating line breaks in
982 strings: a single CR (carriage return) character, a single LF (line-
983 feed) character, the two-character sequence CRLF, any of the three pre-
984 ceding, or any Unicode newline sequence. The Unicode newline sequences
985 are the three just mentioned, plus the single characters VT (vertical
986 tab, U+000B), FF (formfeed, U+000C), NEL (next line, U+0085), LS (line
987 separator, U+2028), and PS (paragraph separator, U+2029).
988
989 Each of the first three conventions is used by at least one operating
990 system as its standard newline sequence. When PCRE is built, a default
991 can be specified. The default default is LF, which is the Unix stan-
992 dard. When PCRE is run, the default can be overridden, either when a
993 pattern is compiled, or when it is matched.
994
995 At compile time, the newline convention can be specified by the options
996 argument of pcre_compile(), or it can be specified by special text at
997 the start of the pattern itself; this overrides any other settings. See
998 the pcrepattern page for details of the special character sequences.
999
1000 In the PCRE documentation the word "newline" is used to mean "the char-
1001 acter or pair of characters that indicate a line break". The choice of
1002 newline convention affects the handling of the dot, circumflex, and
1003 dollar metacharacters, the handling of #-comments in /x mode, and, when
1004 CRLF is a recognized line ending sequence, the match position advance-
1005 ment for a non-anchored pattern. There is more detail about this in the
1006 section on pcre_exec() options below.
1007
1008 The choice of newline convention does not affect the interpretation of
1009 the \n or \r escape sequences, nor does it affect what \R matches,
1010 which is controlled in a similar way, but by separate options.
1011
1012
1013 MULTITHREADING
1014
1015 The PCRE functions can be used in multi-threading applications, with
1016 the proviso that the memory management functions pointed to by
1017 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
1018 callout function pointed to by pcre_callout, are shared by all threads.
1019
1020 The compiled form of a regular expression is not altered during match-
1021 ing, so the same compiled pattern can safely be used by several threads
1022 at once.
1023
1024
1025 SAVING PRECOMPILED PATTERNS FOR LATER USE
1026
1027 The compiled form of a regular expression can be saved and re-used at a
1028 later time, possibly by a different program, and even on a host other
1029 than the one on which it was compiled. Details are given in the
1030 pcreprecompile documentation. However, compiling a regular expression
1031 with one version of PCRE for use with a different version is not guar-
1032 anteed to work and may cause crashes.
1033
1034
1035 CHECKING BUILD-TIME OPTIONS
1036
1037 int pcre_config(int what, void *where);
1038
1039 The function pcre_config() makes it possible for a PCRE client to dis-
1040 cover which optional features have been compiled into the PCRE library.
1041 The pcrebuild documentation has more details about these optional fea-
1042 tures.
1043
1044 The first argument for pcre_config() is an integer, specifying which
1045 information is required; the second argument is a pointer to a variable
1046 into which the information is placed. The following information is
1047 available:
1048
1049 PCRE_CONFIG_UTF8
1050
1051 The output is an integer that is set to one if UTF-8 support is avail-
1052 able; otherwise it is set to zero.
1053
1054 PCRE_CONFIG_UNICODE_PROPERTIES
1055
1056 The output is an integer that is set to one if support for Unicode
1057 character properties is available; otherwise it is set to zero.
1058
1059 PCRE_CONFIG_NEWLINE
1060
1061 The output is an integer whose value specifies the default character
1062 sequence that is recognized as meaning "newline". The four values that
1063 are supported are: 10 for LF, 13 for CR, 3338 for CRLF, -2 for ANYCRLF,
1064 and -1 for ANY. Though they are derived from ASCII, the same values
1065 are returned in EBCDIC environments. The default should normally corre-
1066 spond to the standard sequence for your operating system.
1067
1068 PCRE_CONFIG_BSR
1069
1070 The output is an integer whose value indicates what character sequences
1071 the \R escape sequence matches by default. A value of 0 means that \R
1072 matches any Unicode line ending sequence; a value of 1 means that \R
1073 matches only CR, LF, or CRLF. The default can be overridden when a pat-
1074 tern is compiled or matched.
1075
1076 PCRE_CONFIG_LINK_SIZE
1077
1078 The output is an integer that contains the number of bytes used for
1079 internal linkage in compiled regular expressions. The value is 2, 3, or
1080 4. Larger values allow larger regular expressions to be compiled, at
1081 the expense of slower matching. The default value of 2 is sufficient
1082 for all but the most massive patterns, since it allows the compiled
1083 pattern to be up to 64K in size.
1084
1085 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
1086
1087 The output is an integer that contains the threshold above which the
1088 POSIX interface uses malloc() for output vectors. Further details are
1089 given in the pcreposix documentation.
1090
1091 PCRE_CONFIG_MATCH_LIMIT
1092
1093 The output is a long integer that gives the default limit for the num-
1094 ber of internal matching function calls in a pcre_exec() execution.
1095 Further details are given with pcre_exec() below.
1096
1097 PCRE_CONFIG_MATCH_LIMIT_RECURSION
1098
1099 The output is a long integer that gives the default limit for the depth
1100 of recursion when calling the internal matching function in a
1101 pcre_exec() execution. Further details are given with pcre_exec()
1102 below.
1103
1104 PCRE_CONFIG_STACKRECURSE
1105
1106 The output is an integer that is set to one if internal recursion when
1107 running pcre_exec() is implemented by recursive function calls that use
1108 the stack to remember their state. This is the usual way that PCRE is
1109 compiled. The output is zero if PCRE was compiled to use blocks of data
1110 on the heap instead of recursive function calls. In this case,
1111 pcre_stack_malloc and pcre_stack_free are called to manage memory
1112 blocks on the heap, thus avoiding the use of the stack.
1113
1114
1115 COMPILING A PATTERN
1116
1117 pcre *pcre_compile(const char *pattern, int options,
1118 const char **errptr, int *erroffset,
1119 const unsigned char *tableptr);
1120
1121 pcre *pcre_compile2(const char *pattern, int options,
1122 int *errorcodeptr,
1123 const char **errptr, int *erroffset,
1124 const unsigned char *tableptr);
1125
1126 Either of the functions pcre_compile() or pcre_compile2() can be called
1127 to compile a pattern into an internal form. The only difference between
1128 the two interfaces is that pcre_compile2() has an additional argument,
1129 errorcodeptr, via which a numerical error code can be returned.
1130
1131 The pattern is a C string terminated by a binary zero, and is passed in
1132 the pattern argument. A pointer to a single block of memory that is
1133 obtained via pcre_malloc is returned. This contains the compiled code
1134 and related data. The pcre type is defined for the returned block; this
1135 is a typedef for a structure whose contents are not externally defined.
1136 It is up to the caller to free the memory (via pcre_free) when it is no
1137 longer required.
1138
1139 Although the compiled code of a PCRE regex is relocatable, that is, it
1140 does not depend on memory location, the complete pcre data block is not
1141 fully relocatable, because it may contain a copy of the tableptr argu-
1142 ment, which is an address (see below).
1143
1144 The options argument contains various bit settings that affect the com-
1145 pilation. It should be zero if no options are required. The available
1146 options are described below. Some of them (in particular, those that
1147 are compatible with Perl, but also some others) can also be set and
1148 unset from within the pattern (see the detailed description in the
1149 pcrepattern documentation). For those options that can be different in
1150 different parts of the pattern, the contents of the options argument
1151 specifies their initial settings at the start of compilation and execu-
1152 tion. The PCRE_ANCHORED and PCRE_NEWLINE_xxx options can be set at the
1153 time of matching as well as at compile time.
1154
1155 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
1156 if compilation of a pattern fails, pcre_compile() returns NULL, and
1157 sets the variable pointed to by errptr to point to a textual error mes-
1158 sage. This is a static string that is part of the library. You must not
1159 try to free it. The offset from the start of the pattern to the charac-
1160 ter where the error was discovered is placed in the variable pointed to
1161 by erroffset, which must not be NULL. If it is, an immediate error is
1162 given.
1163
1164 If pcre_compile2() is used instead of pcre_compile(), and the error-
1165 codeptr argument is not NULL, a non-zero error code number is returned
1166 via this argument in the event of an error. This is in addition to the
1167 textual error message. Error codes and messages are listed below.
1168
1169 If the final argument, tableptr, is NULL, PCRE uses a default set of
1170 character tables that are built when PCRE is compiled, using the
1171 default C locale. Otherwise, tableptr must be an address that is the
1172 result of a call to pcre_maketables(). This value is stored with the
1173 compiled pattern, and used again by pcre_exec(), unless another table
1174 pointer is passed to it. For more discussion, see the section on locale
1175 support below.
1176
1177 This code fragment shows a typical straightforward call to pcre_com-
1178 pile():
1179
1180 pcre *re;
1181 const char *error;
1182 int erroffset;
1183 re = pcre_compile(
1184 "^A.*Z", /* the pattern */
1185 0, /* default options */
1186 &error, /* for error message */
1187 &erroffset, /* for error offset */
1188 NULL); /* use default character tables */
1189
1190 The following names for option bits are defined in the pcre.h header
1191 file:
1192
1193 PCRE_ANCHORED
1194
1195 If this bit is set, the pattern is forced to be "anchored", that is, it
1196 is constrained to match only at the first matching point in the string
1197 that is being searched (the "subject string"). This effect can also be
1198 achieved by appropriate constructs in the pattern itself, which is the
1199 only way to do it in Perl.
1200
1201 PCRE_AUTO_CALLOUT
1202
1203 If this bit is set, pcre_compile() automatically inserts callout items,
1204 all with number 255, before each pattern item. For discussion of the
1205 callout facility, see the pcrecallout documentation.
1206
1207 PCRE_BSR_ANYCRLF
1208 PCRE_BSR_UNICODE
1209
1210 These options (which are mutually exclusive) control what the \R escape
1211 sequence matches. The choice is either to match only CR, LF, or CRLF,
1212 or to match any Unicode newline sequence. The default is specified when
1213 PCRE is built. It can be overridden from within the pattern, or by set-
1214 ting an option when a compiled pattern is matched.
1215
1216 PCRE_CASELESS
1217
1218 If this bit is set, letters in the pattern match both upper and lower
1219 case letters. It is equivalent to Perl's /i option, and it can be
1220 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
1221 always understands the concept of case for characters whose values are
1222 less than 128, so caseless matching is always possible. For characters
1223 with higher values, the concept of case is supported if PCRE is com-
1224 piled with Unicode property support, but not otherwise. If you want to
1225 use caseless matching for characters 128 and above, you must ensure
1226 that PCRE is compiled with Unicode property support as well as with
1227 UTF-8 support.
1228
1229 PCRE_DOLLAR_ENDONLY
1230
1231 If this bit is set, a dollar metacharacter in the pattern matches only
1232 at the end of the subject string. Without this option, a dollar also
1233 matches immediately before a newline at the end of the string (but not
1234 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
1235 if PCRE_MULTILINE is set. There is no equivalent to this option in
1236 Perl, and no way to set it within a pattern.
1237
1238 PCRE_DOTALL
1239
1240 If this bit is set, a dot metacharater in the pattern matches all char-
1241 acters, including those that indicate newline. Without it, a dot does
1242 not match when the current position is at a newline. This option is
1243 equivalent to Perl's /s option, and it can be changed within a pattern
1244 by a (?s) option setting. A negative class such as [^a] always matches
1245 newline characters, independent of the setting of this option.
1246
1247 PCRE_DUPNAMES
1248
1249 If this bit is set, names used to identify capturing subpatterns need
1250 not be unique. This can be helpful for certain types of pattern when it
1251 is known that only one instance of the named subpattern can ever be
1252 matched. There are more details of named subpatterns below; see also
1253 the pcrepattern documentation.
1254
1255 PCRE_EXTENDED
1256
1257 If this bit is set, whitespace data characters in the pattern are
1258 totally ignored except when escaped or inside a character class. White-
1259 space does not include the VT character (code 11). In addition, charac-
1260 ters between an unescaped # outside a character class and the next new-
1261 line, inclusive, are also ignored. This is equivalent to Perl's /x
1262 option, and it can be changed within a pattern by a (?x) option set-
1263 ting.
1264
1265 This option makes it possible to include comments inside complicated
1266 patterns. Note, however, that this applies only to data characters.
1267 Whitespace characters may never appear within special character
1268 sequences in a pattern, for example within the sequence (?( which
1269 introduces a conditional subpattern.
1270
1271 PCRE_EXTRA
1272
1273 This option was invented in order to turn on additional functionality
1274 of PCRE that is incompatible with Perl, but it is currently of very
1275 little use. When set, any backslash in a pattern that is followed by a
1276 letter that has no special meaning causes an error, thus reserving
1277 these combinations for future expansion. By default, as in Perl, a
1278 backslash followed by a letter with no special meaning is treated as a
1279 literal. (Perl can, however, be persuaded to give a warning for this.)
1280 There are at present no other features controlled by this option. It
1281 can also be set by a (?X) option setting within a pattern.
1282
1283 PCRE_FIRSTLINE
1284
1285 If this option is set, an unanchored pattern is required to match
1286 before or at the first newline in the subject string, though the
1287 matched text may continue over the newline.
1288
1289 PCRE_JAVASCRIPT_COMPAT
1290
1291 If this option is set, PCRE's behaviour is changed in some ways so that
1292 it is compatible with JavaScript rather than Perl. The changes are as
1293 follows:
1294
1295 (1) A lone closing square bracket in a pattern causes a compile-time
1296 error, because this is illegal in JavaScript (by default it is treated
1297 as a data character). Thus, the pattern AB]CD becomes illegal when this
1298 option is set.
1299
1300 (2) At run time, a back reference to an unset subpattern group matches
1301 an empty string (by default this causes the current matching alterna-
1302 tive to fail). A pattern such as (\1)(a) succeeds when this option is
1303 set (assuming it can find an "a" in the subject), whereas it fails by
1304 default, for Perl compatibility.
1305
1306 PCRE_MULTILINE
1307
1308 By default, PCRE treats the subject string as consisting of a single
1309 line of characters (even if it actually contains newlines). The "start
1310 of line" metacharacter (^) matches only at the start of the string,
1311 while the "end of line" metacharacter ($) matches only at the end of
1312 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1313 is set). This is the same as Perl.
1314
1315 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1316 constructs match immediately following or immediately before internal
1317 newlines in the subject string, respectively, as well as at the very
1318 start and end. This is equivalent to Perl's /m option, and it can be
1319 changed within a pattern by a (?m) option setting. If there are no new-
1320 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1321 setting PCRE_MULTILINE has no effect.
1322
1323 PCRE_NEWLINE_CR
1324 PCRE_NEWLINE_LF
1325 PCRE_NEWLINE_CRLF
1326 PCRE_NEWLINE_ANYCRLF
1327 PCRE_NEWLINE_ANY
1328
1329 These options override the default newline definition that was chosen
1330 when PCRE was built. Setting the first or the second specifies that a
1331 newline is indicated by a single character (CR or LF, respectively).
1332 Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
1333 two-character CRLF sequence. Setting PCRE_NEWLINE_ANYCRLF specifies
1334 that any of the three preceding sequences should be recognized. Setting
1335 PCRE_NEWLINE_ANY specifies that any Unicode newline sequence should be
1336 recognized. The Unicode newline sequences are the three just mentioned,
1337 plus the single characters VT (vertical tab, U+000B), FF (formfeed,
1338 U+000C), NEL (next line, U+0085), LS (line separator, U+2028), and PS
1339 (paragraph separator, U+2029). The last two are recognized only in
1340 UTF-8 mode.
1341
1342 The newline setting in the options word uses three bits that are
1343 treated as a number, giving eight possibilities. Currently only six are
1344 used (default plus the five values above). This means that if you set
1345 more than one newline option, the combination may or may not be sensi-
1346 ble. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equivalent to
1347 PCRE_NEWLINE_CRLF, but other combinations may yield unused numbers and
1348 cause an error.
1349
1350 The only time that a line break is specially recognized when compiling
1351 a pattern is if PCRE_EXTENDED is set, and an unescaped # outside a
1352 character class is encountered. This indicates a comment that lasts
1353 until after the next line break sequence. In other circumstances, line
1354 break sequences are treated as literal data, except that in
1355 PCRE_EXTENDED mode, both CR and LF are treated as whitespace characters
1356 and are therefore ignored.
1357
1358 The newline option that is set at compile time becomes the default that
1359 is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1360
1361 PCRE_NO_AUTO_CAPTURE
1362
1363 If this option is set, it disables the use of numbered capturing paren-
1364 theses in the pattern. Any opening parenthesis that is not followed by
1365 ? behaves as if it were followed by ?: but named parentheses can still
1366 be used for capturing (and they acquire numbers in the usual way).
1367 There is no equivalent of this option in Perl.
1368
1369 PCRE_UNGREEDY
1370
1371 This option inverts the "greediness" of the quantifiers so that they
1372 are not greedy by default, but become greedy if followed by "?". It is
1373 not compatible with Perl. It can also be set by a (?U) option setting
1374 within the pattern.
1375
1376 PCRE_UTF8
1377
1378 This option causes PCRE to regard both the pattern and the subject as
1379 strings of UTF-8 characters instead of single-byte character strings.
1380 However, it is available only when PCRE is built to include UTF-8 sup-
1381 port. If not, the use of this option provokes an error. Details of how
1382 this option changes the behaviour of PCRE are given in the section on
1383 UTF-8 support in the main pcre page.
1384
1385 PCRE_NO_UTF8_CHECK
1386
1387 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1388 automatically checked. There is a discussion about the validity of
1389 UTF-8 strings in the main pcre page. If an invalid UTF-8 sequence of
1390 bytes is found, pcre_compile() returns an error. If you already know
1391 that your pattern is valid, and you want to skip this check for perfor-
1392 mance reasons, you can set the PCRE_NO_UTF8_CHECK option. When it is
1393 set, the effect of passing an invalid UTF-8 string as a pattern is
1394 undefined. It may cause your program to crash. Note that this option
1395 can also be passed to pcre_exec() and pcre_dfa_exec(), to suppress the
1396 UTF-8 validity checking of subject strings.
1397
1398
1399 COMPILATION ERROR CODES
1400
1401 The following table lists the error codes than may be returned by
1402 pcre_compile2(), along with the error messages that may be returned by
1403 both compiling functions. As PCRE has developed, some error codes have
1404 fallen out of use. To avoid confusion, they have not been re-used.
1405
1406 0 no error
1407 1 \ at end of pattern
1408 2 \c at end of pattern
1409 3 unrecognized character follows \
1410 4 numbers out of order in {} quantifier
1411 5 number too big in {} quantifier
1412 6 missing terminating ] for character class
1413 7 invalid escape sequence in character class
1414 8 range out of order in character class
1415 9 nothing to repeat
1416 10 [this code is not in use]
1417 11 internal error: unexpected repeat
1418 12 unrecognized character after (? or (?-
1419 13 POSIX named classes are supported only within a class
1420 14 missing )
1421 15 reference to non-existent subpattern
1422 16 erroffset passed as NULL
1423 17 unknown option bit(s) set
1424 18 missing ) after comment
1425 19 [this code is not in use]
1426 20 regular expression is too large
1427 21 failed to get memory
1428 22 unmatched parentheses
1429 23 internal error: code overflow
1430 24 unrecognized character after (?<
1431 25 lookbehind assertion is not fixed length
1432 26 malformed number or name after (?(
1433 27 conditional group contains more than two branches
1434 28 assertion expected after (?(
1435 29 (?R or (?[+-]digits must be followed by )
1436 30 unknown POSIX class name
1437 31 POSIX collating elements are not supported
1438 32 this version of PCRE is not compiled with PCRE_UTF8 support
1439 33 [this code is not in use]
1440 34 character value in \x{...} sequence is too large
1441 35 invalid condition (?(0)
1442 36 \C not allowed in lookbehind assertion
1443 37 PCRE does not support \L, \l, \N, \U, or \u
1444 38 number after (?C is > 255
1445 39 closing ) for (?C expected
1446 40 recursive call could loop indefinitely
1447 41 unrecognized character after (?P
1448 42 syntax error in subpattern name (missing terminator)
1449 43 two named subpatterns have the same name
1450 44 invalid UTF-8 string
1451 45 support for \P, \p, and \X has not been compiled
1452 46 malformed \P or \p sequence
1453 47 unknown property name after \P or \p
1454 48 subpattern name is too long (maximum 32 characters)
1455 49 too many named subpatterns (maximum 10000)
1456 50 [this code is not in use]
1457 51 octal value is greater than \377 (not in UTF-8 mode)
1458 52 internal error: overran compiling workspace
1459 53 internal error: previously-checked referenced subpattern not
1460 found
1461 54 DEFINE group contains more than one branch
1462 55 repeating a DEFINE group is not allowed
1463 56 inconsistent NEWLINE options
1464 57 \g is not followed by a braced, angle-bracketed, or quoted
1465 name/number or by a plain number
1466 58 a numbered reference must not be zero
1467 59 (*VERB) with an argument is not supported
1468 60 (*VERB) not recognized
1469 61 number is too big
1470 62 subpattern name expected
1471 63 digit expected after (?+
1472 64 ] is an invalid data character in JavaScript compatibility mode
1473
1474 The numbers 32 and 10000 in errors 48 and 49 are defaults; different
1475 values may be used if the limits were changed when PCRE was built.
1476
1477
1478 STUDYING A PATTERN
1479
1480 pcre_extra *pcre_study(const pcre *code, int options
1481 const char **errptr);
1482
1483 If a compiled pattern is going to be used several times, it is worth
1484 spending more time analyzing it in order to speed up the time taken for
1485 matching. The function pcre_study() takes a pointer to a compiled pat-
1486 tern as its first argument. If studying the pattern produces additional
1487 information that will help speed up matching, pcre_study() returns a
1488 pointer to a pcre_extra block, in which the study_data field points to
1489 the results of the study.
1490
1491 The returned value from pcre_study() can be passed directly to
1492 pcre_exec(). However, a pcre_extra block also contains other fields
1493 that can be set by the caller before the block is passed; these are
1494 described below in the section on matching a pattern.
1495
1496 If studying the pattern does not produce any additional information
1497 pcre_study() returns NULL. In that circumstance, if the calling program
1498 wants to pass any of the other fields to pcre_exec(), it must set up
1499 its own pcre_extra block.
1500
1501 The second argument of pcre_study() contains option bits. At present,
1502 no options are defined, and this argument should always be zero.
1503
1504 The third argument for pcre_study() is a pointer for an error message.
1505 If studying succeeds (even if no data is returned), the variable it
1506 points to is set to NULL. Otherwise it is set to point to a textual
1507 error message. This is a static string that is part of the library. You
1508 must not try to free it. You should test the error pointer for NULL
1509 after calling pcre_study(), to be sure that it has run successfully.
1510
1511 This is a typical call to pcre_study():
1512
1513 pcre_extra *pe;
1514 pe = pcre_study(
1515 re, /* result of pcre_compile() */
1516 0, /* no options exist */
1517 &error); /* set to NULL or points to a message */
1518
1519 At present, studying a pattern is useful only for non-anchored patterns
1520 that do not have a single fixed starting character. A bitmap of possi-
1521 ble starting bytes is created.
1522
1523
1524 LOCALE SUPPORT
1525
1526 PCRE handles caseless matching, and determines whether characters are
1527 letters, digits, or whatever, by reference to a set of tables, indexed
1528 by character value. When running in UTF-8 mode, this applies only to
1529 characters with codes less than 128. Higher-valued codes never match
1530 escapes such as \w or \d, but can be tested with \p if PCRE is built
1531 with Unicode character property support. The use of locales with Uni-
1532 code is discouraged. If you are handling characters with codes greater
1533 than 128, you should either use UTF-8 and Unicode, or use locales, but
1534 not try to mix the two.
1535
1536 PCRE contains an internal set of tables that are used when the final
1537 argument of pcre_compile() is NULL. These are sufficient for many
1538 applications. Normally, the internal tables recognize only ASCII char-
1539 acters. However, when PCRE is built, it is possible to cause the inter-
1540 nal tables to be rebuilt in the default "C" locale of the local system,
1541 which may cause them to be different.
1542
1543 The internal tables can always be overridden by tables supplied by the
1544 application that calls PCRE. These may be created in a different locale
1545 from the default. As more and more applications change to using Uni-
1546 code, the need for this locale support is expected to die away.
1547
1548 External tables are built by calling the pcre_maketables() function,
1549 which has no arguments, in the relevant locale. The result can then be
1550 passed to pcre_compile() or pcre_exec() as often as necessary. For
1551 example, to build and use tables that are appropriate for the French
1552 locale (where accented characters with values greater than 128 are
1553 treated as letters), the following code could be used:
1554
1555 setlocale(LC_CTYPE, "fr_FR");
1556 tables = pcre_maketables();
1557 re = pcre_compile(..., tables);
1558
1559 The locale name "fr_FR" is used on Linux and other Unix-like systems;
1560 if you are using Windows, the name for the French locale is "french".
1561
1562 When pcre_maketables() runs, the tables are built in memory that is
1563 obtained via pcre_malloc. It is the caller's responsibility to ensure
1564 that the memory containing the tables remains available for as long as
1565 it is needed.
1566
1567 The pointer that is passed to pcre_compile() is saved with the compiled
1568 pattern, and the same tables are used via this pointer by pcre_study()
1569 and normally also by pcre_exec(). Thus, by default, for any single pat-
1570 tern, compilation, studying and matching all happen in the same locale,
1571 but different patterns can be compiled in different locales.
1572
1573 It is possible to pass a table pointer or NULL (indicating the use of
1574 the internal tables) to pcre_exec(). Although not intended for this
1575 purpose, this facility could be used to match a pattern in a different
1576 locale from the one in which it was compiled. Passing table pointers at
1577 run time is discussed below in the section on matching a pattern.
1578
1579
1580 INFORMATION ABOUT A PATTERN
1581
1582 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1583 int what, void *where);
1584
1585 The pcre_fullinfo() function returns information about a compiled pat-
1586 tern. It replaces the obsolete pcre_info() function, which is neverthe-
1587 less retained for backwards compability (and is documented below).
