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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 654 - (show annotations)
Tue Aug 2 11:00:40 2011 UTC (8 years, 1 month ago) by ph10
File MIME type: text/plain
File size: 376481 byte(s)
Documentation and general text tidies in preparation for test 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. Some features that appeared in Python and PCRE
23 before they appeared in Perl are also available using the Python syn-
24 tax, there is some support for one or two .NET and Oniguruma syntax
25 items, and there is an option for requesting some minor changes that
26 give better JavaScript compatibility.
27
28 The current implementation of PCRE corresponds approximately with Perl
29 5.12, including support for UTF-8 encoded strings and Unicode general
30 category properties. However, UTF-8 and Unicode support has to be
31 explicitly enabled; it is not the default. The Unicode tables corre-
32 spond to Unicode release 6.0.0.
33
34 In addition to the Perl-compatible matching function, PCRE contains an
35 alternative function that matches the same compiled patterns in a dif-
36 ferent way. In certain circumstances, the alternative function has some
37 advantages. For a discussion of the two matching algorithms, see the
38 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 and
59 NON-UNIX-USE files 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 pcreperform discussion of performance issues
93 pcreposix the POSIX-compatible C API
94 pcreprecompile details of saving and re-using precompiled patterns
95 pcresample discussion of the pcredemo program
96 pcrestack discussion of stack usage
97 pcresyntax quick syntax reference
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 strings of 1-byte characters.
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 is given.
185 At compile time, the only additional information is the offset to the
186 first byte of the failing character. The runtime functions (pcre_exec()
187 and pcre_dfa_exec()), pass back this information as well as a more
188 detailed reason code if the caller has provided memory in which to do
189 this.
190
191 In some situations, you may already know that your strings are valid,
192 and therefore want to skip these checks in order to improve perfor-
193 mance. If you set the PCRE_NO_UTF8_CHECK flag at compile time or at run
194 time, PCRE assumes that the pattern or subject it is given (respec-
195 tively) contains only valid UTF-8 codes. In this case, it does not
196 diagnose an invalid UTF-8 string.
197
198 If you pass an invalid UTF-8 string when PCRE_NO_UTF8_CHECK is set,
199 what happens depends on why the string is invalid. If the string con-
200 forms to the "old" definition of UTF-8 (RFC 2279), it is processed as a
201 string of characters in the range 0 to 0x7FFFFFFF. In other words,
202 apart from the initial validity test, PCRE (when in UTF-8 mode) handles
203 strings according to the more liberal rules of RFC 2279. However, if
204 the string does not even conform to RFC 2279, the result is undefined.
205 Your program may crash.
206
207 If you want to process strings of values in the full range 0 to
208 0x7FFFFFFF, encoded in a UTF-8-like manner as per the old RFC, you can
209 set PCRE_NO_UTF8_CHECK to bypass the more restrictive test. However, in
210 this situation, you will have to apply your own validity check.
211
212 General comments about UTF-8 mode
213
214 1. An unbraced hexadecimal escape sequence (such as \xb3) matches a
215 two-byte UTF-8 character if the value is greater than 127.
216
217 2. Octal numbers up to \777 are recognized, and match two-byte UTF-8
218 characters for values greater than \177.
219
220 3. Repeat quantifiers apply to complete UTF-8 characters, not to indi-
221 vidual bytes, for example: \x{100}{3}.
222
223 4. The dot metacharacter matches one UTF-8 character instead of a sin-
224 gle byte.
225
226 5. The escape sequence \C can be used to match a single byte in UTF-8
227 mode, but its use can lead to some strange effects. This facility is
228 not available in the alternative matching function, pcre_dfa_exec().
229
230 6. The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly
231 test characters of any code value, but, by default, the characters that
232 PCRE recognizes as digits, spaces, or word characters remain the same
233 set as before, all with values less than 256. This remains true even
234 when PCRE is built to include Unicode property support, because to do
235 otherwise would slow down PCRE in many common cases. Note in particular
236 that this applies to \b and \B, because they are defined in terms of \w
237 and \W. If you really want to test for a wider sense of, say, "digit",
238 you can use explicit Unicode property tests such as \p{Nd}. Alterna-
239 tively, if you set the PCRE_UCP option, the way that the character
240 escapes work is changed so that Unicode properties are used to deter-
241 mine which characters match. There are more details in the section on
242 generic character types in the pcrepattern documentation.
243
244 7. Similarly, characters that match the POSIX named character classes
245 are all low-valued characters, unless the PCRE_UCP option is set.
246
247 8. However, the horizontal and vertical whitespace matching escapes
248 (\h, \H, \v, and \V) do match all the appropriate Unicode characters,
249 whether or not PCRE_UCP is set.
250
251 9. Case-insensitive matching applies only to characters whose values
252 are less than 128, unless PCRE is built with Unicode property support.