1588
1589 The first argument for pcre_fullinfo() is a pointer to the compiled
1590 pattern. The second argument is the result of pcre_study(), or NULL if
1591 the pattern was not studied. The third argument specifies which piece
1592 of information is required, and the fourth argument is a pointer to a
1593 variable to receive the data. The yield of the function is zero for
1594 success, or one of the following negative numbers:
1595
1596 PCRE_ERROR_NULL the argument code was NULL
1597 the argument where was NULL
1598 PCRE_ERROR_BADMAGIC the "magic number" was not found
1599 PCRE_ERROR_BADOPTION the value of what was invalid
1600
1601 The "magic number" is placed at the start of each compiled pattern as
1602 an simple check against passing an arbitrary memory pointer. Here is a
1603 typical call of pcre_fullinfo(), to obtain the length of the compiled
1604 pattern:
1605
1606 int rc;
1607 size_t length;
1608 rc = pcre_fullinfo(
1609 re, /* result of pcre_compile() */
1610 pe, /* result of pcre_study(), or NULL */
1611 PCRE_INFO_SIZE, /* what is required */
1612 &length); /* where to put the data */
1613
1614 The possible values for the third argument are defined in pcre.h, and
1615 are as follows:
1616
1617 PCRE_INFO_BACKREFMAX
1618
1619 Return the number of the highest back reference in the pattern. The
1620 fourth argument should point to an int variable. Zero is returned if
1621 there are no back references.
1622
1623 PCRE_INFO_CAPTURECOUNT
1624
1625 Return the number of capturing subpatterns in the pattern. The fourth
1626 argument should point to an int variable.
1627
1628 PCRE_INFO_DEFAULT_TABLES
1629
1630 Return a pointer to the internal default character tables within PCRE.
1631 The fourth argument should point to an unsigned char * variable. This
1632 information call is provided for internal use by the pcre_study() func-
1633 tion. External callers can cause PCRE to use its internal tables by
1634 passing a NULL table pointer.
1635
1636 PCRE_INFO_FIRSTBYTE
1637
1638 Return information about the first byte of any matched string, for a
1639 non-anchored pattern. The fourth argument should point to an int vari-
1640 able. (This option used to be called PCRE_INFO_FIRSTCHAR; the old name
1641 is still recognized for backwards compatibility.)
1642
1643 If there is a fixed first byte, for example, from a pattern such as
1644 (cat|cow|coyote), its value is returned. Otherwise, if either
1645
1646 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
1647 branch starts with "^", or
1648
1649 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
1650 set (if it were set, the pattern would be anchored),
1651
1652 -1 is returned, indicating that the pattern matches only at the start
1653 of a subject string or after any newline within the string. Otherwise
1654 -2 is returned. For anchored patterns, -2 is returned.
1655
1656 PCRE_INFO_FIRSTTABLE
1657
1658 If the pattern was studied, and this resulted in the construction of a
1659 256-bit table indicating a fixed set of bytes for the first byte in any
1660 matching string, a pointer to the table is returned. Otherwise NULL is
1661 returned. The fourth argument should point to an unsigned char * vari-
1662 able.
1663
1664 PCRE_INFO_HASCRORLF
1665
1666 Return 1 if the pattern contains any explicit matches for CR or LF
1667 characters, otherwise 0. The fourth argument should point to an int
1668 variable. An explicit match is either a literal CR or LF character, or
1669 \r or \n.
1670
1671 PCRE_INFO_JCHANGED
1672
1673 Return 1 if the (?J) or (?-J) option setting is used in the pattern,
1674 otherwise 0. The fourth argument should point to an int variable. (?J)
1675 and (?-J) set and unset the local PCRE_DUPNAMES option, respectively.
1676
1677 PCRE_INFO_LASTLITERAL
1678
1679 Return the value of the rightmost literal byte that must exist in any
1680 matched string, other than at its start, if such a byte has been
1681 recorded. The fourth argument should point to an int variable. If there
1682 is no such byte, -1 is returned. For anchored patterns, a last literal
1683 byte is recorded only if it follows something of variable length. For
1684 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
1685 /^a\dz\d/ the returned value is -1.
1686
1687 PCRE_INFO_NAMECOUNT
1688 PCRE_INFO_NAMEENTRYSIZE
1689 PCRE_INFO_NAMETABLE
1690
1691 PCRE supports the use of named as well as numbered capturing parenthe-
1692 ses. The names are just an additional way of identifying the parenthe-
1693 ses, which still acquire numbers. Several convenience functions such as
1694 pcre_get_named_substring() are provided for extracting captured sub-
1695 strings by name. It is also possible to extract the data directly, by
1696 first converting the name to a number in order to access the correct
1697 pointers in the output vector (described with pcre_exec() below). To do
1698 the conversion, you need to use the name-to-number map, which is
1699 described by these three values.
1700
1701 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
1702 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
1703 of each entry; both of these return an int value. The entry size
1704 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
1705 a pointer to the first entry of the table (a pointer to char). The
1706 first two bytes of each entry are the number of the capturing parenthe-
1707 sis, most significant byte first. The rest of the entry is the corre-
1708 sponding name, zero terminated. The names are in alphabetical order.
1709 When PCRE_DUPNAMES is set, duplicate names are in order of their paren-
1710 theses numbers. For example, consider the following pattern (assume
1711 PCRE_EXTENDED is set, so white space - including newlines - is
1712 ignored):
1713
1714 (?<date> (?<year>(\d\d)?\d\d) -
1715 (?<month>\d\d) - (?<day>\d\d) )
1716
1717 There are four named subpatterns, so the table has four entries, and
1718 each entry in the table is eight bytes long. The table is as follows,
1719 with non-printing bytes shows in hexadecimal, and undefined bytes shown
1720 as ??:
1721
1722 00 01 d a t e 00 ??
1723 00 05 d a y 00 ?? ??
1724 00 04 m o n t h 00
1725 00 02 y e a r 00 ??
1726
1727 When writing code to extract data from named subpatterns using the
1728 name-to-number map, remember that the length of the entries is likely
1729 to be different for each compiled pattern.
1730
1731 PCRE_INFO_OKPARTIAL
1732
1733 Return 1 if the pattern can be used for partial matching with
1734 pcre_exec(), otherwise 0. The fourth argument should point to an int
1735 variable. From release 8.00, this always returns 1, because the
1736 restrictions that previously applied to partial matching have been
1737 lifted. The pcrepartial documentation gives details of partial match-
1738 ing.
1739
1740 PCRE_INFO_OPTIONS
1741
1742 Return a copy of the options with which the pattern was compiled. The
1743 fourth argument should point to an unsigned long int variable. These
1744 option bits are those specified in the call to pcre_compile(), modified
1745 by any top-level option settings at the start of the pattern itself. In
1746 other words, they are the options that will be in force when matching
1747 starts. For example, if the pattern /(?im)abc(?-i)d/ is compiled with
1748 the PCRE_EXTENDED option, the result is PCRE_CASELESS, PCRE_MULTILINE,
1749 and PCRE_EXTENDED.
1750
1751 A pattern is automatically anchored by PCRE if all of its top-level
1752 alternatives begin with one of the following:
1753
1754 ^ unless PCRE_MULTILINE is set
1755 \A always
1756 \G always
1757 .* if PCRE_DOTALL is set and there are no back
1758 references to the subpattern in which .* appears
1759
1760 For such patterns, the PCRE_ANCHORED bit is set in the options returned
1761 by pcre_fullinfo().
1762
1763 PCRE_INFO_SIZE
1764
1765 Return the size of the compiled pattern, that is, the value that was
1766 passed as the argument to pcre_malloc() when PCRE was getting memory in
1767 which to place the compiled data. The fourth argument should point to a
1768 size_t variable.
1769
1770 PCRE_INFO_STUDYSIZE
1771
1772 Return the size of the data block pointed to by the study_data field in
1773 a pcre_extra block. That is, it is the value that was passed to
1774 pcre_malloc() when PCRE was getting memory into which to place the data
1775 created by pcre_study(). The fourth argument should point to a size_t
1776 variable.
1777
1778
1779 OBSOLETE INFO FUNCTION
1780
1781 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
1782
1783 The pcre_info() function is now obsolete because its interface is too
1784 restrictive to return all the available data about a compiled pattern.
1785 New programs should use pcre_fullinfo() instead. The yield of
1786 pcre_info() is the number of capturing subpatterns, or one of the fol-
1787 lowing negative numbers:
1788
1789 PCRE_ERROR_NULL the argument code was NULL
1790 PCRE_ERROR_BADMAGIC the "magic number" was not found
1791
1792 If the optptr argument is not NULL, a copy of the options with which
1793 the pattern was compiled is placed in the integer it points to (see
1794 PCRE_INFO_OPTIONS above).
1795
1796 If the pattern is not anchored and the firstcharptr argument is not
1797 NULL, it is used to pass back information about the first character of
1798 any matched string (see PCRE_INFO_FIRSTBYTE above).
1799
1800
1801 REFERENCE COUNTS
1802
1803 int pcre_refcount(pcre *code, int adjust);
1804
1805 The pcre_refcount() function is used to maintain a reference count in
1806 the data block that contains a compiled pattern. It is provided for the
1807 benefit of applications that operate in an object-oriented manner,
1808 where different parts of the application may be using the same compiled
1809 pattern, but you want to free the block when they are all done.
1810
1811 When a pattern is compiled, the reference count field is initialized to
1812 zero. It is changed only by calling this function, whose action is to
1813 add the adjust value (which may be positive or negative) to it. The
1814 yield of the function is the new value. However, the value of the count
1815 is constrained to lie between 0 and 65535, inclusive. If the new value
1816 is outside these limits, it is forced to the appropriate limit value.
1817
1818 Except when it is zero, the reference count is not correctly preserved
1819 if a pattern is compiled on one host and then transferred to a host
1820 whose byte-order is different. (This seems a highly unlikely scenario.)
1821
1822
1823 MATCHING A PATTERN: THE TRADITIONAL FUNCTION
1824
1825 int pcre_exec(const pcre *code, const pcre_extra *extra,
1826 const char *subject, int length, int startoffset,
1827 int options, int *ovector, int ovecsize);
1828
1829 The function pcre_exec() is called to match a subject string against a
1830 compiled pattern, which is passed in the code argument. If the pattern
1831 has been studied, the result of the study should be passed in the extra
1832 argument. This function is the main matching facility of the library,
1833 and it operates in a Perl-like manner. For specialist use there is also
1834 an alternative matching function, which is described below in the sec-
1835 tion about the pcre_dfa_exec() function.
1836
1837 In most applications, the pattern will have been compiled (and option-
1838 ally studied) in the same process that calls pcre_exec(). However, it
1839 is possible to save compiled patterns and study data, and then use them
1840 later in different processes, possibly even on different hosts. For a
1841 discussion about this, see the pcreprecompile documentation.
1842
1843 Here is an example of a simple call to pcre_exec():
1844
1845 int rc;
1846 int ovector[30];
1847 rc = pcre_exec(
1848 re, /* result of pcre_compile() */
1849 NULL, /* we didn't study the pattern */
1850 "some string", /* the subject string */
1851 11, /* the length of the subject string */
1852 0, /* start at offset 0 in the subject */
1853 0, /* default options */
1854 ovector, /* vector of integers for substring information */
1855 30); /* number of elements (NOT size in bytes) */
1856
1857 Extra data for pcre_exec()
1858
1859 If the extra argument is not NULL, it must point to a pcre_extra data
1860 block. The pcre_study() function returns such a block (when it doesn't
1861 return NULL), but you can also create one for yourself, and pass addi-
1862 tional information in it. The pcre_extra block contains the following
1863 fields (not necessarily in this order):
1864
1865 unsigned long int flags;
1866 void *study_data;
1867 unsigned long int match_limit;
1868 unsigned long int match_limit_recursion;
1869 void *callout_data;
1870 const unsigned char *tables;
1871
1872 The flags field is a bitmap that specifies which of the other fields
1873 are set. The flag bits are:
1874
1875 PCRE_EXTRA_STUDY_DATA
1876 PCRE_EXTRA_MATCH_LIMIT
1877 PCRE_EXTRA_MATCH_LIMIT_RECURSION
1878 PCRE_EXTRA_CALLOUT_DATA
1879 PCRE_EXTRA_TABLES
1880
1881 Other flag bits should be set to zero. The study_data field is set in
1882 the pcre_extra block that is returned by pcre_study(), together with
1883 the appropriate flag bit. You should not set this yourself, but you may
1884 add to the block by setting the other fields and their corresponding
1885 flag bits.
1886
1887 The match_limit field provides a means of preventing PCRE from using up
1888 a vast amount of resources when running patterns that are not going to
1889 match, but which have a very large number of possibilities in their
1890 search trees. The classic example is the use of nested unlimited
1891 repeats.
1892
1893 Internally, PCRE uses a function called match() which it calls repeat-
1894 edly (sometimes recursively). The limit set by match_limit is imposed
1895 on the number of times this function is called during a match, which
1896 has the effect of limiting the amount of backtracking that can take
1897 place. For patterns that are not anchored, the count restarts from zero
1898 for each position in the subject string.
1899
1900 The default value for the limit can be set when PCRE is built; the
1901 default default is 10 million, which handles all but the most extreme
1902 cases. You can override the default by suppling pcre_exec() with a
1903 pcre_extra block in which match_limit is set, and
1904 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
1905 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
1906
1907 The match_limit_recursion field is similar to match_limit, but instead
1908 of limiting the total number of times that match() is called, it limits
1909 the depth of recursion. The recursion depth is a smaller number than
1910 the total number of calls, because not all calls to match() are recur-
1911 sive. This limit is of use only if it is set smaller than match_limit.
1912
1913 Limiting the recursion depth limits the amount of stack that can be
1914 used, or, when PCRE has been compiled to use memory on the heap instead
1915 of the stack, the amount of heap memory that can be used.
1916
1917 The default value for match_limit_recursion can be set when PCRE is
1918 built; the default default is the same value as the default for
1919 match_limit. You can override the default by suppling pcre_exec() with
1920 a pcre_extra block in which match_limit_recursion is set, and
1921 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
1922 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
1923
1924 The callout_data field is used in conjunction with the "callout" fea-
1925 ture, and is described in the pcrecallout documentation.
1926
1927 The tables field is used to pass a character tables pointer to
1928 pcre_exec(); this overrides the value that is stored with the compiled
1929 pattern. A non-NULL value is stored with the compiled pattern only if
1930 custom tables were supplied to pcre_compile() via its tableptr argu-
1931 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
1932 PCRE's internal tables to be used. This facility is helpful when re-
1933 using patterns that have been saved after compiling with an external
1934 set of tables, because the external tables might be at a different
1935 address when pcre_exec() is called. See the pcreprecompile documenta-
1936 tion for a discussion of saving compiled patterns for later use.
1937
1938 Option bits for pcre_exec()
1939
1940 The unused bits of the options argument for pcre_exec() must be zero.
1941 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
1942 PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NOTEMPTY_ATSTART,
1943 PCRE_NO_START_OPTIMIZE, PCRE_NO_UTF8_CHECK, PCRE_PARTIAL_SOFT, and
1944 PCRE_PARTIAL_HARD.
1945
1946 PCRE_ANCHORED
1947
1948 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
1949 matching position. If a pattern was compiled with PCRE_ANCHORED, or
1950 turned out to be anchored by virtue of its contents, it cannot be made
1951 unachored at matching time.
1952
1953 PCRE_BSR_ANYCRLF
1954 PCRE_BSR_UNICODE
1955
1956 These options (which are mutually exclusive) control what the \R escape
1957 sequence matches. The choice is either to match only CR, LF, or CRLF,
1958 or to match any Unicode newline sequence. These options override the
1959 choice that was made or defaulted when the pattern was compiled.
1960
1961 PCRE_NEWLINE_CR
1962 PCRE_NEWLINE_LF
1963 PCRE_NEWLINE_CRLF
1964 PCRE_NEWLINE_ANYCRLF
1965 PCRE_NEWLINE_ANY
1966
1967 These options override the newline definition that was chosen or
1968 defaulted when the pattern was compiled. For details, see the descrip-
1969 tion of pcre_compile() above. During matching, the newline choice
1970 affects the behaviour of the dot, circumflex, and dollar metacharac-
1971 ters. It may also alter the way the match position is advanced after a
1972 match failure for an unanchored pattern.
1973
1974 When PCRE_NEWLINE_CRLF, PCRE_NEWLINE_ANYCRLF, or PCRE_NEWLINE_ANY is
1975 set, and a match attempt for an unanchored pattern fails when the cur-
1976 rent position is at a CRLF sequence, and the pattern contains no
1977 explicit matches for CR or LF characters, the match position is
1978 advanced by two characters instead of one, in other words, to after the
1979 CRLF.
1980
1981 The above rule is a compromise that makes the most common cases work as
1982 expected. For example, if the pattern is .+A (and the PCRE_DOTALL
1983 option is not set), it does not match the string "\r\nA" because, after
1984 failing at the start, it skips both the CR and the LF before retrying.
1985 However, the pattern [\r\n]A does match that string, because it con-
1986 tains an explicit CR or LF reference, and so advances only by one char-
1987 acter after the first failure.
1988
1989 An explicit match for CR of LF is either a literal appearance of one of
1990 those characters, or one of the \r or \n escape sequences. Implicit
1991 matches such as [^X] do not count, nor does \s (which includes CR and
1992 LF in the characters that it matches).
1993
1994 Notwithstanding the above, anomalous effects may still occur when CRLF
1995 is a valid newline sequence and explicit \r or \n escapes appear in the
1996 pattern.
1997
1998 PCRE_NOTBOL
1999
2000 This option specifies that first character of the subject string is not
2001 the beginning of a line, so the circumflex metacharacter should not
2002 match before it. Setting this without PCRE_MULTILINE (at compile time)
2003 causes circumflex never to match. This option affects only the behav-
2004 iour of the circumflex metacharacter. It does not affect \A.
2005
2006 PCRE_NOTEOL
2007
2008 This option specifies that the end of the subject string is not the end
2009 of a line, so the dollar metacharacter should not match it nor (except
2010 in multiline mode) a newline immediately before it. Setting this with-
2011 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
2012 option affects only the behaviour of the dollar metacharacter. It does
2013 not affect \Z or \z.
2014
2015 PCRE_NOTEMPTY
2016
2017 An empty string is not considered to be a valid match if this option is
2018 set. If there are alternatives in the pattern, they are tried. If all
2019 the alternatives match the empty string, the entire match fails. For
2020 example, if the pattern
2021
2022 a?b?
2023
2024 is applied to a string not beginning with "a" or "b", it matches an
2025 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
2026 match is not valid, so PCRE searches further into the string for occur-
2027 rences of "a" or "b".
2028
2029 PCRE_NOTEMPTY_ATSTART
2030
2031 This is like PCRE_NOTEMPTY, except that an empty string match that is
2032 not at the start of the subject is permitted. If the pattern is
2033 anchored, such a match can occur only if the pattern contains \K.
2034
2035 Perl has no direct equivalent of PCRE_NOTEMPTY or
2036 PCRE_NOTEMPTY_ATSTART, but it does make a special case of a pattern
2037 match of the empty string within its split() function, and when using
2038 the /g modifier. It is possible to emulate Perl's behaviour after
2039 matching a null string by first trying the match again at the same off-
2040 set with PCRE_NOTEMPTY_ATSTART and PCRE_ANCHORED, and then if that
2041 fails, by advancing the starting offset (see below) and trying an ordi-
2042 nary match again. There is some code that demonstrates how to do this
2043 in the pcredemo sample program.
2044
2045 PCRE_NO_START_OPTIMIZE
2046
2047 There are a number of optimizations that pcre_exec() uses at the start
2048 of a match, in order to speed up the process. For example, if it is
2049 known that a match must start with a specific character, it searches
2050 the subject for that character, and fails immediately if it cannot find
2051 it, without actually running the main matching function. When callouts
2052 are in use, these optimizations can cause them to be skipped. This
2053 option disables the "start-up" optimizations, causing performance to
2054 suffer, but ensuring that the callouts do occur.
2055
2056 PCRE_NO_UTF8_CHECK
2057
2058 When PCRE_UTF8 is set at compile time, the validity of the subject as a
2059 UTF-8 string is automatically checked when pcre_exec() is subsequently
2060 called. The value of startoffset is also checked to ensure that it
2061 points to the start of a UTF-8 character. There is a discussion about
2062 the validity of UTF-8 strings in the section on UTF-8 support in the
2063 main pcre page. If an invalid UTF-8 sequence of bytes is found,
2064 pcre_exec() returns the error PCRE_ERROR_BADUTF8. If startoffset con-
2065 tains an invalid value, PCRE_ERROR_BADUTF8_OFFSET is returned.
2066
2067 If you already know that your subject is valid, and you want to skip
2068 these checks for performance reasons, you can set the
2069 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
2070 do this for the second and subsequent calls to pcre_exec() if you are
2071 making repeated calls to find all the matches in a single subject
2072 string. However, you should be sure that the value of startoffset
2073 points to the start of a UTF-8 character. When PCRE_NO_UTF8_CHECK is
2074 set, the effect of passing an invalid UTF-8 string as a subject, or a
2075 value of startoffset that does not point to the start of a UTF-8 char-
2076 acter, is undefined. Your program may crash.
2077
2078 PCRE_PARTIAL_HARD
2079 PCRE_PARTIAL_SOFT
2080
2081 These options turn on the partial matching feature. For backwards com-
2082 patibility, PCRE_PARTIAL is a synonym for PCRE_PARTIAL_SOFT. A partial
2083 match occurs if the end of the subject string is reached successfully,
2084 but there are not enough subject characters to complete the match. If
2085 this happens when PCRE_PARTIAL_HARD is set, pcre_exec() immediately
2086 returns PCRE_ERROR_PARTIAL. Otherwise, if PCRE_PARTIAL_SOFT is set,
2087 matching continues by testing any other alternatives. Only if they all
2088 fail is PCRE_ERROR_PARTIAL returned (instead of PCRE_ERROR_NOMATCH).
2089 The portion of the string that was inspected when the partial match was
2090 found is set as the first matching string. There is a more detailed
2091 discussion in the pcrepartial documentation.
2092
2093 The string to be matched by pcre_exec()
2094
2095 The subject string is passed to pcre_exec() as a pointer in subject, a
2096 length (in bytes) in length, and a starting byte offset in startoffset.
2097 In UTF-8 mode, the byte offset must point to the start of a UTF-8 char-
2098 acter. Unlike the pattern string, the subject may contain binary zero
2099 bytes. When the starting offset is zero, the search for a match starts
2100 at the beginning of the subject, and this is by far the most common
2101 case.
2102
2103 A non-zero starting offset is useful when searching for another match
2104 in the same subject by calling pcre_exec() again after a previous suc-
2105 cess. Setting startoffset differs from just passing over a shortened
2106 string and setting PCRE_NOTBOL in the case of a pattern that begins
2107 with any kind of lookbehind. For example, consider the pattern
2108
2109 \Biss\B
2110
2111 which finds occurrences of "iss" in the middle of words. (\B matches
2112 only if the current position in the subject is not a word boundary.)
2113 When applied to the string "Mississipi" the first call to pcre_exec()
2114 finds the first occurrence. If pcre_exec() is called again with just
2115 the remainder of the subject, namely "issipi", it does not match,
2116 because \B is always false at the start of the subject, which is deemed
2117 to be a word boundary. However, if pcre_exec() is passed the entire
2118 string again, but with startoffset set to 4, it finds the second occur-
2119 rence of "iss" because it is able to look behind the starting point to
2120 discover that it is preceded by a letter.
2121
2122 If a non-zero starting offset is passed when the pattern is anchored,
2123 one attempt to match at the given offset is made. This can only succeed
2124 if the pattern does not require the match to be at the start of the
2125 subject.
2126
2127 How pcre_exec() returns captured substrings
2128
2129 In general, a pattern matches a certain portion of the subject, and in
2130 addition, further substrings from the subject may be picked out by
2131 parts of the pattern. Following the usage in Jeffrey Friedl's book,
2132 this is called "capturing" in what follows, and the phrase "capturing
2133 subpattern" is used for a fragment of a pattern that picks out a sub-
2134 string. PCRE supports several other kinds of parenthesized subpattern
2135 that do not cause substrings to be captured.
2136
2137 Captured substrings are returned to the caller via a vector of integers
2138 whose address is passed in ovector. The number of elements in the vec-
2139 tor is passed in ovecsize, which must be a non-negative number. Note:
2140 this argument is NOT the size of ovector in bytes.
2141
2142 The first two-thirds of the vector is used to pass back captured sub-
2143 strings, each substring using a pair of integers. The remaining third
2144 of the vector is used as workspace by pcre_exec() while matching cap-
2145 turing subpatterns, and is not available for passing back information.
2146 The number passed in ovecsize should always be a multiple of three. If
2147 it is not, it is rounded down.
2148
2149 When a match is successful, information about captured substrings is
2150 returned in pairs of integers, starting at the beginning of ovector,
2151 and continuing up to two-thirds of its length at the most. The first
2152 element of each pair is set to the byte offset of the first character
2153 in a substring, and the second is set to the byte offset of the first
2154 character after the end of a substring. Note: these values are always
2155 byte offsets, even in UTF-8 mode. They are not character counts.
2156
2157 The first pair of integers, ovector[0] and ovector[1], identify the
2158 portion of the subject string matched by the entire pattern. The next
2159 pair is used for the first capturing subpattern, and so on. The value
2160 returned by pcre_exec() is one more than the highest numbered pair that
2161 has been set. For example, if two substrings have been captured, the
2162 returned value is 3. If there are no capturing subpatterns, the return
2163 value from a successful match is 1, indicating that just the first pair
2164 of offsets has been set.
2165
2166 If a capturing subpattern is matched repeatedly, it is the last portion
2167 of the string that it matched that is returned.
2168
2169 If the vector is too small to hold all the captured substring offsets,
2170 it is used as far as possible (up to two-thirds of its length), and the
2171 function returns a value of zero. If the substring offsets are not of
2172 interest, pcre_exec() may be called with ovector passed as NULL and
2173 ovecsize as zero. However, if the pattern contains back references and
2174 the ovector is not big enough to remember the related substrings, PCRE
2175 has to get additional memory for use during matching. Thus it is usu-
2176 ally advisable to supply an ovector.
2177
2178 The pcre_info() function can be used to find out how many capturing
2179 subpatterns there are in a compiled pattern. The smallest size for
2180 ovector that will allow for n captured substrings, in addition to the
2181 offsets of the substring matched by the whole pattern, is (n+1)*3.
2182
2183 It is possible for capturing subpattern number n+1 to match some part
2184 of the subject when subpattern n has not been used at all. For example,
2185 if the string "abc" is matched against the pattern (a|(z))(bc) the
2186 return from the function is 4, and subpatterns 1 and 3 are matched, but
2187 2 is not. When this happens, both values in the offset pairs corre-
2188 sponding to unused subpatterns are set to -1.
2189
2190 Offset values that correspond to unused subpatterns at the end of the
2191 expression are also set to -1. For example, if the string "abc" is
2192 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
2193 matched. The return from the function is 2, because the highest used
2194 capturing subpattern number is 1. However, you can refer to the offsets
2195 for the second and third capturing subpatterns if you wish (assuming
2196 the vector is large enough, of course).
2197
2198 Some convenience functions are provided for extracting the captured
2199 substrings as separate strings. These are described below.
2200
2201 Error return values from pcre_exec()
2202
2203 If pcre_exec() fails, it returns a negative number. The following are
2204 defined in the header file:
2205
2206 PCRE_ERROR_NOMATCH (-1)
2207
2208 The subject string did not match the pattern.
2209
2210 PCRE_ERROR_NULL (-2)
2211
2212 Either code or subject was passed as NULL, or ovector was NULL and
2213 ovecsize was not zero.
2214
2215 PCRE_ERROR_BADOPTION (-3)
2216
2217 An unrecognized bit was set in the options argument.
2218
2219 PCRE_ERROR_BADMAGIC (-4)
2220
2221 PCRE stores a 4-byte "magic number" at the start of the compiled code,
2222 to catch the case when it is passed a junk pointer and to detect when a
2223 pattern that was compiled in an environment of one endianness is run in
2224 an environment with the other endianness. This is the error that PCRE
2225 gives when the magic number is not present.
2226
2227 PCRE_ERROR_UNKNOWN_OPCODE (-5)
2228
2229 While running the pattern match, an unknown item was encountered in the
2230 compiled pattern. This error could be caused by a bug in PCRE or by
2231 overwriting of the compiled pattern.
2232
2233 PCRE_ERROR_NOMEMORY (-6)
2234
2235 If a pattern contains back references, but the ovector that is passed
2236 to pcre_exec() is not big enough to remember the referenced substrings,
2237 PCRE gets a block of memory at the start of matching to use for this
2238 purpose. If the call via pcre_malloc() fails, this error is given. The
2239 memory is automatically freed at the end of matching.
2240
2241 PCRE_ERROR_NOSUBSTRING (-7)
2242
2243 This error is used by the pcre_copy_substring(), pcre_get_substring(),
2244 and pcre_get_substring_list() functions (see below). It is never
2245 returned by pcre_exec().
2246
2247 PCRE_ERROR_MATCHLIMIT (-8)
2248
2249 The backtracking limit, as specified by the match_limit field in a
2250 pcre_extra structure (or defaulted) was reached. See the description
2251 above.
2252
2253 PCRE_ERROR_CALLOUT (-9)
2254
2255 This error is never generated by pcre_exec() itself. It is provided for
2256 use by callout functions that want to yield a distinctive error code.
2257 See the pcrecallout documentation for details.
2258
2259 PCRE_ERROR_BADUTF8 (-10)
2260
2261 A string that contains an invalid UTF-8 byte sequence was passed as a
2262 subject.
2263
2264 PCRE_ERROR_BADUTF8_OFFSET (-11)
2265
2266 The UTF-8 byte sequence that was passed as a subject was valid, but the
2267 value of startoffset did not point to the beginning of a UTF-8 charac-
2268 ter.
2269
2270 PCRE_ERROR_PARTIAL (-12)
2271
2272 The subject string did not match, but it did match partially. See the
2273 pcrepartial documentation for details of partial matching.
2274
2275 PCRE_ERROR_BADPARTIAL (-13)
2276
2277 This code is no longer in use. It was formerly returned when the
2278 PCRE_PARTIAL option was used with a compiled pattern containing items
2279 that were not supported for partial matching. From release 8.00
2280 onwards, there are no restrictions on partial matching.
2281
2282 PCRE_ERROR_INTERNAL (-14)
2283
2284 An unexpected internal error has occurred. This error could be caused
2285 by a bug in PCRE or by overwriting of the compiled pattern.
2286
2287 PCRE_ERROR_BADCOUNT (-15)
2288
2289 This error is given if the value of the ovecsize argument is negative.
2290
2291 PCRE_ERROR_RECURSIONLIMIT (-21)
2292
2293 The internal recursion limit, as specified by the match_limit_recursion
2294 field in a pcre_extra structure (or defaulted) was reached. See the
2295 description above.
2296
2297 PCRE_ERROR_BADNEWLINE (-23)
2298
2299 An invalid combination of PCRE_NEWLINE_xxx options was given.
2300
2301 Error numbers -16 to -20 and -22 are not used by pcre_exec().
2302
2303
2304 EXTRACTING CAPTURED SUBSTRINGS BY NUMBER
2305
2306 int pcre_copy_substring(const char *subject, int *ovector,
2307 int stringcount, int stringnumber, char *buffer,
2308 int buffersize);
2309
2310 int pcre_get_substring(const char *subject, int *ovector,
2311 int stringcount, int stringnumber,
2312 const char **stringptr);
2313
2314 int pcre_get_substring_list(const char *subject,
2315 int *ovector, int stringcount, const char ***listptr);
2316
2317 Captured substrings can be accessed directly by using the offsets
2318 returned by pcre_exec() in ovector. For convenience, the functions
2319 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
2320 string_list() are provided for extracting captured substrings as new,
2321 separate, zero-terminated strings. These functions identify substrings
2322 by number. The next section describes functions for extracting named
2323 substrings.
2324
2325 A substring that contains a binary zero is correctly extracted and has
2326 a further zero added on the end, but the result is not, of course, a C
2327 string. However, you can process such a string by referring to the
2328 length that is returned by pcre_copy_substring() and pcre_get_sub-
2329 string(). Unfortunately, the interface to pcre_get_substring_list() is
2330 not adequate for handling strings containing binary zeros, because the
2331 end of the final string is not independently indicated.
2332
2333 The first three arguments are the same for all three of these func-
2334 tions: subject is the subject string that has just been successfully
2335 matched, ovector is a pointer to the vector of integer offsets that was
2336 passed to pcre_exec(), and stringcount is the number of substrings that
2337 were captured by the match, including the substring that matched the
2338 entire regular expression. This is the value returned by pcre_exec() if
2339 it is greater than zero. If pcre_exec() returned zero, indicating that
2340 it ran out of space in ovector, the value passed as stringcount should
2341 be the number of elements in the vector divided by three.
2342
2343 The functions pcre_copy_substring() and pcre_get_substring() extract a
2344 single substring, whose number is given as stringnumber. A value of
2345 zero extracts the substring that matched the entire pattern, whereas
2346 higher values extract the captured substrings. For pcre_copy_sub-
2347 string(), the string is placed in buffer, whose length is given by
2348 buffersize, while for pcre_get_substring() a new block of memory is
2349 obtained via pcre_malloc, and its address is returned via stringptr.
2350 The yield of the function is the length of the string, not including
2351 the terminating zero, or one of these error codes:
2352
2353 PCRE_ERROR_NOMEMORY (-6)
2354
2355 The buffer was too small for pcre_copy_substring(), or the attempt to
2356 get memory failed for pcre_get_substring().
2357
2358 PCRE_ERROR_NOSUBSTRING (-7)
2359
2360 There is no substring whose number is stringnumber.
2361
2362 The pcre_get_substring_list() function extracts all available sub-
2363 strings and builds a list of pointers to them. All this is done in a
2364 single block of memory that is obtained via pcre_malloc. The address of
2365 the memory block is returned via listptr, which is also the start of
2366 the list of string pointers. The end of the list is marked by a NULL
2367 pointer. The yield of the function is zero if all went well, or the
2368 error code
2369
2370 PCRE_ERROR_NOMEMORY (-6)
2371
2372 if the attempt to get the memory block failed.
2373
2374 When any of these functions encounter a substring that is unset, which
2375 can happen when capturing subpattern number n+1 matches some part of
2376 the subject, but subpattern n has not been used at all, they return an
2377 empty string. This can be distinguished from a genuine zero-length sub-
2378 string by inspecting the appropriate offset in ovector, which is nega-
2379 tive for unset substrings.
2380
2381 The two convenience functions pcre_free_substring() and pcre_free_sub-
2382 string_list() can be used to free the memory returned by a previous
2383 call of pcre_get_substring() or pcre_get_substring_list(), respec-
2384 tively. They do nothing more than call the function pointed to by
2385 pcre_free, which of course could be called directly from a C program.
2386 However, PCRE is used in some situations where it is linked via a spe-
2387 cial interface to another programming language that cannot use
2388 pcre_free directly; it is for these cases that the functions are pro-
2389 vided.
2390
2391
2392 EXTRACTING CAPTURED SUBSTRINGS BY NAME
2393
2394 int pcre_get_stringnumber(const pcre *code,
2395 const char *name);
2396
2397 int pcre_copy_named_substring(const pcre *code,
2398 const char *subject, int *ovector,
2399 int stringcount, const char *stringname,
2400 char *buffer, int buffersize);
2401
2402 int pcre_get_named_substring(const pcre *code,
2403 const char *subject, int *ovector,
2404 int stringcount, const char *stringname,
2405 const char **stringptr);
2406
2407 To extract a substring by name, you first have to find associated num-
2408 ber. For example, for this pattern
2409
2410 (a+)b(?<xxx>\d+)...
2411
2412 the number of the subpattern called "xxx" is 2. If the name is known to
2413 be unique (PCRE_DUPNAMES was not set), you can find the number from the
2414 name by calling pcre_get_stringnumber(). The first argument is the com-
2415 piled pattern, and the second is the name. The yield of the function is
2416 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
2417 subpattern of that name.
2418
2419 Given the number, you can extract the substring directly, or use one of
2420 the functions described in the previous section. For convenience, there
2421 are also two functions that do the whole job.
2422
2423 Most of the arguments of pcre_copy_named_substring() and
2424 pcre_get_named_substring() are the same as those for the similarly
2425 named functions that extract by number. As these are described in the
2426 previous section, they are not re-described here. There are just two
2427 differences:
2428
2429 First, instead of a substring number, a substring name is given. Sec-
2430 ond, there is an extra argument, given at the start, which is a pointer
2431 to the compiled pattern. This is needed in order to gain access to the
2432 name-to-number translation table.
2433
2434 These functions call pcre_get_stringnumber(), and if it succeeds, they
2435 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
2436 ate. NOTE: If PCRE_DUPNAMES is set and there are duplicate names, the
2437 behaviour may not be what you want (see the next section).
2438
2439 Warning: If the pattern uses the "(?|" feature to set up multiple sub-
2440 patterns with the same number, you cannot use names to distinguish
2441 them, because names are not included in the compiled code. The matching
2442 process uses only numbers.
2443
2444
2445 DUPLICATE SUBPATTERN NAMES
2446
2447 int pcre_get_stringtable_entries(const pcre *code,
2448 const char *name, char **first, char **last);
2449
2450 When a pattern is compiled with the PCRE_DUPNAMES option, names for
2451 subpatterns are not required to be unique. Normally, patterns with
2452 duplicate names are such that in any one match, only one of the named
2453 subpatterns participates. An example is shown in the pcrepattern docu-
2454 mentation.
2455
2456 When duplicates are present, pcre_copy_named_substring() and
2457 pcre_get_named_substring() return the first substring corresponding to
2458 the given name that is set. If none are set, PCRE_ERROR_NOSUBSTRING
2459 (-7) is returned; no data is returned. The pcre_get_stringnumber()
2460 function returns one of the numbers that are associated with the name,
2461 but it is not defined which it is.
2462
2463 If you want to get full details of all captured substrings for a given
2464 name, you must use the pcre_get_stringtable_entries() function. The
2465 first argument is the compiled pattern, and the second is the name. The
2466 third and fourth are pointers to variables which are updated by the
2467 function. After it has run, they point to the first and last entries in
2468 the name-to-number table for the given name. The function itself
2469 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
2470 there are none. The format of the table is described above in the sec-
2471 tion entitled Information about a pattern. Given all the relevant
2472 entries for the name, you can extract each of their numbers, and hence
2473 the captured data, if any.
2474
2475
2476 FINDING ALL POSSIBLE MATCHES
2477
2478 The traditional matching function uses a similar algorithm to Perl,
2479 which stops when it finds the first match, starting at a given point in
2480 the subject. If you want to find all possible matches, or the longest
2481 possible match, consider using the alternative matching function (see
2482 below) instead. If you cannot use the alternative function, but still
2483 need to find all possible matches, you can kludge it up by making use
2484 of the callout facility, which is described in the pcrecallout documen-
2485 tation.
2486
2487 What you have to do is to insert a callout right at the end of the pat-
2488 tern. When your callout function is called, extract and save the cur-
2489 rent matched substring. Then return 1, which forces pcre_exec() to
2490 backtrack and try other alternatives. Ultimately, when it runs out of
2491 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
2492
2493
2494 MATCHING A PATTERN: THE ALTERNATIVE FUNCTION
2495
2496 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
2497 const char *subject, int length, int startoffset,
2498 int options, int *ovector, int ovecsize,
2499 int *workspace, int wscount);
2500
2501 The function pcre_dfa_exec() is called to match a subject string
2502 against a compiled pattern, using a matching algorithm that scans the
2503 subject string just once, and does not backtrack. This has different
2504 characteristics to the normal algorithm, and is not compatible with
2505 Perl. Some of the features of PCRE patterns are not supported. Never-
2506 theless, there are times when this kind of matching can be useful. For
2507 a discussion of the two matching algorithms, and a list of features
2508 that pcre_dfa_exec() does not support, see the pcrematching documenta-
2509 tion.
2510
2511 The arguments for the pcre_dfa_exec() function are the same as for
2512 pcre_exec(), plus two extras. The ovector argument is used in a differ-
2513 ent way, and this is described below. The other common arguments are
2514 used in the same way as for pcre_exec(), so their description is not
2515 repeated here.
2516
2517 The two additional arguments provide workspace for the function. The
2518 workspace vector should contain at least 20 elements. It is used for
2519 keeping track of multiple paths through the pattern tree. More
2520 workspace will be needed for patterns and subjects where there are a
2521 lot of potential matches.
2522
2523 Here is an example of a simple call to pcre_dfa_exec():
2524
2525 int rc;
2526 int ovector[10];
2527 int wspace[20];
2528 rc = pcre_dfa_exec(
2529 re, /* result of pcre_compile() */
2530 NULL, /* we didn't study the pattern */
2531 "some string", /* the subject string */
2532 11, /* the length of the subject string */
2533 0, /* start at offset 0 in the subject */
2534 0, /* default options */
2535 ovector, /* vector of integers for substring information */
2536 10, /* number of elements (NOT size in bytes) */
2537 wspace, /* working space vector */
2538 20); /* number of elements (NOT size in bytes) */
2539
2540 Option bits for pcre_dfa_exec()
2541
2542 The unused bits of the options argument for pcre_dfa_exec() must be
2543 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
2544 LINE_xxx, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY,
2545 PCRE_NOTEMPTY_ATSTART, PCRE_NO_UTF8_CHECK, PCRE_PARTIAL_HARD, PCRE_PAR-
2546 TIAL_SOFT, PCRE_DFA_SHORTEST, and PCRE_DFA_RESTART. All but the last
2547 four of these are exactly the same as for pcre_exec(), so their
2548 description is not repeated here.
2549
2550 PCRE_PARTIAL_HARD
2551 PCRE_PARTIAL_SOFT
2552
2553 These have the same general effect as they do for pcre_exec(), but the
2554 details are slightly different. When PCRE_PARTIAL_HARD is set for
2555 pcre_dfa_exec(), it returns PCRE_ERROR_PARTIAL if the end of the sub-
2556 ject is reached and there is still at least one matching possibility
2557 that requires additional characters. This happens even if some complete
2558 matches have also been found. When PCRE_PARTIAL_SOFT is set, the return
2559 code PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end
2560 of the subject is reached, there have been no complete matches, but
2561 there is still at least one matching possibility. The portion of the
2562 string that was inspected when the longest partial match was found is
2563 set as the first matching string in both cases.
2564
2565 PCRE_DFA_SHORTEST
2566
2567 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
2568 stop as soon as it has found one match. Because of the way the alterna-
2569 tive algorithm works, this is necessarily the shortest possible match
2570 at the first possible matching point in the subject string.
2571
2572 PCRE_DFA_RESTART
2573
2574 When pcre_dfa_exec() returns a partial match, it is possible to call it
2575 again, with additional subject characters, and have it continue with
2576 the same match. The PCRE_DFA_RESTART option requests this action; when
2577 it is set, the workspace and wscount options must reference the same
2578 vector as before because data about the match so far is left in them
2579 after a partial match. There is more discussion of this facility in the
2580 pcrepartial documentation.
2581
2582 Successful returns from pcre_dfa_exec()
2583
2584 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
2585 string in the subject. Note, however, that all the matches from one run
2586 of the function start at the same point in the subject. The shorter
2587 matches are all initial substrings of the longer matches. For example,
2588 if the pattern
2589
2590 <.*>
2591
2592 is matched against the string
2593
2594 This is <something> <something else> <something further> no more
2595
2596 the three matched strings are
2597
2598 <something>
2599 <something> <something else>
2600 <something> <something else> <something further>
2601
2602 On success, the yield of the function is a number greater than zero,
2603 which is the number of matched substrings. The substrings themselves
2604 are returned in ovector. Each string uses two elements; the first is
2605 the offset to the start, and the second is the offset to the end. In
2606 fact, all the strings have the same start offset. (Space could have
2607 been saved by giving this only once, but it was decided to retain some
2608 compatibility with the way pcre_exec() returns data, even though the
2609 meaning of the strings is different.)
2610
2611 The strings are returned in reverse order of length; that is, the long-
2612 est matching string is given first. If there were too many matches to
2613 fit into ovector, the yield of the function is zero, and the vector is
2614 filled with the longest matches.
2615
2616 Error returns from pcre_dfa_exec()
2617
2618 The pcre_dfa_exec() function returns a negative number when it fails.
2619 Many of the errors are the same as for pcre_exec(), and these are
2620 described above. There are in addition the following errors that are
2621 specific to pcre_dfa_exec():
2622
2623 PCRE_ERROR_DFA_UITEM (-16)
2624
2625 This return is given if pcre_dfa_exec() encounters an item in the pat-
2626 tern that it does not support, for instance, the use of \C or a back
2627 reference.
2628
2629 PCRE_ERROR_DFA_UCOND (-17)
2630
2631 This return is given if pcre_dfa_exec() encounters a condition item
2632 that uses a back reference for the condition, or a test for recursion
2633 in a specific group. These are not supported.
2634
2635 PCRE_ERROR_DFA_UMLIMIT (-18)
2636
2637 This return is given if pcre_dfa_exec() is called with an extra block
2638 that contains a setting of the match_limit field. This is not supported
2639 (it is meaningless).
2640
2641 PCRE_ERROR_DFA_WSSIZE (-19)
2642
2643 This return is given if pcre_dfa_exec() runs out of space in the
2644 workspace vector.
2645
2646 PCRE_ERROR_DFA_RECURSE (-20)
2647
2648 When a recursive subpattern is processed, the matching function calls
2649 itself recursively, using private vectors for ovector and workspace.
2650 This error is given if the output vector is not large enough. This
2651 should be extremely rare, as a vector of size 1000 is used.
2652
2653
2654 SEE ALSO
2655
2656 pcrebuild(3), pcrecallout(3), pcrecpp(3)(3), pcrematching(3), pcrepar-
2657 tial(3), pcreposix(3), pcreprecompile(3), pcresample(3), pcrestack(3).
2658
2659
2660 AUTHOR
2661
2662 Philip Hazel
2663 University Computing Service
2664 Cambridge CB2 3QH, England.
2665
2666
2667 REVISION
2668
2669 Last updated: 11 September 2009
2670 Copyright (c) 1997-2009 University of Cambridge.
2671 ------------------------------------------------------------------------------
2672
2673
2674 PCRECALLOUT(3) PCRECALLOUT(3)
2675
2676
2677 NAME
2678 PCRE - Perl-compatible regular expressions
2679
2680
2681 PCRE CALLOUTS
2682
2683 int (*pcre_callout)(pcre_callout_block *);
2684
2685 PCRE provides a feature called "callout", which is a means of temporar-
2686 ily passing control to the caller of PCRE in the middle of pattern
2687 matching. The caller of PCRE provides an external function by putting
2688 its entry point in the global variable pcre_callout. By default, this
2689 variable contains NULL, which disables all calling out.
2690
2691 Within a regular expression, (?C) indicates the points at which the
2692 external function is to be called. Different callout points can be
2693 identified by putting a number less than 256 after the letter C. The
2694 default value is zero. For example, this pattern has two callout
2695 points:
2696
2697 (?C1)abc(?C2)def
2698
2699 If the PCRE_AUTO_CALLOUT option bit is set when pcre_compile() is
2700 called, PCRE automatically inserts callouts, all with number 255,
2701 before each item in the pattern. For example, if PCRE_AUTO_CALLOUT is
2702 used with the pattern
2703
2704 A(\d{2}|--)
2705
2706 it is processed as if it were
2707
2708 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
2709
2710 Notice that there is a callout before and after each parenthesis and
2711 alternation bar. Automatic callouts can be used for tracking the
2712 progress of pattern matching. The pcretest command has an option that
2713 sets automatic callouts; when it is used, the output indicates how the
2714 pattern is matched. This is useful information when you are trying to
2715 optimize the performance of a particular pattern.
2716
2717
2718 MISSING CALLOUTS
2719
2720 You should be aware that, because of optimizations in the way PCRE
2721 matches patterns by default, callouts sometimes do not happen. For
2722 example, if the pattern is
2723
2724 ab(?C4)cd
2725
2726 PCRE knows that any matching string must contain the letter "d". If the
2727 subject string is "abyz", the lack of "d" means that matching doesn't
2728 ever start, and the callout is never reached. However, with "abyd",
2729 though the result is still no match, the callout is obeyed.
2730
2731 You can disable these optimizations by passing the PCRE_NO_START_OPTI-
2732 MIZE option to pcre_exec() or pcre_dfa_exec(). This slows down the
2733 matching process, but does ensure that callouts such as the example
2734 above are obeyed.
2735
2736
2737 THE CALLOUT INTERFACE
2738
2739 During matching, when PCRE reaches a callout point, the external func-
2740 tion defined by pcre_callout is called (if it is set). This applies to
2741 both the pcre_exec() and the pcre_dfa_exec() matching functions. The
2742 only argument to the callout function is a pointer to a pcre_callout
2743 block. This structure contains the following fields:
2744
2745 int version;
2746 int callout_number;
2747 int *offset_vector;
2748 const char *subject;
2749 int subject_length;
2750 int start_match;
2751 int current_position;
2752 int capture_top;
2753 int capture_last;
2754 void *callout_data;
2755 int pattern_position;
2756 int next_item_length;
2757
2758 The version field is an integer containing the version number of the
2759 block format. The initial version was 0; the current version is 1. The
2760 version number will change again in future if additional fields are
2761 added, but the intention is never to remove any of the existing fields.
2762
2763 The callout_number field contains the number of the callout, as com-
2764 piled into the pattern (that is, the number after ?C for manual call-
2765 outs, and 255 for automatically generated callouts).
2766
2767 The offset_vector field is a pointer to the vector of offsets that was
2768 passed by the caller to pcre_exec() or pcre_dfa_exec(). When
2769 pcre_exec() is used, the contents can be inspected in order to extract
2770 substrings that have been matched so far, in the same way as for
2771 extracting substrings after a match has completed. For pcre_dfa_exec()
2772 this field is not useful.
2773
2774 The subject and subject_length fields contain copies of the values that
2775 were passed to pcre_exec().
2776
2777 The start_match field normally contains the offset within the subject
2778 at which the current match attempt started. However, if the escape
2779 sequence \K has been encountered, this value is changed to reflect the
2780 modified starting point. If the pattern is not anchored, the callout
2781 function may be called several times from the same point in the pattern
2782 for different starting points in the subject.
2783
2784 The current_position field contains the offset within the subject of
2785 the current match pointer.
2786
2787 When the pcre_exec() function is used, the capture_top field contains
2788 one more than the number of the highest numbered captured substring so
2789 far. If no substrings have been captured, the value of capture_top is
2790 one. This is always the case when pcre_dfa_exec() is used, because it
2791 does not support captured substrings.
2792
2793 The capture_last field contains the number of the most recently cap-
2794 tured substring. If no substrings have been captured, its value is -1.
2795 This is always the case when pcre_dfa_exec() is used.
2796
2797 The callout_data field contains a value that is passed to pcre_exec()
2798 or pcre_dfa_exec() specifically so that it can be passed back in call-
2799 outs. It is passed in the pcre_callout field of the pcre_extra data
2800 structure. If no such data was passed, the value of callout_data in a
2801 pcre_callout block is NULL. There is a description of the pcre_extra
2802 structure in the pcreapi documentation.
2803
2804 The pattern_position field is present from version 1 of the pcre_call-
2805 out structure. It contains the offset to the next item to be matched in
2806 the pattern string.
2807
2808 The next_item_length field is present from version 1 of the pcre_call-
2809 out structure. It contains the length of the next item to be matched in
2810 the pattern string. When the callout immediately precedes an alterna-
2811 tion bar, a closing parenthesis, or the end of the pattern, the length
2812 is zero. When the callout precedes an opening parenthesis, the length
2813 is that of the entire subpattern.