253 Even when Unicode property support is available, PCRE still uses its
254 own character tables when checking the case of low-valued characters,
255 so as not to degrade performance. The Unicode property information is
256 used only for characters with higher values. Furthermore, PCRE supports
257 case-insensitive matching only when there is a one-to-one mapping
258 between a letter's cases. There are a small number of many-to-one map-
259 pings in Unicode; these are not supported by PCRE.
260
261
262 AUTHOR
263
264 Philip Hazel
265 University Computing Service
266 Cambridge CB2 3QH, England.
267
268 Putting an actual email address here seems to have been a spam magnet,
269 so I've taken it away. If you want to email me, use my two initials,
270 followed by the two digits 10, at the domain cam.ac.uk.
271
272
273 REVISION
274
275 Last updated: 07 May 2011
276 Copyright (c) 1997-2011 University of Cambridge.
277 ------------------------------------------------------------------------------
278
279
280 PCREBUILD(3) PCREBUILD(3)
281
282
283 NAME
284 PCRE - Perl-compatible regular expressions
285
286
287 PCRE BUILD-TIME OPTIONS
288
289 This document describes the optional features of PCRE that can be
290 selected when the library is compiled. It assumes use of the configure
291 script, where the optional features are selected or deselected by pro-
292 viding options to configure before running the make command. However,
293 the same options can be selected in both Unix-like and non-Unix-like
294 environments using the GUI facility of cmake-gui if you are using CMake
295 instead of configure to build PCRE.
296
297 There is a lot more information about building PCRE in non-Unix-like
298 environments in the file called NON_UNIX_USE, which is part of the PCRE
299 distribution. You should consult this file as well as the README file
300 if you are building in a non-Unix-like environment.
301
302 The complete list of options for configure (which includes the standard
303 ones such as the selection of the installation directory) can be
304 obtained by running
305
306 ./configure --help
307
308 The following sections include descriptions of options whose names
309 begin with --enable or --disable. These settings specify changes to the
310 defaults for the configure command. Because of the way that configure
311 works, --enable and --disable always come in pairs, so the complemen-
312 tary option always exists as well, but as it specifies the default, it
313 is not described.
314
315
316 BUILDING SHARED AND STATIC LIBRARIES
317
318 The PCRE building process uses libtool to build both shared and static
319 Unix libraries by default. You can suppress one of these by adding one
320 of
321
322 --disable-shared
323 --disable-static
324
325 to the configure command, as required.
326
327
328 C++ SUPPORT
329
330 By default, the configure script will search for a C++ compiler and C++
331 header files. If it finds them, it automatically builds the C++ wrapper
332 library for PCRE. You can disable this by adding
333
334 --disable-cpp
335
336 to the configure command.
337
338
339 UTF-8 SUPPORT
340
341 To build PCRE with support for UTF-8 Unicode character strings, add
342
343 --enable-utf8
344
345 to the configure command. Of itself, this does not make PCRE treat
346 strings as UTF-8. As well as compiling PCRE with this option, you also
347 have have to set the PCRE_UTF8 option when you call the pcre_compile()
348 or pcre_compile2() functions.
349
350 If you set --enable-utf8 when compiling in an EBCDIC environment, PCRE
351 expects its input to be either ASCII or UTF-8 (depending on the runtime
352 option). It is not possible to support both EBCDIC and UTF-8 codes in
353 the same version of the library. Consequently, --enable-utf8 and
354 --enable-ebcdic are mutually exclusive.
355
356
357 UNICODE CHARACTER PROPERTY SUPPORT
358
359 UTF-8 support allows PCRE to process character values greater than 255
360 in the strings that it handles. On its own, however, it does not pro-
361 vide any facilities for accessing the properties of such characters. If
362 you want to be able to use the pattern escapes \P, \p, and \X, which
363 refer to Unicode character properties, you must add
364
365 --enable-unicode-properties
366
367 to the configure command. This implies UTF-8 support, even if you have
368 not explicitly requested it.
369
370 Including Unicode property support adds around 30K of tables to the
371 PCRE library. Only the general category properties such as Lu and Nd
372 are supported. Details are given in the pcrepattern documentation.
373
374
375 CODE VALUE OF NEWLINE
376
377 By default, PCRE interprets the linefeed (LF) character as indicating
378 the end of a line. This is the normal newline character on Unix-like
379 systems. You can compile PCRE to use carriage return (CR) instead, by
380 adding
381
382 --enable-newline-is-cr
383
384 to the configure command. There is also a --enable-newline-is-lf
385 option, which explicitly specifies linefeed as the newline character.
386
387 Alternatively, you can specify that line endings are to be indicated by
388 the two character sequence CRLF. If you want this, add
389
390 --enable-newline-is-crlf
391
392 to the configure command. There is a fourth option, specified by
393
394 --enable-newline-is-anycrlf
395
396 which causes PCRE to recognize any of the three sequences CR, LF, or
397 CRLF as indicating a line ending. Finally, a fifth option, specified by
398
399 --enable-newline-is-any
400
401 causes PCRE to recognize any Unicode newline sequence.