2814
2815 The pattern_position and next_item_length fields are intended to help
2816 in distinguishing between different automatic callouts, which all have
2817 the same callout number. However, they are set for all callouts.
2818
2819
2820 RETURN VALUES
2821
2822 The external callout function returns an integer to PCRE. If the value
2823 is zero, matching proceeds as normal. If the value is greater than
2824 zero, matching fails at the current point, but the testing of other
2825 matching possibilities goes ahead, just as if a lookahead assertion had
2826 failed. If the value is less than zero, the match is abandoned, and
2827 pcre_exec() (or pcre_dfa_exec()) returns the negative value.
2828
2829 Negative values should normally be chosen from the set of
2830 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
2831 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
2832 reserved for use by callout functions; it will never be used by PCRE
2833 itself.
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: 15 March 2009
2846 Copyright (c) 1997-2009 University of Cambridge.
2847 ------------------------------------------------------------------------------
2848
2849
2850 PCRECOMPAT(3) PCRECOMPAT(3)
2851
2852
2853 NAME
2854 PCRE - Perl-compatible regular expressions
2855
2856
2857 DIFFERENCES BETWEEN PCRE AND PERL
2858
2859 This document describes the differences in the ways that PCRE and Perl
2860 handle regular expressions. The differences described here are mainly
2861 with respect to Perl 5.8, though PCRE versions 7.0 and later contain
2862 some features that are in Perl 5.10.
2863
2864 1. PCRE has only a subset of Perl's UTF-8 and Unicode support. Details
2865 of what it does have are given in the section on UTF-8 support in the
2866 main pcre page.
2867
2868 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl
2869 permits them, but they do not mean what you might think. For example,
2870 (?!a){3} does not assert that the next three characters are not "a". It
2871 just asserts that the next character is not "a" three times.
2872
2873 3. Capturing subpatterns that occur inside negative lookahead asser-
2874 tions are counted, but their entries in the offsets vector are never
2875 set. Perl sets its numerical variables from any such patterns that are
2876 matched before the assertion fails to match something (thereby succeed-
2877 ing), but only if the negative lookahead assertion contains just one
2878 branch.
2879
2880 4. Though binary zero characters are supported in the subject string,
2881 they are not allowed in a pattern string because it is passed as a nor-
2882 mal C string, terminated by zero. The escape sequence \0 can be used in
2883 the pattern to represent a binary zero.
2884
2885 5. The following Perl escape sequences are not supported: \l, \u, \L,
2886 \U, and \N. In fact these are implemented by Perl's general string-han-
2887 dling and are not part of its pattern matching engine. If any of these
2888 are encountered by PCRE, an error is generated.
2889
2890 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
2891 is built with Unicode character property support. The properties that
2892 can be tested with \p and \P are limited to the general category prop-
2893 erties such as Lu and Nd, script names such as Greek or Han, and the
2894 derived properties Any and L&. PCRE does support the Cs (surrogate)
2895 property, which Perl does not; the Perl documentation says "Because
2896 Perl hides the need for the user to understand the internal representa-
2897 tion of Unicode characters, there is no need to implement the somewhat
2898 messy concept of surrogates."
2899
2900 7. PCRE does support the \Q...\E escape for quoting substrings. Charac-
2901 ters in between are treated as literals. This is slightly different
2902 from Perl in that $ and @ are also handled as literals inside the
2903 quotes. In Perl, they cause variable interpolation (but of course PCRE
2904 does not have variables). Note the following examples:
2905
2906 Pattern PCRE matches Perl matches
2907
2908 \Qabc$xyz\E abc$xyz abc followed by the
2909 contents of $xyz
2910 \Qabc\$xyz\E abc\$xyz abc\$xyz
2911 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2912
2913 The \Q...\E sequence is recognized both inside and outside character
2914 classes.
2915
2916 8. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
2917 constructions. However, there is support for recursive patterns. This
2918 is not available in Perl 5.8, but it is in Perl 5.10. Also, the PCRE
2919 "callout" feature allows an external function to be called during pat-
2920 tern matching. See the pcrecallout documentation for details.
2921
2922 9. Subpatterns that are called recursively or as "subroutines" are
2923 always treated as atomic groups in PCRE. This is like Python, but
2924 unlike Perl. There is a discussion of an example that explains this in
2925 more detail in the section on recursion differences from Perl in the
2926 pcrecompat page.
2927
2928 10. There are some differences that are concerned with the settings of
2929 captured strings when part of a pattern is repeated. For example,
2930 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
2931 unset, but in PCRE it is set to "b".
2932
2933 11. PCRE does support Perl 5.10's backtracking verbs (*ACCEPT),
2934 (*FAIL), (*F), (*COMMIT), (*PRUNE), (*SKIP), and (*THEN), but only in
2935 the forms without an argument. PCRE does not support (*MARK).
2936
2937 12. PCRE provides some extensions to the Perl regular expression facil-
2938 ities. Perl 5.10 will include new features that are not in earlier
2939 versions, some of which (such as named parentheses) have been in PCRE
2940 for some time. This list is with respect to Perl 5.10:
2941
2942 (a) Although lookbehind assertions must match fixed length strings,
2943 each alternative branch of a lookbehind assertion can match a different
2944 length of string. Perl requires them all to have the same length.
2945
2946 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
2947 meta-character matches only at the very end of the string.
2948
2949 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
2950 cial meaning is faulted. Otherwise, like Perl, the backslash is quietly
2951 ignored. (Perl can be made to issue a warning.)
2952
2953 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
2954 fiers is inverted, that is, by default they are not greedy, but if fol-
2955 lowed by a question mark they are.
2956
2957 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
2958 tried only at the first matching position in the subject string.
2959
2960 (f) The PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NOTEMPTY_ATSTART,
2961 and PCRE_NO_AUTO_CAPTURE options for pcre_exec() have no Perl equiva-
2962 lents.
2963
2964 (g) The \R escape sequence can be restricted to match only CR, LF, or
2965 CRLF by the PCRE_BSR_ANYCRLF option.
2966
2967 (h) The callout facility is PCRE-specific.
2968
2969 (i) The partial matching facility is PCRE-specific.
2970
2971 (j) Patterns compiled by PCRE can be saved and re-used at a later time,
2972 even on different hosts that have the other endianness.
2973
2974 (k) The alternative matching function (pcre_dfa_exec()) matches in a
2975 different way and is not Perl-compatible.
2976
2977 (l) PCRE recognizes some special sequences such as (*CR) at the start
2978 of a pattern that set overall options that cannot be changed within the
2979 pattern.
2980
2981
2982 AUTHOR
2983
2984 Philip Hazel
2985 University Computing Service
2986 Cambridge CB2 3QH, England.
2987
2988
2989 REVISION
2990
2991 Last updated: 18 September 2009
2992 Copyright (c) 1997-2009 University of Cambridge.
2993 ------------------------------------------------------------------------------
2994
2995
2996 PCREPATTERN(3) PCREPATTERN(3)
2997
2998
2999 NAME
3000 PCRE - Perl-compatible regular expressions
3001
3002
3003 PCRE REGULAR EXPRESSION DETAILS
3004
3005 The syntax and semantics of the regular expressions that are supported
3006 by PCRE are described in detail below. There is a quick-reference syn-
3007 tax summary in the pcresyntax page. PCRE tries to match Perl syntax and
3008 semantics as closely as it can. PCRE also supports some alternative
3009 regular expression syntax (which does not conflict with the Perl syn-
3010 tax) in order to provide some compatibility with regular expressions in
3011 Python, .NET, and Oniguruma.
3012
3013 Perl's regular expressions are described in its own documentation, and
3014 regular expressions in general are covered in a number of books, some
3015 of which have copious examples. Jeffrey Friedl's "Mastering Regular
3016 Expressions", published by O'Reilly, covers regular expressions in
3017 great detail. This description of PCRE's regular expressions is
3018 intended as reference material.
3019
3020 The original operation of PCRE was on strings of one-byte characters.
3021 However, there is now also support for UTF-8 character strings. To use
3022 this, you must build PCRE to include UTF-8 support, and then call
3023 pcre_compile() with the PCRE_UTF8 option. There is also a special
3024 sequence that can be given at the start of a pattern:
3025
3026 (*UTF8)
3027
3028 Starting a pattern with this sequence is equivalent to setting the
3029 PCRE_UTF8 option. This feature is not Perl-compatible. How setting
3030 UTF-8 mode affects pattern matching is mentioned in several places
3031 below. There is also a summary of UTF-8 features in the section on
3032 UTF-8 support in the main pcre page.
3033
3034 The remainder of this document discusses the patterns that are sup-
3035 ported by PCRE when its main matching function, pcre_exec(), is used.
3036 From release 6.0, PCRE offers a second matching function,
3037 pcre_dfa_exec(), which matches using a different algorithm that is not
3038 Perl-compatible. Some of the features discussed below are not available
3039 when pcre_dfa_exec() is used. The advantages and disadvantages of the
3040 alternative function, and how it differs from the normal function, are
3041 discussed in the pcrematching page.
3042
3043
3044 NEWLINE CONVENTIONS
3045
3046 PCRE supports five different conventions for indicating line breaks in
3047 strings: a single CR (carriage return) character, a single LF (line-
3048 feed) character, the two-character sequence CRLF, any of the three pre-
3049 ceding, or any Unicode newline sequence. The pcreapi page has further
3050 discussion about newlines, and shows how to set the newline convention
3051 in the options arguments for the compiling and matching functions.
3052
3053 It is also possible to specify a newline convention by starting a pat-
3054 tern string with one of the following five sequences:
3055
3056 (*CR) carriage return
3057 (*LF) linefeed
3058 (*CRLF) carriage return, followed by linefeed
3059 (*ANYCRLF) any of the three above
3060 (*ANY) all Unicode newline sequences
3061
3062 These override the default and the options given to pcre_compile(). For
3063 example, on a Unix system where LF is the default newline sequence, the
3064 pattern
3065
3066 (*CR)a.b
3067
3068 changes the convention to CR. That pattern matches "a\nb" because LF is
3069 no longer a newline. Note that these special settings, which are not
3070 Perl-compatible, are recognized only at the very start of a pattern,
3071 and that they must be in upper case. If more than one of them is
3072 present, the last one is used.
3073
3074 The newline convention does not affect what the \R escape sequence
3075 matches. By default, this is any Unicode newline sequence, for Perl
3076 compatibility. However, this can be changed; see the description of \R
3077 in the section entitled "Newline sequences" below. A change of \R set-
3078 ting can be combined with a change of newline convention.
3079
3080
3081 CHARACTERS AND METACHARACTERS
3082
3083 A regular expression is a pattern that is matched against a subject
3084 string from left to right. Most characters stand for themselves in a
3085 pattern, and match the corresponding characters in the subject. As a
3086 trivial example, the pattern
3087
3088 The quick brown fox
3089
3090 matches a portion of a subject string that is identical to itself. When
3091 caseless matching is specified (the PCRE_CASELESS option), letters are
3092 matched independently of case. In UTF-8 mode, PCRE always understands
3093 the concept of case for characters whose values are less than 128, so
3094 caseless matching is always possible. For characters with higher val-
3095 ues, the concept of case is supported if PCRE is compiled with Unicode
3096 property support, but not otherwise. If you want to use caseless
3097 matching for characters 128 and above, you must ensure that PCRE is
3098 compiled with Unicode property support as well as with UTF-8 support.
3099
3100 The power of regular expressions comes from the ability to include
3101 alternatives and repetitions in the pattern. These are encoded in the
3102 pattern by the use of metacharacters, which do not stand for themselves
3103 but instead are interpreted in some special way.
3104
3105 There are two different sets of metacharacters: those that are recog-
3106 nized anywhere in the pattern except within square brackets, and those
3107 that are recognized within square brackets. Outside square brackets,
3108 the metacharacters are as follows:
3109
3110 \ general escape character with several uses
3111 ^ assert start of string (or line, in multiline mode)
3112 $ assert end of string (or line, in multiline mode)
3113 . match any character except newline (by default)
3114 [ start character class definition
3115 | start of alternative branch
3116 ( start subpattern
3117 ) end subpattern
3118 ? extends the meaning of (
3119 also 0 or 1 quantifier
3120 also quantifier minimizer
3121 * 0 or more quantifier
3122 + 1 or more quantifier
3123 also "possessive quantifier"
3124 { start min/max quantifier
3125
3126 Part of a pattern that is in square brackets is called a "character
3127 class". In a character class the only metacharacters are:
3128
3129 \ general escape character
3130 ^ negate the class, but only if the first character
3131 - indicates character range
3132 [ POSIX character class (only if followed by POSIX
3133 syntax)
3134 ] terminates the character class
3135
3136 The following sections describe the use of each of the metacharacters.
3137
3138
3139 BACKSLASH
3140
3141 The backslash character has several uses. Firstly, if it is followed by
3142 a non-alphanumeric character, it takes away any special meaning that
3143 character may have. This use of backslash as an escape character
3144 applies both inside and outside character classes.
3145
3146 For example, if you want to match a * character, you write \* in the
3147 pattern. This escaping action applies whether or not the following
3148 character would otherwise be interpreted as a metacharacter, so it is
3149 always safe to precede a non-alphanumeric with backslash to specify
3150 that it stands for itself. In particular, if you want to match a back-
3151 slash, you write \\.
3152
3153 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
3154 the pattern (other than in a character class) and characters between a
3155 # outside a character class and the next newline are ignored. An escap-
3156 ing backslash can be used to include a whitespace or # character as
3157 part of the pattern.
3158
3159 If you want to remove the special meaning from a sequence of charac-
3160 ters, you can do so by putting them between \Q and \E. This is differ-
3161 ent from Perl in that $ and @ are handled as literals in \Q...\E
3162 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
3163 tion. Note the following examples:
3164
3165 Pattern PCRE matches Perl matches
3166
3167 \Qabc$xyz\E abc$xyz abc followed by the
3168 contents of $xyz
3169 \Qabc\$xyz\E abc\$xyz abc\$xyz
3170 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3171
3172 The \Q...\E sequence is recognized both inside and outside character
3173 classes.
3174
3175 Non-printing characters
3176
3177 A second use of backslash provides a way of encoding non-printing char-
3178 acters in patterns in a visible manner. There is no restriction on the
3179 appearance of non-printing characters, apart from the binary zero that
3180 terminates a pattern, but when a pattern is being prepared by text
3181 editing, it is usually easier to use one of the following escape
3182 sequences than the binary character it represents:
3183
3184 \a alarm, that is, the BEL character (hex 07)
3185 \cx "control-x", where x is any character
3186 \e escape (hex 1B)
3187 \f formfeed (hex 0C)
3188 \n linefeed (hex 0A)
3189 \r carriage return (hex 0D)
3190 \t tab (hex 09)
3191 \ddd character with octal code ddd, or backreference
3192 \xhh character with hex code hh
3193 \x{hhh..} character with hex code hhh..
3194
3195 The precise effect of \cx is as follows: if x is a lower case letter,
3196 it is converted to upper case. Then bit 6 of the character (hex 40) is
3197 inverted. Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c;
3198 becomes hex 7B.
3199
3200 After \x, from zero to two hexadecimal digits are read (letters can be
3201 in upper or lower case). Any number of hexadecimal digits may appear
3202 between \x{ and }, but the value of the character code must be less
3203 than 256 in non-UTF-8 mode, and less than 2**31 in UTF-8 mode. That is,
3204 the maximum value in hexadecimal is 7FFFFFFF. Note that this is bigger
3205 than the largest Unicode code point, which is 10FFFF.
3206
3207 If characters other than hexadecimal digits appear between \x{ and },
3208 or if there is no terminating }, this form of escape is not recognized.
3209 Instead, the initial \x will be interpreted as a basic hexadecimal
3210 escape, with no following digits, giving a character whose value is
3211 zero.
3212
3213 Characters whose value is less than 256 can be defined by either of the
3214 two syntaxes for \x. There is no difference in the way they are han-
3215 dled. For example, \xdc is exactly the same as \x{dc}.
3216
3217 After \0 up to two further octal digits are read. If there are fewer
3218 than two digits, just those that are present are used. Thus the
3219 sequence \0\x\07 specifies two binary zeros followed by a BEL character
3220 (code value 7). Make sure you supply two digits after the initial zero
3221 if the pattern character that follows is itself an octal digit.
3222
3223 The handling of a backslash followed by a digit other than 0 is compli-
3224 cated. Outside a character class, PCRE reads it and any following dig-
3225 its as a decimal number. If the number is less than 10, or if there
3226 have been at least that many previous capturing left parentheses in the
3227 expression, the entire sequence is taken as a back reference. A
3228 description of how this works is given later, following the discussion
3229 of parenthesized subpatterns.
3230
3231 Inside a character class, or if the decimal number is greater than 9
3232 and there have not been that many capturing subpatterns, PCRE re-reads
3233 up to three octal digits following the backslash, and uses them to gen-
3234 erate a data character. Any subsequent digits stand for themselves. In
3235 non-UTF-8 mode, the value of a character specified in octal must be
3236 less than \400. In UTF-8 mode, values up to \777 are permitted. For
3237 example:
3238
3239 \040 is another way of writing a space
3240 \40 is the same, provided there are fewer than 40
3241 previous capturing subpatterns
3242 \7 is always a back reference
3243 \11 might be a back reference, or another way of
3244 writing a tab
3245 \011 is always a tab
3246 \0113 is a tab followed by the character "3"
3247 \113 might be a back reference, otherwise the
3248 character with octal code 113
3249 \377 might be a back reference, otherwise
3250 the byte consisting entirely of 1 bits
3251 \81 is either a back reference, or a binary zero
3252 followed by the two characters "8" and "1"
3253
3254 Note that octal values of 100 or greater must not be introduced by a
3255 leading zero, because no more than three octal digits are ever read.
3256
3257 All the sequences that define a single character value can be used both
3258 inside and outside character classes. In addition, inside a character
3259 class, the sequence \b is interpreted as the backspace character (hex
3260 08), and the sequences \R and \X are interpreted as the characters "R"
3261 and "X", respectively. Outside a character class, these sequences have
3262 different meanings (see below).
3263
3264 Absolute and relative back references
3265
3266 The sequence \g followed by an unsigned or a negative number, option-
3267 ally enclosed in braces, is an absolute or relative back reference. A
3268 named back reference can be coded as \g{name}. Back references are dis-
3269 cussed later, following the discussion of parenthesized subpatterns.
3270
3271 Absolute and relative subroutine calls
3272
3273 For compatibility with Oniguruma, the non-Perl syntax \g followed by a
3274 name or a number enclosed either in angle brackets or single quotes, is
3275 an alternative syntax for referencing a subpattern as a "subroutine".
3276 Details are discussed later. Note that \g{...} (Perl syntax) and
3277 \g<...> (Oniguruma syntax) are not synonymous. The former is a back
3278 reference; the latter is a subroutine call.
3279
3280 Generic character types
3281
3282 Another use of backslash is for specifying generic character types. The
3283 following are always recognized:
3284
3285 \d any decimal digit
3286 \D any character that is not a decimal digit
3287 \h any horizontal whitespace character
3288 \H any character that is not a horizontal whitespace character
3289 \s any whitespace character
3290 \S any character that is not a whitespace character
3291 \v any vertical whitespace character
3292 \V any character that is not a vertical whitespace character
3293 \w any "word" character
3294 \W any "non-word" character
3295
3296 Each pair of escape sequences partitions the complete set of characters
3297 into two disjoint sets. Any given character matches one, and only one,
3298 of each pair.
3299
3300 These character type sequences can appear both inside and outside char-
3301 acter classes. They each match one character of the appropriate type.
3302 If the current matching point is at the end of the subject string, all
3303 of them fail, since there is no character to match.
3304
3305 For compatibility with Perl, \s does not match the VT character (code
3306 11). This makes it different from the the POSIX "space" class. The \s
3307 characters are HT (9), LF (10), FF (12), CR (13), and space (32). If
3308 "use locale;" is included in a Perl script, \s may match the VT charac-
3309 ter. In PCRE, it never does.
3310
3311 In UTF-8 mode, characters with values greater than 128 never match \d,
3312 \s, or \w, and always match \D, \S, and \W. This is true even when Uni-
3313 code character property support is available. These sequences retain
3314 their original meanings from before UTF-8 support was available, mainly
3315 for efficiency reasons. Note that this also affects \b, because it is
3316 defined in terms of \w and \W.
3317
3318 The sequences \h, \H, \v, and \V are Perl 5.10 features. In contrast to
3319 the other sequences, these do match certain high-valued codepoints in
3320 UTF-8 mode. The horizontal space characters are:
3321
3322 U+0009 Horizontal tab
3323 U+0020 Space
3324 U+00A0 Non-break space
3325 U+1680 Ogham space mark
3326 U+180E Mongolian vowel separator
3327 U+2000 En quad
3328 U+2001 Em quad
3329 U+2002 En space
3330 U+2003 Em space
3331 U+2004 Three-per-em space
3332 U+2005 Four-per-em space
3333 U+2006 Six-per-em space
3334 U+2007 Figure space
3335 U+2008 Punctuation space
3336 U+2009 Thin space
3337 U+200A Hair space
3338 U+202F Narrow no-break space
3339 U+205F Medium mathematical space
3340 U+3000 Ideographic space
3341
3342 The vertical space characters are:
3343
3344 U+000A Linefeed
3345 U+000B Vertical tab
3346 U+000C Formfeed
3347 U+000D Carriage return
3348 U+0085 Next line
3349 U+2028 Line separator
3350 U+2029 Paragraph separator
3351
3352 A "word" character is an underscore or any character less than 256 that
3353 is a letter or digit. The definition of letters and digits is con-
3354 trolled by PCRE's low-valued character tables, and may vary if locale-
3355 specific matching is taking place (see "Locale support" in the pcreapi
3356 page). For example, in a French locale such as "fr_FR" in Unix-like
3357 systems, or "french" in Windows, some character codes greater than 128
3358 are used for accented letters, and these are matched by \w. The use of
3359 locales with Unicode is discouraged.
3360
3361 Newline sequences
3362
3363 Outside a character class, by default, the escape sequence \R matches
3364 any Unicode newline sequence. This is a Perl 5.10 feature. In non-UTF-8
3365 mode \R is equivalent to the following:
3366
3367 (?>\r\n|\n|\x0b|\f|\r|\x85)
3368
3369 This is an example of an "atomic group", details of which are given
3370 below. This particular group matches either the two-character sequence
3371 CR followed by LF, or one of the single characters LF (linefeed,
3372 U+000A), VT (vertical tab, U+000B), FF (formfeed, U+000C), CR (carriage
3373 return, U+000D), or NEL (next line, U+0085). The two-character sequence
3374 is treated as a single unit that cannot be split.
3375
3376 In UTF-8 mode, two additional characters whose codepoints are greater
3377 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
3378 rator, U+2029). Unicode character property support is not needed for
3379 these characters to be recognized.
3380
3381 It is possible to restrict \R to match only CR, LF, or CRLF (instead of
3382 the complete set of Unicode line endings) by setting the option
3383 PCRE_BSR_ANYCRLF either at compile time or when the pattern is matched.
3384 (BSR is an abbrevation for "backslash R".) This can be made the default
3385 when PCRE is built; if this is the case, the other behaviour can be
3386 requested via the PCRE_BSR_UNICODE option. It is also possible to
3387 specify these settings by starting a pattern string with one of the
3388 following sequences:
3389
3390 (*BSR_ANYCRLF) CR, LF, or CRLF only
3391 (*BSR_UNICODE) any Unicode newline sequence
3392
3393 These override the default and the options given to pcre_compile(), but
3394 they can be overridden by options given to pcre_exec(). Note that these
3395 special settings, which are not Perl-compatible, are recognized only at
3396 the very start of a pattern, and that they must be in upper case. If
3397 more than one of them is present, the last one is used. They can be
3398 combined with a change of newline convention, for example, a pattern
3399 can start with:
3400
3401 (*ANY)(*BSR_ANYCRLF)
3402
3403 Inside a character class, \R matches the letter "R".
3404
3405 Unicode character properties
3406
3407 When PCRE is built with Unicode character property support, three addi-
3408 tional escape sequences that match characters with specific properties
3409 are available. When not in UTF-8 mode, these sequences are of course
3410 limited to testing characters whose codepoints are less than 256, but
3411 they do work in this mode. The extra escape sequences are:
3412
3413 \p{xx} a character with the xx property
3414 \P{xx} a character without the xx property
3415 \X an extended Unicode sequence
3416
3417 The property names represented by xx above are limited to the Unicode
3418 script names, the general category properties, and "Any", which matches
3419 any character (including newline). Other properties such as "InMusical-
3420 Symbols" are not currently supported by PCRE. Note that \P{Any} does
3421 not match any characters, so always causes a match failure.
3422
3423 Sets of Unicode characters are defined as belonging to certain scripts.
3424 A character from one of these sets can be matched using a script name.
3425 For example:
3426
3427 \p{Greek}
3428 \P{Han}
3429
3430 Those that are not part of an identified script are lumped together as
3431 "Common". The current list of scripts is:
3432
3433 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
3434 Buhid, Canadian_Aboriginal, Cherokee, Common, Coptic, Cuneiform,
3435 Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian, Glagolitic,
3436 Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
3437 gana, Inherited, Kannada, Katakana, Kharoshthi, Khmer, Lao, Latin,
3438 Limbu, Linear_B, Malayalam, Mongolian, Myanmar, New_Tai_Lue, Nko,
3439 Ogham, Old_Italic, Old_Persian, Oriya, Osmanya, Phags_Pa, Phoenician,
3440 Runic, Shavian, Sinhala, Syloti_Nagri, Syriac, Tagalog, Tagbanwa,
3441 Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Yi.
3442
3443 Each character has exactly one general category property, specified by
3444 a two-letter abbreviation. For compatibility with Perl, negation can be
3445 specified by including a circumflex between the opening brace and the
3446 property name. For example, \p{^Lu} is the same as \P{Lu}.
3447
3448 If only one letter is specified with \p or \P, it includes all the gen-
3449 eral category properties that start with that letter. In this case, in
3450 the absence of negation, the curly brackets in the escape sequence are
3451 optional; these two examples have the same effect:
3452
3453 \p{L}
3454 \pL
3455
3456 The following general category property codes are supported:
3457
3458 C Other
3459 Cc Control
3460 Cf Format
3461 Cn Unassigned
3462 Co Private use
3463 Cs Surrogate
3464
3465 L Letter
3466 Ll Lower case letter
3467 Lm Modifier letter
3468 Lo Other letter
3469 Lt Title case letter
3470 Lu Upper case letter
3471
3472 M Mark
3473 Mc Spacing mark
3474 Me Enclosing mark
3475 Mn Non-spacing mark
3476
3477 N Number
3478 Nd Decimal number
3479 Nl Letter number
3480 No Other number
3481
3482 P Punctuation
3483 Pc Connector punctuation
3484 Pd Dash punctuation
3485 Pe Close punctuation
3486 Pf Final punctuation
3487 Pi Initial punctuation
3488 Po Other punctuation
3489 Ps Open punctuation
3490
3491 S Symbol
3492 Sc Currency symbol
3493 Sk Modifier symbol
3494 Sm Mathematical symbol
3495 So Other symbol
3496
3497 Z Separator
3498 Zl Line separator
3499 Zp Paragraph separator
3500 Zs Space separator
3501
3502 The special property L& is also supported: it matches a character that
3503 has the Lu, Ll, or Lt property, in other words, a letter that is not
3504 classified as a modifier or "other".
3505
3506 The Cs (Surrogate) property applies only to characters in the range
3507 U+D800 to U+DFFF. Such characters are not valid in UTF-8 strings (see
3508 RFC 3629) and so cannot be tested by PCRE, unless UTF-8 validity check-
3509 ing has been turned off (see the discussion of PCRE_NO_UTF8_CHECK in
3510 the pcreapi page). Perl does not support the Cs property.
3511
3512 The long synonyms for property names that Perl supports (such as
3513 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
3514 any of these properties with "Is".
3515
3516 No character that is in the Unicode table has the Cn (unassigned) prop-
3517 erty. Instead, this property is assumed for any code point that is not
3518 in the Unicode table.
3519
3520 Specifying caseless matching does not affect these escape sequences.
3521 For example, \p{Lu} always matches only upper case letters.
3522
3523 The \X escape matches any number of Unicode characters that form an
3524 extended Unicode sequence. \X is equivalent to
3525
3526 (?>\PM\pM*)
3527
3528 That is, it matches a character without the "mark" property, followed
3529 by zero or more characters with the "mark" property, and treats the
3530 sequence as an atomic group (see below). Characters with the "mark"
3531 property are typically accents that affect the preceding character.
3532 None of them have codepoints less than 256, so in non-UTF-8 mode \X
3533 matches any one character.
3534
3535 Matching characters by Unicode property is not fast, because PCRE has
3536 to search a structure that contains data for over fifteen thousand
3537 characters. That is why the traditional escape sequences such as \d and
3538 \w do not use Unicode properties in PCRE.
3539
3540 Resetting the match start
3541
3542 The escape sequence \K, which is a Perl 5.10 feature, causes any previ-
3543 ously matched characters not to be included in the final matched
3544 sequence. For example, the pattern:
3545
3546 foo\Kbar
3547
3548 matches "foobar", but reports that it has matched "bar". This feature
3549 is similar to a lookbehind assertion (described below). However, in
3550 this case, the part of the subject before the real match does not have
3551 to be of fixed length, as lookbehind assertions do. The use of \K does
3552 not interfere with the setting of captured substrings. For example,
3553 when the pattern
3554
3555 (foo)\Kbar
3556
3557 matches "foobar", the first substring is still set to "foo".
3558
3559 Simple assertions
3560
3561 The final use of backslash is for certain simple assertions. An asser-
3562 tion specifies a condition that has to be met at a particular point in
3563 a match, without consuming any characters from the subject string. The
3564 use of subpatterns for more complicated assertions is described below.
3565 The backslashed assertions are:
3566
3567 \b matches at a word boundary
3568 \B matches when not at a word boundary
3569 \A matches at the start of the subject
3570 \Z matches at the end of the subject
3571 also matches before a newline at the end of the subject
3572 \z matches only at the end of the subject
3573 \G matches at the first matching position in the subject
3574
3575 These assertions may not appear in character classes (but note that \b
3576 has a different meaning, namely the backspace character, inside a char-
3577 acter class).
3578
3579 A word boundary is a position in the subject string where the current
3580 character and the previous character do not both match \w or \W (i.e.
3581 one matches \w and the other matches \W), or the start or end of the
3582 string if the first or last character matches \w, respectively.
3583
3584 The \A, \Z, and \z assertions differ from the traditional circumflex
3585 and dollar (described in the next section) in that they only ever match
3586 at the very start and end of the subject string, whatever options are
3587 set. Thus, they are independent of multiline mode. These three asser-
3588 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
3589 affect only the behaviour of the circumflex and dollar metacharacters.
3590 However, if the startoffset argument of pcre_exec() is non-zero, indi-
3591 cating that matching is to start at a point other than the beginning of
3592 the subject, \A can never match. The difference between \Z and \z is
3593 that \Z matches before a newline at the end of the string as well as at
3594 the very end, whereas \z matches only at the end.
3595
3596 The \G assertion is true only when the current matching position is at
3597 the start point of the match, as specified by the startoffset argument
3598 of pcre_exec(). It differs from \A when the value of startoffset is
3599 non-zero. By calling pcre_exec() multiple times with appropriate argu-
3600 ments, you can mimic Perl's /g option, and it is in this kind of imple-
3601 mentation where \G can be useful.
3602
3603 Note, however, that PCRE's interpretation of \G, as the start of the
3604 current match, is subtly different from Perl's, which defines it as the
3605 end of the previous match. In Perl, these can be different when the
3606 previously matched string was empty. Because PCRE does just one match
3607 at a time, it cannot reproduce this behaviour.
3608
3609 If all the alternatives of a pattern begin with \G, the expression is
3610 anchored to the starting match position, and the "anchored" flag is set
3611 in the compiled regular expression.
3612
3613
3614 CIRCUMFLEX AND DOLLAR
3615
3616 Outside a character class, in the default matching mode, the circumflex
3617 character is an assertion that is true only if the current matching
3618 point is at the start of the subject string. If the startoffset argu-
3619 ment of pcre_exec() is non-zero, circumflex can never match if the
3620 PCRE_MULTILINE option is unset. Inside a character class, circumflex
3621 has an entirely different meaning (see below).
3622
3623 Circumflex need not be the first character of the pattern if a number
3624 of alternatives are involved, but it should be the first thing in each
3625 alternative in which it appears if the pattern is ever to match that
3626 branch. If all possible alternatives start with a circumflex, that is,
3627 if the pattern is constrained to match only at the start of the sub-
3628 ject, it is said to be an "anchored" pattern. (There are also other
3629 constructs that can cause a pattern to be anchored.)
3630
3631 A dollar character is an assertion that is true only if the current
3632 matching point is at the end of the subject string, or immediately
3633 before a newline at the end of the string (by default). Dollar need not
3634 be the last character of the pattern if a number of alternatives are
3635 involved, but it should be the last item in any branch in which it
3636 appears. Dollar has no special meaning in a character class.
3637
3638 The meaning of dollar can be changed so that it matches only at the
3639 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
3640 compile time. This does not affect the \Z assertion.
3641
3642 The meanings of the circumflex and dollar characters are changed if the
3643 PCRE_MULTILINE option is set. When this is the case, a circumflex
3644 matches immediately after internal newlines as well as at the start of
3645 the subject string. It does not match after a newline that ends the
3646 string. A dollar matches before any newlines in the string, as well as
3647 at the very end, when PCRE_MULTILINE is set. When newline is specified
3648 as the two-character sequence CRLF, isolated CR and LF characters do
3649 not indicate newlines.
3650
3651 For example, the pattern /^abc$/ matches the subject string "def\nabc"
3652 (where \n represents a newline) in multiline mode, but not otherwise.
3653 Consequently, patterns that are anchored in single line mode because
3654 all branches start with ^ are not anchored in multiline mode, and a
3655 match for circumflex is possible when the startoffset argument of
3656 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
3657 PCRE_MULTILINE is set.
3658
3659 Note that the sequences \A, \Z, and \z can be used to match the start
3660 and end of the subject in both modes, and if all branches of a pattern
3661 start with \A it is always anchored, whether or not PCRE_MULTILINE is
3662 set.
3663
3664
3665 FULL STOP (PERIOD, DOT)
3666
3667 Outside a character class, a dot in the pattern matches any one charac-
3668 ter in the subject string except (by default) a character that signi-
3669 fies the end of a line. In UTF-8 mode, the matched character may be
3670 more than one byte long.
3671
3672 When a line ending is defined as a single character, dot never matches
3673 that character; when the two-character sequence CRLF is used, dot does
3674 not match CR if it is immediately followed by LF, but otherwise it
3675 matches all characters (including isolated CRs and LFs). When any Uni-
3676 code line endings are being recognized, dot does not match CR or LF or
3677 any of the other line ending characters.
3678
3679 The behaviour of dot with regard to newlines can be changed. If the
3680 PCRE_DOTALL option is set, a dot matches any one character, without
3681 exception. If the two-character sequence CRLF is present in the subject
3682 string, it takes two dots to match it.
3683
3684 The handling of dot is entirely independent of the handling of circum-
3685 flex and dollar, the only relationship being that they both involve
3686 newlines. Dot has no special meaning in a character class.
3687
3688
3689 MATCHING A SINGLE BYTE
3690
3691 Outside a character class, the escape sequence \C matches any one byte,
3692 both in and out of UTF-8 mode. Unlike a dot, it always matches any
3693 line-ending characters. The feature is provided in Perl in order to
3694 match individual bytes in UTF-8 mode. Because it breaks up UTF-8 char-
3695 acters into individual bytes, what remains in the string may be a mal-
3696 formed UTF-8 string. For this reason, the \C escape sequence is best
3697 avoided.
3698
3699 PCRE does not allow \C to appear in lookbehind assertions (described
3700 below), because in UTF-8 mode this would make it impossible to calcu-
3701 late the length of the lookbehind.
3702
3703
3704 SQUARE BRACKETS AND CHARACTER CLASSES
3705
3706 An opening square bracket introduces a character class, terminated by a
3707 closing square bracket. A closing square bracket on its own is not spe-
3708 cial. If a closing square bracket is required as a member of the class,
3709 it should be the first data character in the class (after an initial
3710 circumflex, if present) or escaped with a backslash.
3711
3712 A character class matches a single character in the subject. In UTF-8
3713 mode, the character may occupy more than one byte. A matched character
3714 must be in the set of characters defined by the class, unless the first
3715 character in the class definition is a circumflex, in which case the
3716 subject character must not be in the set defined by the class. If a
3717 circumflex is actually required as a member of the class, ensure it is
3718 not the first character, or escape it with a backslash.
3719
3720 For example, the character class [aeiou] matches any lower case vowel,
3721 while [^aeiou] matches any character that is not a lower case vowel.
3722 Note that a circumflex is just a convenient notation for specifying the
3723 characters that are in the class by enumerating those that are not. A
3724 class that starts with a circumflex is not an assertion: it still con-
3725 sumes a character from the subject string, and therefore it fails if
3726 the current pointer is at the end of the string.
3727
3728 In UTF-8 mode, characters with values greater than 255 can be included
3729 in a class as a literal string of bytes, or by using the \x{ escaping
3730 mechanism.
3731
3732 When caseless matching is set, any letters in a class represent both
3733 their upper case and lower case versions, so for example, a caseless
3734 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
3735 match "A", whereas a caseful version would. In UTF-8 mode, PCRE always
3736 understands the concept of case for characters whose values are less
3737 than 128, so caseless matching is always possible. For characters with
3738 higher values, the concept of case is supported if PCRE is compiled
3739 with Unicode property support, but not otherwise. If you want to use
3740 caseless matching for characters 128 and above, you must ensure that
3741 PCRE is compiled with Unicode property support as well as with UTF-8
3742 support.
3743
3744 Characters that might indicate line breaks are never treated in any
3745 special way when matching character classes, whatever line-ending
3746 sequence is in use, and whatever setting of the PCRE_DOTALL and
3747 PCRE_MULTILINE options is used. A class such as [^a] always matches one
3748 of these characters.
3749
3750 The minus (hyphen) character can be used to specify a range of charac-
3751 ters in a character class. For example, [d-m] matches any letter
3752 between d and m, inclusive. If a minus character is required in a
3753 class, it must be escaped with a backslash or appear in a position
3754 where it cannot be interpreted as indicating a range, typically as the
3755 first or last character in the class.
3756
3757 It is not possible to have the literal character "]" as the end charac-
3758 ter of a range. A pattern such as [W-]46] is interpreted as a class of
3759 two characters ("W" and "-") followed by a literal string "46]", so it
3760 would match "W46]" or "-46]". However, if the "]" is escaped with a
3761 backslash it is interpreted as the end of range, so [W-\]46] is inter-
3762 preted as a class containing a range followed by two other characters.
3763 The octal or hexadecimal representation of "]" can also be used to end
3764 a range.
3765
3766 Ranges operate in the collating sequence of character values. They can
3767 also be used for characters specified numerically, for example
3768 [\000-\037]. In UTF-8 mode, ranges can include characters whose values
3769 are greater than 255, for example [\x{100}-\x{2ff}].
3770
3771 If a range that includes letters is used when caseless matching is set,
3772 it matches the letters in either case. For example, [W-c] is equivalent
3773 to [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if
3774 character tables for a French locale are in use, [\xc8-\xcb] matches
3775 accented E characters in both cases. In UTF-8 mode, PCRE supports the
3776 concept of case for characters with values greater than 128 only when
3777 it is compiled with Unicode property support.
3778
3779 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
3780 in a character class, and add the characters that they match to the
3781 class. For example, [\dABCDEF] matches any hexadecimal digit. A circum-
3782 flex can conveniently be used with the upper case character types to
3783 specify a more restricted set of characters than the matching lower
3784 case type. For example, the class [^\W_] matches any letter or digit,
3785 but not underscore.
3786
3787 The only metacharacters that are recognized in character classes are
3788 backslash, hyphen (only where it can be interpreted as specifying a
3789 range), circumflex (only at the start), opening square bracket (only
3790 when it can be interpreted as introducing a POSIX class name - see the
3791 next section), and the terminating closing square bracket. However,
3792 escaping other non-alphanumeric characters does no harm.
3793
3794
3795 POSIX CHARACTER CLASSES
3796
3797 Perl supports the POSIX notation for character classes. This uses names
3798 enclosed by [: and :] within the enclosing square brackets. PCRE also
3799 supports this notation. For example,
3800
3801 [01[:alpha:]%]
3802
3803 matches "0", "1", any alphabetic character, or "%". The supported class
3804 names are
3805
3806 alnum letters and digits
3807 alpha letters
3808 ascii character codes 0 - 127
3809 blank space or tab only
3810 cntrl control characters
3811 digit decimal digits (same as \d)
3812 graph printing characters, excluding space
3813 lower lower case letters
3814 print printing characters, including space
3815 punct printing characters, excluding letters and digits
3816 space white space (not quite the same as \s)
3817 upper upper case letters
3818 word "word" characters (same as \w)
3819 xdigit hexadecimal digits
3820
3821 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
3822 and space (32). Notice that this list includes the VT character (code
3823 11). This makes "space" different to \s, which does not include VT (for
3824 Perl compatibility).
3825
3826 The name "word" is a Perl extension, and "blank" is a GNU extension
3827 from Perl 5.8. Another Perl extension is negation, which is indicated
3828 by a ^ character after the colon. For example,
3829
3830 [12[:^digit:]]
3831
3832 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
3833 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
3834 these are not supported, and an error is given if they are encountered.
3835
3836 In UTF-8 mode, characters with values greater than 128 do not match any
3837 of the POSIX character classes.
3838
3839
3840 VERTICAL BAR
3841
3842 Vertical bar characters are used to separate alternative patterns. For
3843 example, the pattern
3844
3845 gilbert|sullivan
3846
3847 matches either "gilbert" or "sullivan". Any number of alternatives may
3848 appear, and an empty alternative is permitted (matching the empty
3849 string). The matching process tries each alternative in turn, from left
3850 to right, and the first one that succeeds is used. If the alternatives
3851 are within a subpattern (defined below), "succeeds" means matching the
3852 rest of the main pattern as well as the alternative in the subpattern.
3853
3854
3855 INTERNAL OPTION SETTING
3856
3857 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
3858 PCRE_EXTENDED options (which are Perl-compatible) can be changed from
3859 within the pattern by a sequence of Perl option letters enclosed
3860 between "(?" and ")". The option letters are
3861
3862 i for PCRE_CASELESS
3863 m for PCRE_MULTILINE
3864 s for PCRE_DOTALL
3865 x for PCRE_EXTENDED
3866
3867 For example, (?im) sets caseless, multiline matching. It is also possi-
3868 ble to unset these options by preceding the letter with a hyphen, and a
3869 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
3870 LESS and PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED,
3871 is also permitted. If a letter appears both before and after the
3872 hyphen, the option is unset.
3873
3874 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA
3875 can be changed in the same way as the Perl-compatible options by using
3876 the characters J, U and X respectively.
3877
3878 When one of these option changes occurs at top level (that is, not
3879 inside subpattern parentheses), the change applies to the remainder of
3880 the pattern that follows. If the change is placed right at the start of
3881 a pattern, PCRE extracts it into the global options (and it will there-
3882 fore show up in data extracted by the pcre_fullinfo() function).
3883
3884 An option change within a subpattern (see below for a description of
3885 subpatterns) affects only that part of the current pattern that follows
3886 it, so
3887
3888 (a(?i)b)c
3889
3890 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
3891 used). By this means, options can be made to have different settings
3892 in different parts of the pattern. Any changes made in one alternative
3893 do carry on into subsequent branches within the same subpattern. For
3894 example,
3895
3896 (a(?i)b|c)
3897
3898 matches "ab", "aB", "c", and "C", even though when matching "C" the
3899 first branch is abandoned before the option setting. This is because
3900 the effects of option settings happen at compile time. There would be
3901 some very weird behaviour otherwise.
3902
3903 Note: There are other PCRE-specific options that can be set by the
3904 application when the compile or match functions are called. In some
3905 cases the pattern can contain special leading sequences such as (*CRLF)
3906 to override what the application has set or what has been defaulted.
3907 Details are given in the section entitled "Newline sequences" above.
3908 There is also the (*UTF8) leading sequence that can be used to set
3909 UTF-8 mode; this is equivalent to setting the PCRE_UTF8 option.
3910
3911
3912 SUBPATTERNS
3913
3914 Subpatterns are delimited by parentheses (round brackets), which can be
3915 nested. Turning part of a pattern into a subpattern does two things:
3916
3917 1. It localizes a set of alternatives. For example, the pattern
3918
3919 cat(aract|erpillar|)
3920
3921 matches one of the words "cat", "cataract", or "caterpillar". Without
3922 the parentheses, it would match "cataract", "erpillar" or an empty
3923 string.
3924
3925 2. It sets up the subpattern as a capturing subpattern. This means
3926 that, when the whole pattern matches, that portion of the subject
3927 string that matched the subpattern is passed back to the caller via the
3928 ovector argument of pcre_exec(). Opening parentheses are counted from
3929 left to right (starting from 1) to obtain numbers for the capturing
3930 subpatterns.
3931
3932 For example, if the string "the red king" is matched against the pat-
3933 tern
3934
3935 the ((red|white) (king|queen))
3936
3937 the captured substrings are "red king", "red", and "king", and are num-
3938 bered 1, 2, and 3, respectively.
3939
3940 The fact that plain parentheses fulfil two functions is not always
3941 helpful. There are often times when a grouping subpattern is required
3942 without a capturing requirement. If an opening parenthesis is followed
3943 by a question mark and a colon, the subpattern does not do any captur-
3944 ing, and is not counted when computing the number of any subsequent
3945 capturing subpatterns. For example, if the string "the white queen" is
3946 matched against the pattern
3947
3948 the ((?:red|white) (king|queen))
3949
3950 the captured substrings are "white queen" and "queen", and are numbered
3951 1 and 2. The maximum number of capturing subpatterns is 65535.
3952
3953 As a convenient shorthand, if any option settings are required at the
3954 start of a non-capturing subpattern, the option letters may appear
3955 between the "?" and the ":". Thus the two patterns
3956
3957 (?i:saturday|sunday)
3958 (?:(?i)saturday|sunday)
3959
3960 match exactly the same set of strings. Because alternative branches are
3961 tried from left to right, and options are not reset until the end of
3962 the subpattern is reached, an option setting in one branch does affect
3963 subsequent branches, so the above patterns match "SUNDAY" as well as
3964 "Saturday".
3965
3966
3967 DUPLICATE SUBPATTERN NUMBERS
3968
3969 Perl 5.10 introduced a feature whereby each alternative in a subpattern
3970 uses the same numbers for its capturing parentheses. Such a subpattern
3971 starts with (?| and is itself a non-capturing subpattern. For example,
3972 consider this pattern:
3973
3974 (?|(Sat)ur|(Sun))day
3975
3976 Because the two alternatives are inside a (?| group, both sets of cap-
3977 turing parentheses are numbered one. Thus, when the pattern matches,
3978 you can look at captured substring number one, whichever alternative
3979 matched. This construct is useful when you want to capture part, but
3980 not all, of one of a number of alternatives. Inside a (?| group, paren-
3981 theses are numbered as usual, but the number is reset at the start of
3982 each branch. The numbers of any capturing buffers that follow the sub-
3983 pattern start after the highest number used in any branch. The follow-
3984 ing example is taken from the Perl documentation. The numbers under-
3985 neath show in which buffer the captured content will be stored.
3986
3987 # before ---------------branch-reset----------- after
3988 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
3989 # 1 2 2 3 2 3 4
3990
3991 A backreference or a recursive call to a numbered subpattern always
3992 refers to the first one in the pattern with the given number.
3993
3994 An alternative approach to using this "branch reset" feature is to use
3995 duplicate named subpatterns, as described in the next section.
3996
3997
3998 NAMED SUBPATTERNS
3999
4000 Identifying capturing parentheses by number is simple, but it can be
4001 very hard to keep track of the numbers in complicated regular expres-
4002 sions. Furthermore, if an expression is modified, the numbers may
4003 change. To help with this difficulty, PCRE supports the naming of sub-
4004 patterns. This feature was not added to Perl until release 5.10. Python
4005 had the feature earlier, and PCRE introduced it at release 4.0, using
4006 the Python syntax. PCRE now supports both the Perl and the Python syn-
4007 tax.
4008
4009 In PCRE, a subpattern can be named in one of three ways: (?<name>...)
4010 or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
4011 to capturing parentheses from other parts of the pattern, such as back-
4012 references, recursion, and conditions, can be made by name as well as
4013 by number.
4014
4015 Names consist of up to 32 alphanumeric characters and underscores.
4016 Named capturing parentheses are still allocated numbers as well as
4017 names, exactly as if the names were not present. The PCRE API provides
4018 function calls for extracting the name-to-number translation table from
4019 a compiled pattern. There is also a convenience function for extracting
4020 a captured substring by name.
4021
4022 By default, a name must be unique within a pattern, but it is possible
4023 to relax this constraint by setting the PCRE_DUPNAMES option at compile
4024 time. This can be useful for patterns where only one instance of the
4025 named parentheses can match. Suppose you want to match the name of a
4026 weekday, either as a 3-letter abbreviation or as the full name, and in
4027 both cases you want to extract the abbreviation. This pattern (ignoring
4028 the line breaks) does the job:
4029
4030 (?<DN>Mon|Fri|Sun)(?:day)?|
4031 (?<DN>Tue)(?:sday)?|
4032 (?<DN>Wed)(?:nesday)?|
4033 (?<DN>Thu)(?:rsday)?|
4034 (?<DN>Sat)(?:urday)?
4035
4036 There are five capturing substrings, but only one is ever set after a
4037 match. (An alternative way of solving this problem is to use a "branch
4038 reset" subpattern, as described in the previous section.)
4039
4040 The convenience function for extracting the data by name returns the
4041 substring for the first (and in this example, the only) subpattern of
4042 that name that matched. This saves searching to find which numbered
4043 subpattern it was. If you make a reference to a non-unique named sub-
4044 pattern from elsewhere in the pattern, the one that corresponds to the
4045 lowest number is used. For further details of the interfaces for han-
4046 dling named subpatterns, see the pcreapi documentation.
4047
4048 Warning: You cannot use different names to distinguish between two sub-
4049 patterns with the same number (see the previous section) because PCRE
4050 uses only the numbers when matching.
4051
4052
4053 REPETITION
4054
4055 Repetition is specified by quantifiers, which can follow any of the
4056 following items:
4057
4058 a literal data character
4059 the dot metacharacter
4060 the \C escape sequence
4061 the \X escape sequence (in UTF-8 mode with Unicode properties)
4062 the \R escape sequence
4063 an escape such as \d that matches a single character
4064 a character class
4065 a back reference (see next section)
4066 a parenthesized subpattern (unless it is an assertion)
4067
4068 The general repetition quantifier specifies a minimum and maximum num-
4069 ber of permitted matches, by giving the two numbers in curly brackets
4070 (braces), separated by a comma. The numbers must be less than 65536,
4071 and the first must be less than or equal to the second. For example:
4072
4073 z{2,4}
4074
4075 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
4076 special character. If the second number is omitted, but the comma is
4077 present, there is no upper limit; if the second number and the comma
4078 are both omitted, the quantifier specifies an exact number of required
4079 matches. Thus
4080
4081 [aeiou]{3,}
4082
4083 matches at least 3 successive vowels, but may match many more, while
4084
4085 \d{8}
4086
4087 matches exactly 8 digits. An opening curly bracket that appears in a
4088 position where a quantifier is not allowed, or one that does not match
4089 the syntax of a quantifier, is taken as a literal character. For exam-
4090 ple, {,6} is not a quantifier, but a literal string of four characters.
4091
4092 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to
4093 individual bytes. Thus, for example, \x{100}{2} matches two UTF-8 char-
4094 acters, each of which is represented by a two-byte sequence. Similarly,
4095 when Unicode property support is available, \X{3} matches three Unicode
4096 extended sequences, each of which may be several bytes long (and they
4097 may be of different lengths).
4098
4099 The quantifier {0} is permitted, causing the expression to behave as if
4100 the previous item and the quantifier were not present. This may be use-
4101 ful for subpatterns that are referenced as subroutines from elsewhere
4102 in the pattern. Items other than subpatterns that have a {0} quantifier
4103 are omitted from the compiled pattern.
4104
4105 For convenience, the three most common quantifiers have single-charac-
4106 ter abbreviations:
4107
4108 * is equivalent to {0,}
4109 + is equivalent to {1,}
4110 ? is equivalent to {0,1}
4111
4112 It is possible to construct infinite loops by following a subpattern
4113 that can match no characters with a quantifier that has no upper limit,
4114 for example:
4115
4116 (a?)*
4117
4118 Earlier versions of Perl and PCRE used to give an error at compile time
4119 for such patterns. However, because there are cases where this can be
4120 useful, such patterns are now accepted, but if any repetition of the
4121 subpattern does in fact match no characters, the loop is forcibly bro-
4122 ken.
4123
4124 By default, the quantifiers are "greedy", that is, they match as much
4125 as possible (up to the maximum number of permitted times), without
4126 causing the rest of the pattern to fail. The classic example of where
4127 this gives problems is in trying to match comments in C programs. These
4128 appear between /* and */ and within the comment, individual * and /
4129 characters may appear. An attempt to match C comments by applying the
4130 pattern
4131
4132 /\*.*\*/
4133
4134 to the string
4135
4136 /* first comment */ not comment /* second comment */
4137
4138 fails, because it matches the entire string owing to the greediness of
4139 the .* item.
4140
4141 However, if a quantifier is followed by a question mark, it ceases to
4142 be greedy, and instead matches the minimum number of times possible, so
4143 the pattern
4144
4145 /\*.*?\*/
4146
4147 does the right thing with the C comments. The meaning of the various
4148 quantifiers is not otherwise changed, just the preferred number of
4149 matches. Do not confuse this use of question mark with its use as a
4150 quantifier in its own right. Because it has two uses, it can sometimes
4151 appear doubled, as in
4152
4153 \d??\d
4154
4155 which matches one digit by preference, but can match two if that is the
4156 only way the rest of the pattern matches.
4157
4158 If the PCRE_UNGREEDY option is set (an option that is not available in
4159 Perl), the quantifiers are not greedy by default, but individual ones
4160 can be made greedy by following them with a question mark. In other
4161 words, it inverts the default behaviour.
4162
4163 When a parenthesized subpattern is quantified with a minimum repeat
4164 count that is greater than 1 or with a limited maximum, more memory is
4165 required for the compiled pattern, in proportion to the size of the
4166 minimum or maximum.
4167
4168 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
4169 alent to Perl's /s) is set, thus allowing the dot to match newlines,
4170 the pattern is implicitly anchored, because whatever follows will be
4171 tried against every character position in the subject string, so there
4172 is no point in retrying the overall match at any position after the
4173 first. PCRE normally treats such a pattern as though it were preceded
4174 by \A.
4175
4176 In cases where it is known that the subject string contains no new-
4177 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
4178 mization, or alternatively using ^ to indicate anchoring explicitly.
4179
4180 However, there is one situation where the optimization cannot be used.
4181 When .* is inside capturing parentheses that are the subject of a
4182 backreference elsewhere in the pattern, a match at the start may fail
4183 where a later one succeeds. Consider, for example:
4184
4185 (.*)abc\1
4186
4187 If the subject is "xyz123abc123" the match point is the fourth charac-
4188 ter. For this reason, such a pattern is not implicitly anchored.
4189
4190 When a capturing subpattern is repeated, the value captured is the sub-
4191 string that matched the final iteration. For example, after
4192
4193 (tweedle[dume]{3}\s*)+
4194
4195 has matched "tweedledum tweedledee" the value of the captured substring
4196 is "tweedledee". However, if there are nested capturing subpatterns,
4197 the corresponding captured values may have been set in previous itera-
4198 tions. For example, after
4199
4200 /(a|(b))+/
4201
4202 matches "aba" the value of the second captured substring is "b".
4203
4204
4205 ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS
4206
4207 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
4208 repetition, failure of what follows normally causes the repeated item
4209 to be re-evaluated to see if a different number of repeats allows the
4210 rest of the pattern to match. Sometimes it is useful to prevent this,
4211 either to change the nature of the match, or to cause it fail earlier
4212 than it otherwise might, when the author of the pattern knows there is
4213 no point in carrying on.
4214
4215 Consider, for example, the pattern \d+foo when applied to the subject
4216 line
4217
4218 123456bar
4219
4220 After matching all 6 digits and then failing to match "foo", the normal
4221 action of the matcher is to try again with only 5 digits matching the
4222 \d+ item, and then with 4, and so on, before ultimately failing.
4223 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
4224 the means for specifying that once a subpattern has matched, it is not
4225 to be re-evaluated in this way.
4226
4227 If we use atomic grouping for the previous example, the matcher gives
4228 up immediately on failing to match "foo" the first time. The notation
4229 is a kind of special parenthesis, starting with (?> as in this example:
4230
4231 (?>\d+)foo
4232
4233 This kind of parenthesis "locks up" the part of the pattern it con-
4234 tains once it has matched, and a failure further into the pattern is
4235 prevented from backtracking into it. Backtracking past it to previous
4236 items, however, works as normal.
4237
4238 An alternative description is that a subpattern of this type matches
4239 the string of characters that an identical standalone pattern would
4240 match, if anchored at the current point in the subject string.
4241
4242 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
4243 such as the above example can be thought of as a maximizing repeat that
4244 must swallow everything it can. So, while both \d+ and \d+? are pre-
4245 pared to adjust the number of digits they match in order to make the
4246 rest of the pattern match, (?>\d+) can only match an entire sequence of
4247 digits.
4248
4249 Atomic groups in general can of course contain arbitrarily complicated
4250 subpatterns, and can be nested. However, when the subpattern for an
4251 atomic group is just a single repeated item, as in the example above, a
4252 simpler notation, called a "possessive quantifier" can be used. This
4253 consists of an additional + character following a quantifier. Using
4254 this notation, the previous example can be rewritten as
4255
4256 \d++foo
4257
4258 Note that a possessive quantifier can be used with an entire group, for
4259 example:
4260
4261 (abc|xyz){2,3}+
4262
4263 Possessive quantifiers are always greedy; the setting of the
4264 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
4265 simpler forms of atomic group. However, there is no difference in the
4266 meaning of a possessive quantifier and the equivalent atomic group,
4267 though there may be a performance difference; possessive quantifiers
4268 should be slightly faster.
4269
4270 The possessive quantifier syntax is an extension to the Perl 5.8 syn-
4271 tax. Jeffrey Friedl originated the idea (and the name) in the first
4272 edition of his book. Mike McCloskey liked it, so implemented it when he
4273 built Sun's Java package, and PCRE copied it from there. It ultimately
4274 found its way into Perl at release 5.10.
4275
4276 PCRE has an optimization that automatically "possessifies" certain sim-
4277 ple pattern constructs. For example, the sequence A+B is treated as
4278 A++B because there is no point in backtracking into a sequence of A's
4279 when B must follow.
4280
4281 When a pattern contains an unlimited repeat inside a subpattern that
4282 can itself be repeated an unlimited number of times, the use of an
4283 atomic group is the only way to avoid some failing matches taking a
4284 very long time indeed. The pattern
4285
4286 (\D+|<\d+>)*[!?]
4287
4288 matches an unlimited number of substrings that either consist of non-
4289 digits, or digits enclosed in <>, followed by either ! or ?. When it
4290 matches, it runs quickly. However, if it is applied to
4291
4292 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
4293
4294 it takes a long time before reporting failure. This is because the
4295 string can be divided between the internal \D+ repeat and the external
4296 * repeat in a large number of ways, and all have to be tried. (The
4297 example uses [!?] rather than a single character at the end, because
4298 both PCRE and Perl have an optimization that allows for fast failure
4299 when a single character is used. They remember the last single charac-
4300 ter that is required for a match, and fail early if it is not present
4301 in the string.) If the pattern is changed so that it uses an atomic
4302 group, like this:
4303
4304 ((?>\D+)|<\d+>)*[!?]
4305
4306 sequences of non-digits cannot be broken, and failure happens quickly.
4307
4308
4309 BACK REFERENCES
4310
4311 Outside a character class, a backslash followed by a digit greater than
4312 0 (and possibly further digits) is a back reference to a capturing sub-
4313 pattern earlier (that is, to its left) in the pattern, provided there
4314 have been that many previous capturing left parentheses.
4315
4316 However, if the decimal number following the backslash is less than 10,
4317 it is always taken as a back reference, and causes an error only if
4318 there are not that many capturing left parentheses in the entire pat-
4319 tern. In other words, the parentheses that are referenced need not be
4320 to the left of the reference for numbers less than 10. A "forward back
4321 reference" of this type can make sense when a repetition is involved
4322 and the subpattern to the right has participated in an earlier itera-
4323 tion.
4324
4325 It is not possible to have a numerical "forward back reference" to a
4326 subpattern whose number is 10 or more using this syntax because a
4327 sequence such as \50 is interpreted as a character defined in octal.
4328 See the subsection entitled "Non-printing characters" above for further
4329 details of the handling of digits following a backslash. There is no
4330 such problem when named parentheses are used. A back reference to any
4331 subpattern is possible using named parentheses (see below).
4332
4333 Another way of avoiding the ambiguity inherent in the use of digits
4334 following a backslash is to use the \g escape sequence, which is a fea-
4335 ture introduced in Perl 5.10. This escape must be followed by an
4336 unsigned number or a negative number, optionally enclosed in braces.
4337 These examples are all identical:
4338
4339 (ring), \1
4340 (ring), \g1
4341 (ring), \g{1}
4342
4343 An unsigned number specifies an absolute reference without the ambigu-
4344 ity that is present in the older syntax. It is also useful when literal
4345 digits follow the reference. A negative number is a relative reference.
4346 Consider this example:
4347
4348 (abc(def)ghi)\g{-1}
4349
4350 The sequence \g{-1} is a reference to the most recently started captur-
4351 ing subpattern before \g, that is, is it equivalent to \2. Similarly,
4352 \g{-2} would be equivalent to \1. The use of relative references can be
4353 helpful in long patterns, and also in patterns that are created by
4354 joining together fragments that contain references within themselves.
4355
4356 A back reference matches whatever actually matched the capturing sub-
4357 pattern in the current subject string, rather than anything matching
4358 the subpattern itself (see "Subpatterns as subroutines" below for a way
4359 of doing that). So the pattern
4360
4361 (sens|respons)e and \1ibility
4362
4363 matches "sense and sensibility" and "response and responsibility", but
4364 not "sense and responsibility". If caseful matching is in force at the
4365 time of the back reference, the case of letters is relevant. For exam-
4366 ple,
4367
4368 ((?i)rah)\s+\1
4369
4370 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
4371 original capturing subpattern is matched caselessly.
4372
4373 There are several different ways of writing back references to named
4374 subpatterns. The .NET syntax \k{name} and the Perl syntax \k<name> or
4375 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's
4376 unified back reference syntax, in which \g can be used for both numeric
4377 and named references, is also supported. We could rewrite the above
4378 example in any of the following ways:
4379
4380 (?<p1>(?i)rah)\s+\k<p1>
4381 (?'p1'(?i)rah)\s+\k{p1}
4382 (?P<p1>(?i)rah)\s+(?P=p1)
4383 (?<p1>(?i)rah)\s+\g{p1}
4384
4385 A subpattern that is referenced by name may appear in the pattern
4386 before or after the reference.
4387
4388 There may be more than one back reference to the same subpattern. If a
4389 subpattern has not actually been used in a particular match, any back
4390 references to it always fail. For example, the pattern
4391
4392 (a|(bc))\2
4393
4394 always fails if it starts to match "a" rather than "bc". Because there
4395 may be many capturing parentheses in a pattern, all digits following
4396 the backslash are taken as part of a potential back reference number.
4397 If the pattern continues with a digit character, some delimiter must be
4398 used to terminate the back reference. If the PCRE_EXTENDED option is
4399 set, this can be whitespace. Otherwise an empty comment (see "Com-
4400 ments" below) can be used.
4401
4402 A back reference that occurs inside the parentheses to which it refers
4403 fails when the subpattern is first used, so, for example, (a\1) never
4404 matches. However, such references can be useful inside repeated sub-
4405 patterns. For example, the pattern
4406
4407 (a|b\1)+
4408
4409 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
4410 ation of the subpattern, the back reference matches the character
4411 string corresponding to the previous iteration. In order for this to
4412 work, the pattern must be such that the first iteration does not need
4413 to match the back reference. This can be done using alternation, as in
4414 the example above, or by a quantifier with a minimum of zero.
4415
4416
4417 ASSERTIONS
4418
4419 An assertion is a test on the characters following or preceding the
4420 current matching point that does not actually consume any characters.
4421 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
4422 described above.
4423
4424 More complicated assertions are coded as subpatterns. There are two
4425 kinds: those that look ahead of the current position in the subject
4426 string, and those that look behind it. An assertion subpattern is
4427 matched in the normal way, except that it does not cause the current
4428 matching position to be changed.
4429
4430 Assertion subpatterns are not capturing subpatterns, and may not be
4431 repeated, because it makes no sense to assert the same thing several
4432 times. If any kind of assertion contains capturing subpatterns within
4433 it, these are counted for the purposes of numbering the capturing sub-
4434 patterns in the whole pattern. However, substring capturing is carried
4435 out only for positive assertions, because it does not make sense for
4436 negative assertions.
4437
4438 Lookahead assertions
4439
4440 Lookahead assertions start with (?= for positive assertions and (?! for
4441 negative assertions. For example,
4442
4443 \w+(?=;)
4444
4445 matches a word followed by a semicolon, but does not include the semi-
4446 colon in the match, and
4447
4448 foo(?!bar)
4449
4450 matches any occurrence of "foo" that is not followed by "bar". Note
4451 that the apparently similar pattern
4452
4453 (?!foo)bar
4454
4455 does not find an occurrence of "bar" that is preceded by something
4456 other than "foo"; it finds any occurrence of "bar" whatsoever, because
4457 the assertion (?!foo) is always true when the next three characters are
4458 "bar". A lookbehind assertion is needed to achieve the other effect.
4459
4460 If you want to force a matching failure at some point in a pattern, the
4461 most convenient way to do it is with (?!) because an empty string
4462 always matches, so an assertion that requires there not to be an empty
4463 string must always fail.
4464
4465 Lookbehind assertions
4466
4467 Lookbehind assertions start with (?<= for positive assertions and (?<!
4468 for negative assertions. For example,
4469
4470 (?<!foo)bar
4471
4472 does find an occurrence of "bar" that is not preceded by "foo". The
4473 contents of a lookbehind assertion are restricted such that all the
4474 strings it matches must have a fixed length. However, if there are sev-
4475 eral top-level alternatives, they do not all have to have the same
4476 fixed length. Thus
4477
4478 (?<=bullock|donkey)
4479
4480 is permitted, but
4481
4482 (?<!dogs?|cats?)
4483
4484 causes an error at compile time. Branches that match different length
4485 strings are permitted only at the top level of a lookbehind assertion.
4486 This is an extension compared with Perl (at least for 5.8), which
4487 requires all branches to match the same length of string. An assertion
4488 such as
4489
4490 (?<=ab(c|de))
4491
4492 is not permitted, because its single top-level branch can match two
4493 different lengths, but it is acceptable if rewritten to use two top-
4494 level branches:
4495
4496 (?<=abc|abde)
4497
4498 In some cases, the Perl 5.10 escape sequence \K (see above) can be used
4499 instead of a lookbehind assertion; this is not restricted to a fixed-
4500 length.
4501
4502 The implementation of lookbehind assertions is, for each alternative,
4503 to temporarily move the current position back by the fixed length and
4504 then try to match. If there are insufficient characters before the cur-
4505 rent position, the assertion fails.
4506
4507 PCRE does not allow the \C escape (which matches a single byte in UTF-8
4508 mode) to appear in lookbehind assertions, because it makes it impossi-
4509 ble to calculate the length of the lookbehind. The \X and \R escapes,
4510 which can match different numbers of bytes, are also not permitted.
4511
4512 Possessive quantifiers can be used in conjunction with lookbehind
4513 assertions to specify efficient matching at the end of the subject
4514 string. Consider a simple pattern such as
4515
4516 abcd$
4517
4518 when applied to a long string that does not match. Because matching
4519 proceeds from left to right, PCRE will look for each "a" in the subject
4520 and then see if what follows matches the rest of the pattern. If the
4521 pattern is specified as
4522
4523 ^.*abcd$
4524
4525 the initial .* matches the entire string at first, but when this fails
4526 (because there is no following "a"), it backtracks to match all but the
4527 last character, then all but the last two characters, and so on. Once
4528 again the search for "a" covers the entire string, from right to left,
4529 so we are no better off. However, if the pattern is written as
4530
4531 ^.*+(?<=abcd)
4532
4533 there can be no backtracking for the .*+ item; it can match only the
4534 entire string. The subsequent lookbehind assertion does a single test
4535 on the last four characters. If it fails, the match fails immediately.
4536 For long strings, this approach makes a significant difference to the
4537 processing time.
4538
4539 Using multiple assertions
4540
4541 Several assertions (of any sort) may occur in succession. For example,
4542
4543 (?<=\d{3})(?<!999)foo
4544
4545 matches "foo" preceded by three digits that are not "999". Notice that
4546 each of the assertions is applied independently at the same point in
4547 the subject string. First there is a check that the previous three
4548 characters are all digits, and then there is a check that the same
4549 three characters are not "999". This pattern does not match "foo" pre-
4550 ceded by six characters, the first of which are digits and the last
4551 three of which are not "999". For example, it doesn't match "123abc-
4552 foo". A pattern to do that is
4553
4554 (?<=\d{3}...)(?<!999)foo
4555
4556 This time the first assertion looks at the preceding six characters,
4557 checking that the first three are digits, and then the second assertion
4558 checks that the preceding three characters are not "999".
4559
4560 Assertions can be nested in any combination. For example,
4561
4562 (?<=(?<!foo)bar)baz
4563
4564 matches an occurrence of "baz" that is preceded by "bar" which in turn
4565 is not preceded by "foo", while
4566
4567 (?<=\d{3}(?!999)...)foo
4568
4569 is another pattern that matches "foo" preceded by three digits and any
4570 three characters that are not "999".
4571
4572
4573 CONDITIONAL SUBPATTERNS
4574
4575 It is possible to cause the matching process to obey a subpattern con-
4576 ditionally or to choose between two alternative subpatterns, depending
4577 on the result of an assertion, or whether a previous capturing subpat-
4578 tern matched or not. The two possible forms of conditional subpattern
4579 are
4580
4581 (?(condition)yes-pattern)
4582 (?(condition)yes-pattern|no-pattern)
4583
4584 If the condition is satisfied, the yes-pattern is used; otherwise the
4585 no-pattern (if present) is used. If there are more than two alterna-
4586 tives in the subpattern, a compile-time error occurs.
4587
4588 There are four kinds of condition: references to subpatterns, refer-
4589 ences to recursion, a pseudo-condition called DEFINE, and assertions.
4590
4591 Checking for a used subpattern by number
4592
4593 If the text between the parentheses consists of a sequence of digits,
4594 the condition is true if the capturing subpattern of that number has
4595 previously matched. An alternative notation is to precede the digits
4596 with a plus or minus sign. In this case, the subpattern number is rela-
4597 tive rather than absolute. The most recently opened parentheses can be
4598 referenced by (?(-1), the next most recent by (?(-2), and so on. In
4599 looping constructs it can also make sense to refer to subsequent groups
4600 with constructs such as (?(+2).
4601
4602 Consider the following pattern, which contains non-significant white
4603 space to make it more readable (assume the PCRE_EXTENDED option) and to
4604 divide it into three parts for ease of discussion:
4605
4606 ( \( )? [^()]+ (?(1) \) )
4607
4608 The first part matches an optional opening parenthesis, and if that
4609 character is present, sets it as the first captured substring. The sec-
4610 ond part matches one or more characters that are not parentheses. The
4611 third part is a conditional subpattern that tests whether the first set
4612 of parentheses matched or not. If they did, that is, if subject started
4613 with an opening parenthesis, the condition is true, and so the yes-pat-
4614 tern is executed and a closing parenthesis is required. Otherwise,
4615 since no-pattern is not present, the subpattern matches nothing. In
4616 other words, this pattern matches a sequence of non-parentheses,
4617 optionally enclosed in parentheses.
4618
4619 If you were embedding this pattern in a larger one, you could use a
4620 relative reference:
4621
4622 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
4623
4624 This makes the fragment independent of the parentheses in the larger
4625 pattern.
4626
4627 Checking for a used subpattern by name
4628
4629 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
4630 used subpattern by name. For compatibility with earlier versions of
4631 PCRE, which had this facility before Perl, the syntax (?(name)...) is
4632 also recognized. However, there is a possible ambiguity with this syn-
4633 tax, because subpattern names may consist entirely of digits. PCRE
4634 looks first for a named subpattern; if it cannot find one and the name
4635 consists entirely of digits, PCRE looks for a subpattern of that num-
4636 ber, which must be greater than zero. Using subpattern names that con-
4637 sist entirely of digits is not recommended.
4638
4639 Rewriting the above example to use a named subpattern gives this:
4640
4641 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
4642
4643
4644 Checking for pattern recursion
4645
4646 If the condition is the string (R), and there is no subpattern with the
4647 name R, the condition is true if a recursive call to the whole pattern
4648 or any subpattern has been made. If digits or a name preceded by amper-
4649 sand follow the letter R, for example:
4650
4651 (?(R3)...) or (?(R&name)...)
4652
4653 the condition is true if the most recent recursion is into the subpat-
4654 tern whose number or name is given. This condition does not check the
4655 entire recursion stack.
4656
4657 At "top level", all these recursion test conditions are false. Recur-
4658 sive patterns are described below.
4659
4660 Defining subpatterns for use by reference only
4661
4662 If the condition is the string (DEFINE), and there is no subpattern
4663 with the name DEFINE, the condition is always false. In this case,
4664 there may be only one alternative in the subpattern. It is always
4665 skipped if control reaches this point in the pattern; the idea of
4666 DEFINE is that it can be used to define "subroutines" that can be ref-
4667 erenced from elsewhere. (The use of "subroutines" is described below.)
4668 For example, a pattern to match an IPv4 address could be written like
4669 this (ignore whitespace and line breaks):
4670
4671 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
4672 \b (?&byte) (\.(?&byte)){3} \b
4673
4674 The first part of the pattern is a DEFINE group inside which a another
4675 group named "byte" is defined. This matches an individual component of
4676 an IPv4 address (a number less than 256). When matching takes place,
4677 this part of the pattern is skipped because DEFINE acts like a false
4678 condition.
4679
4680 The rest of the pattern uses references to the named group to match the
4681 four dot-separated components of an IPv4 address, insisting on a word
4682 boundary at each end.
4683
4684 Assertion conditions
4685
4686 If the condition is not in any of the above formats, it must be an
4687 assertion. This may be a positive or negative lookahead or lookbehind
4688 assertion. Consider this pattern, again containing non-significant
4689 white space, and with the two alternatives on the second line:
4690
4691 (?(?=[^a-z]*[a-z])
4692 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
4693
4694 The condition is a positive lookahead assertion that matches an
4695 optional sequence of non-letters followed by a letter. In other words,
4696 it tests for the presence of at least one letter in the subject. If a
4697 letter is found, the subject is matched against the first alternative;
4698 otherwise it is matched against the second. This pattern matches
4699 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
4700 letters and dd are digits.
4701
4702
4703 COMMENTS
4704
4705 The sequence (?# marks the start of a comment that continues up to the
4706 next closing parenthesis. Nested parentheses are not permitted. The
4707 characters that make up a comment play no part in the pattern matching
4708 at all.
4709
4710 If the PCRE_EXTENDED option is set, an unescaped # character outside a
4711 character class introduces a comment that continues to immediately
4712 after the next newline in the pattern.
4713
4714
4715 RECURSIVE PATTERNS
4716
4717 Consider the problem of matching a string in parentheses, allowing for
4718 unlimited nested parentheses. Without the use of recursion, the best
4719 that can be done is to use a pattern that matches up to some fixed
4720 depth of nesting. It is not possible to handle an arbitrary nesting
4721 depth.
4722
4723 For some time, Perl has provided a facility that allows regular expres-
4724 sions to recurse (amongst other things). It does this by interpolating
4725 Perl code in the expression at run time, and the code can refer to the
4726 expression itself. A Perl pattern using code interpolation to solve the
4727 parentheses problem can be created like this:
4728
4729 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
4730
4731 The (?p{...}) item interpolates Perl code at run time, and in this case
4732 refers recursively to the pattern in which it appears.
4733
4734 Obviously, PCRE cannot support the interpolation of Perl code. Instead,
4735 it supports special syntax for recursion of the entire pattern, and
4736 also for individual subpattern recursion. After its introduction in
4737 PCRE and Python, this kind of recursion was subsequently introduced
4738 into Perl at release 5.10.
4739
4740 A special item that consists of (? followed by a number greater than
4741 zero and a closing parenthesis is a recursive call of the subpattern of
4742 the given number, provided that it occurs inside that subpattern. (If
4743 not, it is a "subroutine" call, which is described in the next sec-
4744 tion.) The special item (?R) or (?0) is a recursive call of the entire
4745 regular expression.
4746
4747 This PCRE pattern solves the nested parentheses problem (assume the
4748 PCRE_EXTENDED option is set so that white space is ignored):
4749
4750 \( ( (?>[^()]+) | (?R) )* \)
4751
4752 First it matches an opening parenthesis. Then it matches any number of
4753 substrings which can either be a sequence of non-parentheses, or a
4754 recursive match of the pattern itself (that is, a correctly parenthe-
4755 sized substring). Finally there is a closing parenthesis.
4756
4757 If this were part of a larger pattern, you would not want to recurse
4758 the entire pattern, so instead you could use this:
4759
4760 ( \( ( (?>[^()]+) | (?1) )* \) )
4761
4762 We have put the pattern into parentheses, and caused the recursion to
4763 refer to them instead of the whole pattern.
4764
4765 In a larger pattern, keeping track of parenthesis numbers can be
4766 tricky. This is made easier by the use of relative references. (A Perl
4767 5.10 feature.) Instead of (?1) in the pattern above you can write
4768 (?-2) to refer to the second most recently opened parentheses preceding
4769 the recursion. In other words, a negative number counts capturing
4770 parentheses leftwards from the point at which it is encountered.
4771
4772 It is also possible to refer to subsequently opened parentheses, by
4773 writing references such as (?+2). However, these cannot be recursive
4774 because the reference is not inside the parentheses that are refer-
4775 enced. They are always "subroutine" calls, as described in the next
4776 section.
4777
4778 An alternative approach is to use named parentheses instead. The Perl
4779 syntax for this is (?&name); PCRE's earlier syntax (?P>name) is also
4780 supported. We could rewrite the above example as follows:
4781
4782 (?<pn> \( ( (?>[^()]+) | (?&pn) )* \) )
4783
4784 If there is more than one subpattern with the same name, the earliest
4785 one is used.
4786
4787 This particular example pattern that we have been looking at contains
4788 nested unlimited repeats, and so the use of atomic grouping for match-
4789 ing strings of non-parentheses is important when applying the pattern
4790 to strings that do not match. For example, when this pattern is applied
4791 to
4792
4793 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
4794
4795 it yields "no match" quickly. However, if atomic grouping is not used,
4796 the match runs for a very long time indeed because there are so many
4797 different ways the + and * repeats can carve up the subject, and all
4798 have to be tested before failure can be reported.
4799
4800 At the end of a match, the values set for any capturing subpatterns are
4801 those from the outermost level of the recursion at which the subpattern
4802 value is set. If you want to obtain intermediate values, a callout
4803 function can be used (see below and the pcrecallout documentation). If
4804 the pattern above is matched against
4805
4806 (ab(cd)ef)
4807
4808 the value for the capturing parentheses is "ef", which is the last
4809 value taken on at the top level. If additional parentheses are added,
4810 giving
4811
4812 \( ( ( (?>[^()]+) | (?R) )* ) \)
4813 ^ ^
4814 ^ ^
4815
4816 the string they capture is "ab(cd)ef", the contents of the top level
4817 parentheses. If there are more than 15 capturing parentheses in a pat-
4818 tern, PCRE has to obtain extra memory to store data during a recursion,
4819 which it does by using pcre_malloc, freeing it via pcre_free after-
4820 wards. If no memory can be obtained, the match fails with the
4821 PCRE_ERROR_NOMEMORY error.
4822
4823 Do not confuse the (?R) item with the condition (R), which tests for
4824 recursion. Consider this pattern, which matches text in angle brack-
4825 ets, allowing for arbitrary nesting. Only digits are allowed in nested
4826 brackets (that is, when recursing), whereas any characters are permit-
4827 ted at the outer level.
4828
4829 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
4830
4831 In this pattern, (?(R) is the start of a conditional subpattern, with
4832 two different alternatives for the recursive and non-recursive cases.
4833 The (?R) item is the actual recursive call.
4834
4835 Recursion difference from Perl
4836
4837 In PCRE (like Python, but unlike Perl), a recursive subpattern call is
4838 always treated as an atomic group. That is, once it has matched some of
4839 the subject string, it is never re-entered, even if it contains untried
4840 alternatives and there is a subsequent matching failure. This can be
4841 illustrated by the following pattern, which purports to match a palin-
4842 dromic string that contains an odd number of characters (for example,
4843 "a", "aba", "abcba", "abcdcba"):
4844
4845 ^(.|(.)(?1)\2)$
4846
4847 The idea is that it either matches a single character, or two identical
4848 characters surrounding a sub-palindrome. In Perl, this pattern works;
4849 in PCRE it does not if the pattern is longer than three characters.
4850 Consider the subject string "abcba":
4851
4852 At the top level, the first character is matched, but as it is not at
4853 the end of the string, the first alternative fails; the second alterna-
4854 tive is taken and the recursion kicks in. The recursive call to subpat-
4855 tern 1 successfully matches the next character ("b"). (Note that the
4856 beginning and end of line tests are not part of the recursion).
4857
4858 Back at the top level, the next character ("c") is compared with what
4859 subpattern 2 matched, which was "a". This fails. Because the recursion
4860 is treated as an atomic group, there are now no backtracking points,
4861 and so the entire match fails. (Perl is able, at this point, to re-
4862 enter the recursion and try the second alternative.) However, if the
4863 pattern is written with the alternatives in the other order, things are
4864 different:
4865
4866 ^((.)(?1)\2|.)$
4867
4868 This time, the recursing alternative is tried first, and continues to
4869 recurse until it runs out of characters, at which point the recursion
4870 fails. But this time we do have another alternative to try at the
4871 higher level. That is the big difference: in the previous case the
4872 remaining alternative is at a deeper recursion level, which PCRE cannot
4873 use.
4874
4875 To change the pattern so that matches all palindromic strings, not just
4876 those with an odd number of characters, it is tempting to change the
4877 pattern to this:
4878
4879 ^((.)(?1)\2|.?)$
4880
4881 Again, this works in Perl, but not in PCRE, and for the same reason.
4882 When a deeper recursion has matched a single character, it cannot be
4883 entered again in order to match an empty string. The solution is to
4884 separate the two cases, and write out the odd and even cases as alter-
4885 natives at the higher level:
4886
4887 ^(?:((.)(?1)\2|)|((.)(?3)\4|.))
4888
4889 If you want to match typical palindromic phrases, the pattern has to
4890 ignore all non-word characters, which can be done like this:
4891
4892 ^\W*+(?:((.)\W*+(?1)\W*+\2|)|((.)\W*+(?3)\W*+4|\W*+.\W*+))\W*+$
4893
4894 If run with the PCRE_CASELESS option, this pattern matches phrases such
4895 as "A man, a plan, a canal: Panama!" and it works well in both PCRE and
4896 Perl. Note the use of the possessive quantifier *+ to avoid backtrack-
4897 ing into sequences of non-word characters. Without this, PCRE takes a
4898 great deal longer (ten times or more) to match typical phrases, and
4899 Perl takes so long that you think it has gone into a loop.
4900
4901
4902 SUBPATTERNS AS SUBROUTINES
4903
4904 If the syntax for a recursive subpattern reference (either by number or
4905 by name) is used outside the parentheses to which it refers, it oper-
4906 ates like a subroutine in a programming language. The "called" subpat-
4907 tern may be defined before or after the reference. A numbered reference
4908 can be absolute or relative, as in these examples:
4909
4910 (...(absolute)...)...(?2)...
4911 (...(relative)...)...(?-1)...
4912 (...(?+1)...(relative)...
4913
4914 An earlier example pointed out that the pattern
4915
4916 (sens|respons)e and \1ibility
4917
4918 matches "sense and sensibility" and "response and responsibility", but
4919 not "sense and responsibility". If instead the pattern
4920
4921 (sens|respons)e and (?1)ibility
4922
4923 is used, it does match "sense and responsibility" as well as the other
4924 two strings. Another example is given in the discussion of DEFINE
4925 above.
4926
4927 Like recursive subpatterns, a "subroutine" call is always treated as an
4928 atomic group. That is, once it has matched some of the subject string,
4929 it is never re-entered, even if it contains untried alternatives and
4930 there is a subsequent matching failure.
4931
4932 When a subpattern is used as a subroutine, processing options such as
4933 case-independence are fixed when the subpattern is defined. They cannot
4934 be changed for different calls. For example, consider this pattern:
4935
4936 (abc)(?i:(?-1))
4937
4938 It matches "abcabc". It does not match "abcABC" because the change of
4939 processing option does not affect the called subpattern.
4940
4941
4942 ONIGURUMA SUBROUTINE SYNTAX
4943
4944 For compatibility with Oniguruma, the non-Perl syntax \g followed by a
4945 name or a number enclosed either in angle brackets or single quotes, is
4946 an alternative syntax for referencing a subpattern as a subroutine,
4947 possibly recursively. Here are two of the examples used above, rewrit-
4948 ten using this syntax:
4949
4950 (?<pn> \( ( (?>[^()]+) | \g<pn> )* \) )
4951 (sens|respons)e and \g'1'ibility
4952
4953 PCRE supports an extension to Oniguruma: if a number is preceded by a
4954 plus or a minus sign it is taken as a relative reference. For example:
4955
4956 (abc)(?i:\g<-1>)
4957
4958 Note that \g{...} (Perl syntax) and \g<...> (Oniguruma syntax) are not
4959 synonymous. The former is a back reference; the latter is a subroutine
4960 call.
4961
4962
4963 CALLOUTS
4964
4965 Perl has a feature whereby using the sequence (?{...}) causes arbitrary
4966 Perl code to be obeyed in the middle of matching a regular expression.
4967 This makes it possible, amongst other things, to extract different sub-
4968 strings that match the same pair of parentheses when there is a repeti-
4969 tion.
4970
4971 PCRE provides a similar feature, but of course it cannot obey arbitrary
4972 Perl code. The feature is called "callout". The caller of PCRE provides
4973 an external function by putting its entry point in the global variable
4974 pcre_callout. By default, this variable contains NULL, which disables
4975 all calling out.
4976
4977 Within a regular expression, (?C) indicates the points at which the
4978 external function is to be called. If you want to identify different
4979 callout points, you can put a number less than 256 after the letter C.
4980 The default value is zero. For example, this pattern has two callout
4981 points:
4982
4983 (?C1)abc(?C2)def
4984
4985 If the PCRE_AUTO_CALLOUT flag is passed to pcre_compile(), callouts are
4986 automatically installed before each item in the pattern. They are all
4987 numbered 255.
4988
4989 During matching, when PCRE reaches a callout point (and pcre_callout is
4990 set), the external function is called. It is provided with the number
4991 of the callout, the position in the pattern, and, optionally, one item
4992 of data originally supplied by the caller of pcre_exec(). The callout
4993 function may cause matching to proceed, to backtrack, or to fail alto-
4994 gether. A complete description of the interface to the callout function
4995 is given in the pcrecallout documentation.
4996
4997
4998 BACKTRACKING CONTROL
4999
5000 Perl 5.10 introduced a number of "Special Backtracking Control Verbs",
5001 which are described in the Perl documentation as "experimental and sub-
5002 ject to change or removal in a future version of Perl". It goes on to
5003 say: "Their usage in production code should be noted to avoid problems
5004 during upgrades." The same remarks apply to the PCRE features described
5005 in this section.
5006
5007 Since these verbs are specifically related to backtracking, most of
5008 them can be used only when the pattern is to be matched using
5009 pcre_exec(), which uses a backtracking algorithm. With the exception of
5010 (*FAIL), which behaves like a failing negative assertion, they cause an
5011 error if encountered by pcre_dfa_exec().
5012
5013 If any of these verbs are used in an assertion subpattern, their effect
5014 is confined to that subpattern; it does not extend to the surrounding
5015 pattern. Note that assertion subpatterns are processed as anchored at
5016 the point where they are tested.
5017
5018 The new verbs make use of what was previously invalid syntax: an open-
5019 ing parenthesis followed by an asterisk. In Perl, they are generally of
5020 the form (*VERB:ARG) but PCRE does not support the use of arguments, so
5021 its general form is just (*VERB). Any number of these verbs may occur
5022 in a pattern. There are two kinds:
5023
5024 Verbs that act immediately
5025
5026 The following verbs act as soon as they are encountered:
5027
5028 (*ACCEPT)
5029
5030 This verb causes the match to end successfully, skipping the remainder
5031 of the pattern. When inside a recursion, only the innermost pattern is
5032 ended immediately. If the (*ACCEPT) is inside capturing parentheses,
5033 the data so far is captured. (This feature was added to PCRE at release
5034 8.00.) For example:
5035
5036 A((?:A|B(*ACCEPT)|C)D)
5037
5038 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is cap-
5039 tured by the outer parentheses.
5040
5041 (*FAIL) or (*F)
5042
5043 This verb causes the match to fail, forcing backtracking to occur. It
5044 is equivalent to (?!) but easier to read. The Perl documentation notes
5045 that it is probably useful only when combined with (?{}) or (??{}).
5046 Those are, of course, Perl features that are not present in PCRE. The
5047 nearest equivalent is the callout feature, as for example in this pat-
5048 tern:
5049
5050 a+(?C)(*FAIL)
5051
5052 A match with the string "aaaa" always fails, but the callout is taken
5053 before each backtrack happens (in this example, 10 times).
5054
5055 Verbs that act after backtracking
5056
5057 The following verbs do nothing when they are encountered. Matching con-
5058 tinues with what follows, but if there is no subsequent match, a fail-
5059 ure is forced. The verbs differ in exactly what kind of failure
5060 occurs.
5061
5062 (*COMMIT)
5063
5064 This verb causes the whole match to fail outright if the rest of the
5065 pattern does not match. Even if the pattern is unanchored, no further
5066 attempts to find a match by advancing the start point take place. Once
5067 (*COMMIT) has been passed, pcre_exec() is committed to finding a match
5068 at the current starting point, or not at all. For example:
5069
5070 a+(*COMMIT)b
5071
5072 This matches "xxaab" but not "aacaab". It can be thought of as a kind
5073 of dynamic anchor, or "I've started, so I must finish."
5074
5075 (*PRUNE)
5076
5077 This verb causes the match to fail at the current position if the rest
5078 of the pattern does not match. If the pattern is unanchored, the normal
5079 "bumpalong" advance to the next starting character then happens. Back-
5080 tracking can occur as usual to the left of (*PRUNE), or when matching
5081 to the right of (*PRUNE), but if there is no match to the right, back-
5082 tracking cannot cross (*PRUNE). In simple cases, the use of (*PRUNE)
5083 is just an alternative to an atomic group or possessive quantifier, but
5084 there are some uses of (*PRUNE) that cannot be expressed in any other
5085 way.
5086
5087 (*SKIP)
5088
5089 This verb is like (*PRUNE), except that if the pattern is unanchored,
5090 the "bumpalong" advance is not to the next character, but to the posi-
5091 tion in the subject where (*SKIP) was encountered. (*SKIP) signifies
5092 that whatever text was matched leading up to it cannot be part of a
5093 successful match. Consider:
5094
5095 a+(*SKIP)b
5096
5097 If the subject is "aaaac...", after the first match attempt fails
5098 (starting at the first character in the string), the starting point
5099 skips on to start the next attempt at "c". Note that a possessive quan-
5100 tifer does not have the same effect in this example; although it would
5101 suppress backtracking during the first match attempt, the second
5102 attempt would start at the second character instead of skipping on to
5103 "c".
5104
5105 (*THEN)
5106
5107 This verb causes a skip to the next alternation if the rest of the pat-
5108 tern does not match. That is, it cancels pending backtracking, but only
5109 within the current alternation. Its name comes from the observation
5110 that it can be used for a pattern-based if-then-else block:
5111
5112 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
5113
5114 If the COND1 pattern matches, FOO is tried (and possibly further items
5115 after the end of the group if FOO succeeds); on failure the matcher
5116 skips to the second alternative and tries COND2, without backtracking
5117 into COND1. If (*THEN) is used outside of any alternation, it acts
5118 exactly like (*PRUNE).
5119
5120
5121 SEE ALSO
5122
5123 pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
5124
5125
5126 AUTHOR
5127
5128 Philip Hazel
5129 University Computing Service
5130 Cambridge CB2 3QH, England.
5131
5132
5133 REVISION
5134
5135 Last updated: 18 September 2009
5136 Copyright (c) 1997-2009 University of Cambridge.
5137 ------------------------------------------------------------------------------
5138
5139
5140 PCRESYNTAX(3) PCRESYNTAX(3)
5141
5142
5143 NAME
5144 PCRE - Perl-compatible regular expressions
5145
5146
5147 PCRE REGULAR EXPRESSION SYNTAX SUMMARY
5148
5149 The full syntax and semantics of the regular expressions that are sup-
5150 ported by PCRE are described in the pcrepattern documentation. This
5151 document contains just a quick-reference summary of the syntax.
5152
5153
5154 QUOTING
5155
5156 \x where x is non-alphanumeric is a literal x
5157 \Q...\E treat enclosed characters as literal
5158
5159
5160 CHARACTERS
5161
5162 \a alarm, that is, the BEL character (hex 07)
5163 \cx "control-x", where x is any character
5164 \e escape (hex 1B)
5165 \f formfeed (hex 0C)
5166 \n newline (hex 0A)
5167 \r carriage return (hex 0D)
5168 \t tab (hex 09)
5169 \ddd character with octal code ddd, or backreference
5170 \xhh character with hex code hh
5171 \x{hhh..} character with hex code hhh..
5172
5173
5174 CHARACTER TYPES
5175
5176 . any character except newline;
5177 in dotall mode, any character whatsoever
5178 \C one byte, even in UTF-8 mode (best avoided)
5179 \d a decimal digit
5180 \D a character that is not a decimal digit
5181 \h a horizontal whitespace character
5182 \H a character that is not a horizontal whitespace character
5183 \p{xx} a character with the xx property
5184 \P{xx} a character without the xx property
5185 \R a newline sequence
5186 \s a whitespace character
5187 \S a character that is not a whitespace character
5188 \v a vertical whitespace character
5189 \V a character that is not a vertical whitespace character
5190 \w a "word" character
5191 \W a "non-word" character
5192 \X an extended Unicode sequence
5193
5194 In PCRE, \d, \D, \s, \S, \w, and \W recognize only ASCII characters.
5195
5196
5197 GENERAL CATEGORY PROPERTY CODES FOR \p and \P
5198
5199 C Other
5200 Cc Control
5201 Cf Format
5202 Cn Unassigned
5203 Co Private use
5204 Cs Surrogate
5205
5206 L Letter
5207 Ll Lower case letter
5208 Lm Modifier letter
5209 Lo Other letter
5210 Lt Title case letter
5211 Lu Upper case letter
5212 L& Ll, Lu, or Lt
5213
5214 M Mark
5215 Mc Spacing mark
5216 Me Enclosing mark
5217 Mn Non-spacing mark
5218
5219 N Number
5220 Nd Decimal number
5221 Nl Letter number
5222 No Other number
5223
5224 P Punctuation
5225 Pc Connector punctuation
5226 Pd Dash punctuation
5227 Pe Close punctuation
5228 Pf Final punctuation
5229 Pi Initial punctuation
5230 Po Other punctuation
5231 Ps Open punctuation
5232
5233 S Symbol
5234 Sc Currency symbol
5235 Sk Modifier symbol
5236 Sm Mathematical symbol
5237 So Other symbol
5238
5239 Z Separator
5240 Zl Line separator
5241 Zp Paragraph separator
5242 Zs Space separator
5243
5244
5245 SCRIPT NAMES FOR \p AND \P
5246
5247 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
5248 Buhid, Canadian_Aboriginal, Carian, Cham, Cherokee, Common, Coptic, Cu-
5249 neiform, Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian,
5250 Glagolitic, Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo,
5251 Hebrew, Hiragana, Inherited, Kannada, Katakana, Kayah_Li, Kharoshthi,
5252 Khmer, Lao, Latin, Lepcha, Limbu, Linear_B, Lycian, Lydian, Malayalam,
5253 Mongolian, Myanmar, New_Tai_Lue, Nko, Ogham, Old_Italic, Old_Persian,
5254 Ol_Chiki, Oriya, Osmanya, Phags_Pa, Phoenician, Rejang, Runic, Saurash-
5255 tra, Shavian, Sinhala, Sudanese, Syloti_Nagri, Syriac, Tagalog, Tag-
5256 banwa, Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh,
5257 Ugaritic, Vai, Yi.
5258
5259
5260 CHARACTER CLASSES
5261
5262 [...] positive character class
5263 [^...] negative character class
5264 [x-y] range (can be used for hex characters)
5265 [[:xxx:]] positive POSIX named set
5266 [[:^xxx:]] negative POSIX named set
5267
5268 alnum alphanumeric
5269 alpha alphabetic
5270 ascii 0-127
5271 blank space or tab
5272 cntrl control character
5273 digit decimal digit
5274 graph printing, excluding space
5275 lower lower case letter
5276 print printing, including space
5277 punct printing, excluding alphanumeric
5278 space whitespace
5279 upper upper case letter
5280 word same as \w
5281 xdigit hexadecimal digit
5282
5283 In PCRE, POSIX character set names recognize only ASCII characters. You
5284 can use \Q...\E inside a character class.
5285
5286
5287 QUANTIFIERS
5288
5289 ? 0 or 1, greedy
5290 ?+ 0 or 1, possessive
5291 ?? 0 or 1, lazy
5292 * 0 or more, greedy
5293 *+ 0 or more, possessive
5294 *? 0 or more, lazy
5295 + 1 or more, greedy
5296 ++ 1 or more, possessive
5297 +? 1 or more, lazy
5298 {n} exactly n
5299 {n,m} at least n, no more than m, greedy
5300 {n,m}+ at least n, no more than m, possessive
5301 {n,m}? at least n, no more than m, lazy
5302 {n,} n or more, greedy
5303 {n,}+ n or more, possessive
5304 {n,}? n or more, lazy
5305
5306
5307 ANCHORS AND SIMPLE ASSERTIONS
5308
5309 \b word boundary (only ASCII letters recognized)
5310 \B not a word boundary
5311 ^ start of subject
5312 also after internal newline in multiline mode
5313 \A start of subject
5314 $ end of subject
5315 also before newline at end of subject
5316 also before internal newline in multiline mode
5317 \Z end of subject
5318 also before newline at end of subject
5319 \z end of subject
5320 \G first matching position in subject
5321
5322
5323 MATCH POINT RESET
5324
5325 \K reset start of match
5326
5327
5328 ALTERNATION
5329
5330 expr|expr|expr...
5331
5332
5333 CAPTURING
5334
5335 (...) capturing group
5336 (?<name>...) named capturing group (Perl)
5337 (?'name'...) named capturing group (Perl)
5338 (?P<name>...) named capturing group (Python)
5339 (?:...) non-capturing group
5340 (?|...) non-capturing group; reset group numbers for
5341 capturing groups in each alternative
5342
5343
5344 ATOMIC GROUPS
5345
5346 (?>...) atomic, non-capturing group
5347
5348
5349 COMMENT
5350
5351 (?#....) comment (not nestable)
5352
5353
5354 OPTION SETTING
5355
5356 (?i) caseless
5357 (?J) allow duplicate names
5358 (?m) multiline
5359 (?s) single line (dotall)
5360 (?U) default ungreedy (lazy)
5361 (?x) extended (ignore white space)
5362 (?-...) unset option(s)
5363
5364 The following is recognized only at the start of a pattern or after one
5365 of the newline-setting options with similar syntax:
5366
5367 (*UTF8) set UTF-8 mode
5368
5369
5370 LOOKAHEAD AND LOOKBEHIND ASSERTIONS
5371
5372 (?=...) positive look ahead
5373 (?!...) negative look ahead
5374 (?<=...) positive look behind
5375 (?<!...) negative look behind
5376
5377 Each top-level branch of a look behind must be of a fixed length.
5378
5379
5380 BACKREFERENCES
5381
5382 \n reference by number (can be ambiguous)
5383 \gn reference by number
5384 \g{n} reference by number
5385 \g{-n} relative reference by number
5386 \k<name> reference by name (Perl)
5387 \k'name' reference by name (Perl)
5388 \g{name} reference by name (Perl)
5389 \k{name} reference by name (.NET)
5390 (?P=name) reference by name (Python)
5391
5392
5393 SUBROUTINE REFERENCES (POSSIBLY RECURSIVE)
5394
5395 (?R) recurse whole pattern
5396 (?n) call subpattern by absolute number
5397 (?+n) call subpattern by relative number
5398 (?-n) call subpattern by relative number
5399 (?&name) call subpattern by name (Perl)
5400 (?P>name) call subpattern by name (Python)
5401 \g<name> call subpattern by name (Oniguruma)
5402 \g'name' call subpattern by name (Oniguruma)
5403 \g<n> call subpattern by absolute number (Oniguruma)
5404 \g'n' call subpattern by absolute number (Oniguruma)
5405 \g<+n> call subpattern by relative number (PCRE extension)
5406 \g'+n' call subpattern by relative number (PCRE extension)
5407 \g<-n> call subpattern by relative number (PCRE extension)
5408 \g'-n' call subpattern by relative number (PCRE extension)
5409
5410
5411 CONDITIONAL PATTERNS
5412
5413 (?(condition)yes-pattern)
5414 (?(condition)yes-pattern|no-pattern)
5415
5416 (?(n)... absolute reference condition
5417 (?(+n)... relative reference condition
5418 (?(-n)... relative reference condition
5419 (?(<name>)... named reference condition (Perl)
5420 (?('name')... named reference condition (Perl)
5421 (?(name)... named reference condition (PCRE)
5422 (?(R)... overall recursion condition
5423 (?(Rn)... specific group recursion condition
5424 (?(R&name)... specific recursion condition
5425 (?(DEFINE)... define subpattern for reference
5426 (?(assert)... assertion condition
5427
5428
5429 BACKTRACKING CONTROL
5430
5431 The following act immediately they are reached:
5432
5433 (*ACCEPT) force successful match
5434 (*FAIL) force backtrack; synonym (*F)
5435
5436 The following act only when a subsequent match failure causes a back-
5437 track to reach them. They all force a match failure, but they differ in
5438 what happens afterwards. Those that advance the start-of-match point do
5439 so only if the pattern is not anchored.
5440
5441 (*COMMIT) overall failure, no advance of starting point
5442 (*PRUNE) advance to next starting character
5443 (*SKIP) advance start to current matching position
5444 (*THEN) local failure, backtrack to next alternation
5445
5446
5447 NEWLINE CONVENTIONS
5448
5449 These are recognized only at the very start of the pattern or after a
5450 (*BSR_...) or (*UTF8) option.
5451
5452 (*CR) carriage return only
5453 (*LF) linefeed only
5454 (*CRLF) carriage return followed by linefeed
5455 (*ANYCRLF) all three of the above
5456 (*ANY) any Unicode newline sequence
5457
5458
5459 WHAT \R MATCHES
5460
5461 These are recognized only at the very start of the pattern or after a
5462 (*...) option that sets the newline convention or UTF-8 mode.
5463
5464 (*BSR_ANYCRLF) CR, LF, or CRLF
5465 (*BSR_UNICODE) any Unicode newline sequence
5466
5467
5468 CALLOUTS
5469
5470 (?C) callout
5471 (?Cn) callout with data n
5472
5473
5474 SEE ALSO
5475
5476 pcrepattern(3), pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
5477
5478
5479 AUTHOR
5480
5481 Philip Hazel
5482 University Computing Service
5483 Cambridge CB2 3QH, England.
5484
5485
5486 REVISION
5487
5488 Last updated: 11 April 2009
5489 Copyright (c) 1997-2009 University of Cambridge.
5490 ------------------------------------------------------------------------------
5491
5492
5493 PCREPARTIAL(3) PCREPARTIAL(3)
5494
5495
5496 NAME
5497 PCRE - Perl-compatible regular expressions
5498
5499
5500 PARTIAL MATCHING IN PCRE
5501
5502 In normal use of PCRE, if the subject string that is passed to
5503 pcre_exec() or pcre_dfa_exec() matches as far as it goes, but is too
5504 short to match the entire pattern, PCRE_ERROR_NOMATCH is returned.
5505 There are circumstances where it might be helpful to distinguish this
5506 case from other cases in which there is no match.
5507
5508 Consider, for example, an application where a human is required to type
5509 in data for a field with specific formatting requirements. An example
5510 might be a date in the form ddmmmyy, defined by this pattern:
5511
5512 ^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$
5513
5514 If the application sees the user's keystrokes one by one, and can check
5515 that what has been typed so far is potentially valid, it is able to
5516 raise an error as soon as a mistake is made, by beeping and not
5517 reflecting the character that has been typed, for example. This immedi-
5518 ate feedback is likely to be a better user interface than a check that
5519 is delayed until the entire string has been entered. Partial matching
5520 can also sometimes be useful when the subject string is very long and
5521 is not all available at once.
5522
5523 PCRE supports partial matching by means of the PCRE_PARTIAL_SOFT and
5524 PCRE_PARTIAL_HARD options, which can be set when calling pcre_exec() or
5525 pcre_dfa_exec(). For backwards compatibility, PCRE_PARTIAL is a synonym
5526 for PCRE_PARTIAL_SOFT. The essential difference between the two options
5527 is whether or not a partial match is preferred to an alternative com-
5528 plete match, though the details differ between the two matching func-
5529 tions. If both options are set, PCRE_PARTIAL_HARD takes precedence.
5530
5531 Setting a partial matching option disables one of PCRE's optimizations.
5532 PCRE remembers the last literal byte in a pattern, and abandons match-
5533 ing immediately if such a byte is not present in the subject string.
5534 This optimization cannot be used for a subject string that might match
5535 only partially.
5536
5537
5538 PARTIAL MATCHING USING pcre_exec()
5539
5540 A partial match occurs during a call to pcre_exec() whenever the end of
5541 the subject string is reached successfully, but matching cannot con-
5542 tinue because more characters are needed. However, at least one charac-
5543 ter must have been matched. (In other words, a partial match can never
5544 be an empty string.)
5545
5546 If PCRE_PARTIAL_SOFT is set, the partial match is remembered, but
5547 matching continues as normal, and other alternatives in the pattern are
5548 tried. If no complete match can be found, pcre_exec() returns
5549 PCRE_ERROR_PARTIAL instead of PCRE_ERROR_NOMATCH. If there are at least
5550 two slots in the offsets vector, the first of them is set to the offset
5551 of the earliest character that was inspected when the partial match was
5552 found. For convenience, the second offset points to the end of the
5553 string so that a substring can easily be extracted.
5554
5555 For the majority of patterns, the first offset identifies the start of
5556 the partially matched string. However, for patterns that contain look-
5557 behind assertions, or \K, or begin with \b or \B, earlier characters
5558 have been inspected while carrying out the match. For example:
5559
5560 /(?<=abc)123/
5561
5562 This pattern matches "123", but only if it is preceded by "abc". If the
5563 subject string is "xyzabc12", the offsets after a partial match are for
5564 the substring "abc12", because all these characters are needed if
5565 another match is tried with extra characters added.
5566
5567 If there is more than one partial match, the first one that was found
5568 provides the data that is returned. Consider this pattern:
5569
5570 /123\w+X|dogY/
5571
5572 If this is matched against the subject string "abc123dog", both alter-
5573 natives fail to match, but the end of the subject is reached during
5574 matching, so PCRE_ERROR_PARTIAL is returned instead of
5575 PCRE_ERROR_NOMATCH. The offsets are set to 3 and 9, identifying
5576 "123dog" as the first partial match that was found. (In this example,
5577 there are two partial matches, because "dog" on its own partially
5578 matches the second alternative.)
5579
5580 If PCRE_PARTIAL_HARD is set for pcre_exec(), it returns PCRE_ERROR_PAR-
5581 TIAL as soon as a partial match is found, without continuing to search
5582 for possible complete matches. The difference between the two options
5583 can be illustrated by a pattern such as:
5584
5585 /dog(sbody)?/
5586
5587 This matches either "dog" or "dogsbody", greedily (that is, it prefers
5588 the longer string if possible). If it is matched against the string
5589 "dog" with PCRE_PARTIAL_SOFT, it yields a complete match for "dog".
5590 However, if PCRE_PARTIAL_HARD is set, the result is PCRE_ERROR_PARTIAL.
5591 On the other hand, if the pattern is made ungreedy the result is dif-
5592 ferent:
5593
5594 /dog(sbody)??/
5595
5596 In this case the result is always a complete match because pcre_exec()
5597 finds that first, and it never continues after finding a match. It
5598 might be easier to follow this explanation by thinking of the two pat-
5599 terns like this:
5600
5601 /dog(sbody)?/ is the same as /dogsbody|dog/
5602 /dog(sbody)??/ is the same as /dog|dogsbody/
5603
5604 The second pattern will never match "dogsbody" when pcre_exec() is
5605 used, because it will always find the shorter match first.
5606
5607
5608 PARTIAL MATCHING USING pcre_dfa_exec()
5609
5610 The pcre_dfa_exec() function moves along the subject string character
5611 by character, without backtracking, searching for all possible matches
5612 simultaneously. If the end of the subject is reached before the end of
5613 the pattern, there is the possibility of a partial match, again pro-
5614 vided that at least one character has matched.
5615
5616 When PCRE_PARTIAL_SOFT is set, PCRE_ERROR_PARTIAL is returned only if
5617 there have been no complete matches. Otherwise, the complete matches
5618 are returned. However, if PCRE_PARTIAL_HARD is set, a partial match
5619 takes precedence over any complete matches. The portion of the string
5620 that was inspected when the longest partial match was found is set as
5621 the first matching string, provided there are at least two slots in the
5622 offsets vector.
5623
5624 Because pcre_dfa_exec() always searches for all possible matches, and
5625 there is no difference between greedy and ungreedy repetition, its be-
5626 haviour is different from pcre_exec when PCRE_PARTIAL_HARD is set. Con-
5627 sider the string "dog" matched against the ungreedy pattern shown
5628 above:
5629
5630 /dog(sbody)??/
5631
5632 Whereas pcre_exec() stops as soon as it finds the complete match for
5633 "dog", pcre_dfa_exec() also finds the partial match for "dogsbody", and
5634 so returns that when PCRE_PARTIAL_HARD is set.
5635
5636
5637 PARTIAL MATCHING AND WORD BOUNDARIES
5638
5639 If a pattern ends with one of sequences \w or \W, which test for word
5640 boundaries, partial matching with PCRE_PARTIAL_SOFT can give counter-
5641 intuitive results. Consider this pattern:
5642
5643 /\bcat\b/
5644
5645 This matches "cat", provided there is a word boundary at either end. If
5646 the subject string is "the cat", the comparison of the final "t" with a
5647 following character cannot take place, so a partial match is found.
5648 However, pcre_exec() carries on with normal matching, which matches \b
5649 at the end of the subject when the last character is a letter, thus
5650 finding a complete match. The result, therefore, is not PCRE_ERROR_PAR-
5651 TIAL. The same thing happens with pcre_dfa_exec(), because it also
5652 finds the complete match.
5653
5654 Using PCRE_PARTIAL_HARD in this case does yield PCRE_ERROR_PARTIAL,
5655 because then the partial match takes precedence.
5656
5657
5658 FORMERLY RESTRICTED PATTERNS
5659
5660 For releases of PCRE prior to 8.00, because of the way certain internal
5661 optimizations were implemented in the pcre_exec() function, the
5662 PCRE_PARTIAL option (predecessor of PCRE_PARTIAL_SOFT) could not be
5663 used with all patterns. From release 8.00 onwards, the restrictions no
5664 longer apply, and partial matching with pcre_exec() can be requested
5665 for any pattern.
5666
5667 Items that were formerly restricted were repeated single characters and
5668 repeated metasequences. If PCRE_PARTIAL was set for a pattern that did
5669 not conform to the restrictions, pcre_exec() returned the error code
5670 PCRE_ERROR_BADPARTIAL (-13). This error code is no longer in use. The
5671 PCRE_INFO_OKPARTIAL call to pcre_fullinfo() to find out if a compiled
5672 pattern can be used for partial matching now always returns 1.
5673
5674
5675 EXAMPLE OF PARTIAL MATCHING USING PCRETEST
5676
5677 If the escape sequence \P is present in a pcretest data line, the
5678 PCRE_PARTIAL_SOFT option is used for the match. Here is a run of
5679 pcretest that uses the date example quoted above:
5680
5681 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5682 data> 25jun04\P
5683 0: 25jun04
5684 1: jun
5685 data> 25dec3\P
5686 Partial match: 23dec3
5687 data> 3ju\P
5688 Partial match: 3ju
5689 data> 3juj\P
5690 No match
5691 data> j\P
5692 No match
5693
5694 The first data string is matched completely, so pcretest shows the
5695 matched substrings. The remaining four strings do not match the com-
5696 plete pattern, but the first two are partial matches. Similar output is
5697 obtained when pcre_dfa_exec() is used.
5698
5699 If the escape sequence \P is present more than once in a pcretest data
5700 line, the PCRE_PARTIAL_HARD option is set for the match.
5701
5702
5703 MULTI-SEGMENT MATCHING WITH pcre_dfa_exec()
5704
5705 When a partial match has been found using pcre_dfa_exec(), it is possi-
5706 ble to continue the match by providing additional subject data and
5707 calling pcre_dfa_exec() again with the same compiled regular expres-
5708 sion, this time setting the PCRE_DFA_RESTART option. You must pass the
5709 same working space as before, because this is where details of the pre-
5710 vious partial match are stored. Here is an example using pcretest,
5711 using the \R escape sequence to set the PCRE_DFA_RESTART option (\D
5712 specifies the use of pcre_dfa_exec()):
5713
5714 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5715 data> 23ja\P\D
5716 Partial match: 23ja
5717 data> n05\R\D
5718 0: n05
5719
5720 The first call has "23ja" as the subject, and requests partial match-
5721 ing; the second call has "n05" as the subject for the continued
5722 (restarted) match. Notice that when the match is complete, only the
5723 last part is shown; PCRE does not retain the previously partially-
5724 matched string. It is up to the calling program to do that if it needs
5725 to.
5726
5727 You can set the PCRE_PARTIAL_SOFT or PCRE_PARTIAL_HARD options with
5728 PCRE_DFA_RESTART to continue partial matching over multiple segments.
5729 This facility can be used to pass very long subject strings to
5730 pcre_dfa_exec().
5731
5732
5733 MULTI-SEGMENT MATCHING WITH pcre_exec()
5734
5735 From release 8.00, pcre_exec() can also be used to do multi-segment
5736 matching. Unlike pcre_dfa_exec(), it is not possible to restart the
5737 previous match with a new segment of data. Instead, new data must be
5738 added to the previous subject string, and the entire match re-run,
5739 starting from the point where the partial match occurred. Earlier data
5740 can be discarded. Consider an unanchored pattern that matches dates:
5741
5742 re> /\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d/
5743 data> The date is 23ja\P
5744 Partial match: 23ja
5745
5746 The this stage, an application could discard the text preceding "23ja",
5747 add on text from the next segment, and call pcre_exec() again. Unlike
5748 pcre_dfa_exec(), the entire matching string must always be available,
5749 and the complete matching process occurs for each call, so more memory
5750 and more processing time is needed.
5751
5752 Note: If the pattern contains lookbehind assertions, or \K, or starts
5753 with \b or \B, the string that is returned for a partial match will
5754 include characters that precede the partially matched string itself,
5755 because these must be retained when adding on more characters for a
5756 subsequent matching attempt.
5757
5758
5759 ISSUES WITH MULTI-SEGMENT MATCHING
5760
5761 Certain types of pattern may give problems with multi-segment matching,
5762 whichever matching function is used.
5763
5764 1. If the pattern contains tests for the beginning or end of a line,
5765 you need to pass the PCRE_NOTBOL or PCRE_NOTEOL options, as appropri-
5766 ate, when the subject string for any call does not contain the begin-
5767 ning or end of a line.
5768
5769 2. Lookbehind assertions at the start of a pattern are catered for in
5770 the offsets that are returned for a partial match. However, in theory,
5771 a lookbehind assertion later in the pattern could require even earlier
5772 characters to be inspected, and it might not have been reached when a
5773 partial match occurs. This is probably an extremely unlikely case; you
5774 could guard against it to a certain extent by always including extra
5775 characters at the start.
5776
5777 3. Matching a subject string that is split into multiple segments may
5778 not always produce exactly the same result as matching over one single
5779 long string, especially when PCRE_PARTIAL_SOFT is used. The section
5780 "Partial Matching and Word Boundaries" above describes an issue that
5781 arises if the pattern ends with \b or \B. Another kind of difference
5782 may occur when there are multiple matching possibilities, because a
5783 partial match result is given only when there are no completed matches.
5784 This means that as soon as the shortest match has been found, continua-
5785 tion to a new subject segment is no longer possible. Consider again
5786 this pcretest example:
5787
5788 re> /dog(sbody)?/
5789 data> dogsb\P
5790 0: dog
5791 data> do\P\D
5792 Partial match: do
5793 data> gsb\R\P\D
5794 0: g
5795 data> dogsbody\D
5796 0: dogsbody
5797 1: dog
5798
5799 The first data line passes the string "dogsb" to pcre_exec(), setting
5800 the PCRE_PARTIAL_SOFT option. Although the string is a partial match
5801 for "dogsbody", the result is not PCRE_ERROR_PARTIAL, because the
5802 shorter string "dog" is a complete match. Similarly, when the subject
5803 is presented to pcre_dfa_exec() in several parts ("do" and "gsb" being
5804 the first two) the match stops when "dog" has been found, and it is not
5805 possible to continue. On the other hand, if "dogsbody" is presented as
5806 a single string, pcre_dfa_exec() finds both matches.
5807
5808 Because of these problems, it is probably best to use PCRE_PARTIAL_HARD
5809 when matching multi-segment data. The example above then behaves dif-
5810 ferently:
5811
5812 re> /dog(sbody)?/
5813 data> dogsb\P\P
5814 Partial match: dogsb
5815 data> do\P\D
5816 Partial match: do
5817 data> gsb\R\P\P\D
5818 Partial match: gsb
5819
5820
5821 4. Patterns that contain alternatives at the top level which do not all
5822 start with the same pattern item may not work as expected when
5823 pcre_dfa_exec() is used. For example, consider this pattern:
5824
5825 1234|3789
5826
5827 If the first part of the subject is "ABC123", a partial match of the
5828 first alternative is found at offset 3. There is no partial match for
5829 the second alternative, because such a match does not start at the same
5830 point in the subject string. Attempting to continue with the string
5831 "7890" does not yield a match because only those alternatives that
5832 match at one point in the subject are remembered. The problem arises
5833 because the start of the second alternative matches within the first
5834 alternative. There is no problem with anchored patterns or patterns
5835 such as:
5836
5837 1234|ABCD
5838
5839 where no string can be a partial match for both alternatives. This is
5840 not a problem if pcre_exec() is used, because the entire match has to
5841 be rerun each time:
5842
5843 re> /1234|3789/
5844 data> ABC123\P
5845 Partial match: 123
5846 data> 1237890
5847 0: 3789
5848
5849
5850 AUTHOR
5851
5852 Philip Hazel
5853 University Computing Service
5854 Cambridge CB2 3QH, England.
5855
5856
5857 REVISION
5858
5859 Last updated: 05 September 2009
5860 Copyright (c) 1997-2009 University of Cambridge.
5861 ------------------------------------------------------------------------------
5862
5863
5864 PCREPRECOMPILE(3) PCREPRECOMPILE(3)
5865
5866
5867 NAME
5868 PCRE - Perl-compatible regular expressions
5869
5870
5871 SAVING AND RE-USING PRECOMPILED PCRE PATTERNS
5872
5873 If you are running an application that uses a large number of regular
5874 expression patterns, it may be useful to store them in a precompiled
5875 form instead of having to compile them every time the application is
5876 run. If you are not using any private character tables (see the
5877 pcre_maketables() documentation), this is relatively straightforward.
5878 If you are using private tables, it is a little bit more complicated.
5879
5880 If you save compiled patterns to a file, you can copy them to a differ-
5881 ent host and run them there. This works even if the new host has the
5882 opposite endianness to the one on which the patterns were compiled.
5883 There may be a small performance penalty, but it should be insignifi-
5884 cant. However, compiling regular expressions with one version of PCRE
5885 for use with a different version is not guaranteed to work and may
5886 cause crashes.
5887
5888
5889 SAVING A COMPILED PATTERN
5890 The value returned by pcre_compile() points to a single block of memory
5891 that holds the compiled pattern and associated data. You can find the
5892 length of this block in bytes by calling pcre_fullinfo() with an argu-
5893 ment of PCRE_INFO_SIZE. You can then save the data in any appropriate
5894 manner. Here is sample code that compiles a pattern and writes it to a
5895 file. It assumes that the variable fd refers to a file that is open for
5896 output:
5897
5898 int erroroffset, rc, size;
5899 char *error;
5900 pcre *re;
5901
5902 re = pcre_compile("my pattern", 0, &error, &erroroffset, NULL);
5903 if (re == NULL) { ... handle errors ... }
5904 rc = pcre_fullinfo(re, NULL, PCRE_INFO_SIZE, &size);
5905 if (rc < 0) { ... handle errors ... }
5906 rc = fwrite(re, 1, size, fd);
5907 if (rc != size) { ... handle errors ... }
5908
5909 In this example, the bytes that comprise the compiled pattern are
5910 copied exactly. Note that this is binary data that may contain any of
5911 the 256 possible byte values. On systems that make a distinction
5912 between binary and non-binary data, be sure that the file is opened for
5913 binary output.
5914
5915 If you want to write more than one pattern to a file, you will have to
5916 devise a way of separating them. For binary data, preceding each pat-
5917 tern with its length is probably the most straightforward approach.
5918 Another possibility is to write out the data in hexadecimal instead of
5919 binary, one pattern to a line.
5920
5921 Saving compiled patterns in a file is only one possible way of storing
5922 them for later use. They could equally well be saved in a database, or
5923 in the memory of some daemon process that passes them via sockets to
5924 the processes that want them.
5925
5926 If the pattern has been studied, it is also possible to save the study
5927 data in a similar way to the compiled pattern itself. When studying
5928 generates additional information, pcre_study() returns a pointer to a
5929 pcre_extra data block. Its format is defined in the section on matching
5930 a pattern in the pcreapi documentation. The study_data field points to
5931 the binary study data, and this is what you must save (not the
5932 pcre_extra block itself). The length of the study data can be obtained
5933 by calling pcre_fullinfo() with an argument of PCRE_INFO_STUDYSIZE.
5934 Remember to check that pcre_study() did return a non-NULL value before
5935 trying to save the study data.
5936
5937
5938 RE-USING A PRECOMPILED PATTERN
5939
5940 Re-using a precompiled pattern is straightforward. Having reloaded it
5941 into main memory, you pass its pointer to pcre_exec() or
5942 pcre_dfa_exec() in the usual way. This should work even on another
5943 host, and even if that host has the opposite endianness to the one
5944 where the pattern was compiled.
5945
5946 However, if you passed a pointer to custom character tables when the
5947 pattern was compiled (the tableptr argument of pcre_compile()), you
5948 must now pass a similar pointer to pcre_exec() or pcre_dfa_exec(),
5949 because the value saved with the compiled pattern will obviously be
5950 nonsense. A field in a pcre_extra() block is used to pass this data, as
5951 described in the section on matching a pattern in the pcreapi documen-
5952 tation.
5953
5954 If you did not provide custom character tables when the pattern was
5955 compiled, the pointer in the compiled pattern is NULL, which causes
5956 pcre_exec() to use PCRE's internal tables. Thus, you do not need to
5957 take any special action at run time in this case.
5958
5959 If you saved study data with the compiled pattern, you need to create
5960 your own pcre_extra data block and set the study_data field to point to
5961 the reloaded study data. You must also set the PCRE_EXTRA_STUDY_DATA
5962 bit in the flags field to indicate that study data is present. Then
5963 pass the pcre_extra block to pcre_exec() or pcre_dfa_exec() in the
5964 usual way.
5965
5966
5967 COMPATIBILITY WITH DIFFERENT PCRE RELEASES
5968
5969 In general, it is safest to recompile all saved patterns when you
5970 update to a new PCRE release, though not all updates actually require
5971 this. Recompiling is definitely needed for release 7.2.
5972
5973
5974 AUTHOR
5975
5976 Philip Hazel
5977 University Computing Service
5978 Cambridge CB2 3QH, England.
5979
5980
5981 REVISION
5982
5983 Last updated: 13 June 2007
5984 Copyright (c) 1997-2007 University of Cambridge.
5985 ------------------------------------------------------------------------------
5986
5987
5988 PCREPERFORM(3) PCREPERFORM(3)
5989
5990
5991 NAME
5992 PCRE - Perl-compatible regular expressions
5993
5994
5995 PCRE PERFORMANCE
5996
5997 Two aspects of performance are discussed below: memory usage and pro-
5998 cessing time. The way you express your pattern as a regular expression
5999 can affect both of them.
6000
6001
6002 MEMORY USAGE
6003
6004 Patterns are compiled by PCRE into a reasonably efficient byte code, so
6005 that most simple patterns do not use much memory. However, there is one
6006 case where memory usage can be unexpectedly large. When a parenthesized
6007 subpattern has a quantifier with a minimum greater than 1 and/or a lim-
6008 ited maximum, the whole subpattern is repeated in the compiled code.
6009 For example, the pattern
6010
6011 (abc|def){2,4}
6012
6013 is compiled as if it were
6014
6015 (abc|def)(abc|def)((abc|def)(abc|def)?)?
6016
6017 (Technical aside: It is done this way so that backtrack points within
6018 each of the repetitions can be independently maintained.)
6019
6020 For regular expressions whose quantifiers use only small numbers, this
6021 is not usually a problem. However, if the numbers are large, and par-
6022 ticularly if such repetitions are nested, the memory usage can become
6023 an embarrassment. For example, the very simple pattern
6024
6025 ((ab){1,1000}c){1,3}
6026
6027 uses 51K bytes when compiled. When PCRE is compiled with its default
6028 internal pointer size of two bytes, the size limit on a compiled pat-
6029 tern is 64K, and this is reached with the above pattern if the outer
6030 repetition is increased from 3 to 4. PCRE can be compiled to use larger
6031 internal pointers and thus handle larger compiled patterns, but it is
6032 better to try to rewrite your pattern to use less memory if you can.
6033
6034 One way of reducing the memory usage for such patterns is to make use
6035 of PCRE's "subroutine" facility. Re-writing the above pattern as
6036
6037 ((ab)(?2){0,999}c)(?1){0,2}
6038
6039 reduces the memory requirements to 18K, and indeed it remains under 20K
6040 even with the outer repetition increased to 100. However, this pattern
6041 is not exactly equivalent, because the "subroutine" calls are treated
6042 as atomic groups into which there can be no backtracking if there is a
6043 subsequent matching failure. Therefore, PCRE cannot do this kind of
6044 rewriting automatically. Furthermore, there is a noticeable loss of
6045 speed when executing the modified pattern. Nevertheless, if the atomic
6046 grouping is not a problem and the loss of speed is acceptable, this
6047 kind of rewriting will allow you to process patterns that PCRE cannot
6048 otherwise handle.
6049
6050
6051 PROCESSING TIME
6052
6053 Certain items in regular expression patterns are processed more effi-
6054 ciently than others. It is more efficient to use a character class like
6055 [aeiou] than a set of single-character alternatives such as
6056 (a|e|i|o|u). In general, the simplest construction that provides the
6057 required behaviour is usually the most efficient. Jeffrey Friedl's book
6058 contains a lot of useful general discussion about optimizing regular
6059 expressions for efficient performance. This document contains a few
6060 observations about PCRE.
6061
6062 Using Unicode character properties (the \p, \P, and \X escapes) is
6063 slow, because PCRE has to scan a structure that contains data for over
6064 fifteen thousand characters whenever it needs a character's property.
6065 If you can find an alternative pattern that does not use character
6066 properties, it will probably be faster.
6067
6068 When a pattern begins with .* not in parentheses, or in parentheses
6069 that are not the subject of a backreference, and the PCRE_DOTALL option
6070 is set, the pattern is implicitly anchored by PCRE, since it can match
6071 only at the start of a subject string. However, if PCRE_DOTALL is not
6072 set, PCRE cannot make this optimization, because the . metacharacter
6073 does not then match a newline, and if the subject string contains new-
6074 lines, the pattern may match from the character immediately following
6075 one of them instead of from the very start. For example, the pattern
6076
6077 .*second
6078
6079 matches the subject "