402
403 Whatever line ending convention is selected when PCRE is built can be
404 overridden when the library functions are called. At build time it is
405 conventional to use the standard for your operating system.
406
407
408 WHAT \R MATCHES
409
410 By default, the sequence \R in a pattern matches any Unicode newline
411 sequence, whatever has been selected as the line ending sequence. If
412 you specify
413
414 --enable-bsr-anycrlf
415
416 the default is changed so that \R matches only CR, LF, or CRLF. What-
417 ever is selected when PCRE is built can be overridden when the library
418 functions are called.
419
420
421 POSIX MALLOC USAGE
422
423 When PCRE is called through the POSIX interface (see the pcreposix doc-
424 umentation), additional working storage is required for holding the
425 pointers to capturing substrings, because PCRE requires three integers
426 per substring, whereas the POSIX interface provides only two. If the
427 number of expected substrings is small, the wrapper function uses space
428 on the stack, because this is faster than using malloc() for each call.
429 The default threshold above which the stack is no longer used is 10; it
430 can be changed by adding a setting such as
431
432 --with-posix-malloc-threshold=20
433
434 to the configure command.
435
436
437 HANDLING VERY LARGE PATTERNS
438
439 Within a compiled pattern, offset values are used to point from one
440 part to another (for example, from an opening parenthesis to an alter-
441 nation metacharacter). By default, two-byte values are used for these
442 offsets, leading to a maximum size for a compiled pattern of around
443 64K. This is sufficient to handle all but the most gigantic patterns.
444 Nevertheless, some people do want to process truyl enormous patterns,
445 so it is possible to compile PCRE to use three-byte or four-byte off-
446 sets by adding a setting such as
447
448 --with-link-size=3
449
450 to the configure command. The value given must be 2, 3, or 4. Using
451 longer offsets slows down the operation of PCRE because it has to load
452 additional bytes when handling them.
453
454
455 AVOIDING EXCESSIVE STACK USAGE
456
457 When matching with the pcre_exec() function, PCRE implements backtrack-
458 ing by making recursive calls to an internal function called match().
459 In environments where the size of the stack is limited, this can se-
460 verely limit PCRE's operation. (The Unix environment does not usually
461 suffer from this problem, but it may sometimes be necessary to increase
462 the maximum stack size. There is a discussion in the pcrestack docu-
463 mentation.) An alternative approach to recursion that uses memory from
464 the heap to remember data, instead of using recursive function calls,
465 has been implemented to work round the problem of limited stack size.
466 If you want to build a version of PCRE that works this way, add
467
468 --disable-stack-for-recursion
469
470 to the configure command. With this configuration, PCRE will use the
471 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
472 ment functions. By default these point to malloc() and free(), but you
473 can replace the pointers so that your own functions are used instead.
474
475 Separate functions are provided rather than using pcre_malloc and
476 pcre_free because the usage is very predictable: the block sizes
477 requested are always the same, and the blocks are always freed in
478 reverse order. A calling program might be able to implement optimized
479 functions that perform better than malloc() and free(). PCRE runs
480 noticeably more slowly when built in this way. This option affects only
481 the pcre_exec() function; it is not relevant for pcre_dfa_exec().
482
483
484 LIMITING PCRE RESOURCE USAGE
485
486 Internally, PCRE has a function called match(), which it calls repeat-
487 edly (sometimes recursively) when matching a pattern with the
488 pcre_exec() function. By controlling the maximum number of times this
489 function may be called during a single matching operation, a limit can
490 be placed on the resources used by a single call to pcre_exec(). The
491 limit can be changed at run time, as described in the pcreapi documen-
492 tation. The default is 10 million, but this can be changed by adding a
493 setting such as
494
495 --with-match-limit=500000
496
497 to the configure command. This setting has no effect on the
498 pcre_dfa_exec() matching function.
499
500 In some environments it is desirable to limit the depth of recursive
501 calls of match() more strictly than the total number of calls, in order
502 to restrict the maximum amount of stack (or heap, if --disable-stack-
503 for-recursion is specified) that is used. A second limit controls this;
504 it defaults to the value that is set for --with-match-limit, which
505 imposes no additional constraints. However, you can set a lower limit
506 by adding, for example,
507
508 --with-match-limit-recursion=10000
509
510 to the configure command. This value can also be overridden at run
511 time.
512
513
514 CREATING CHARACTER TABLES AT BUILD TIME
515
516 PCRE uses fixed tables for processing characters whose code values are
517 less than 256. By default, PCRE is built with a set of tables that are
518 distributed in the file pcre_chartables.c.dist. These tables are for
519 ASCII codes only. If you add
520
521 --enable-rebuild-chartables
522
523 to the configure command, the distributed tables are no longer used.
524 Instead, a program called dftables is compiled and run. This outputs
525 the source for new set of tables, created in the default locale of your
526 C runtime system. (This method of replacing the tables does not work if
527 you are cross compiling, because dftables is run on the local host. If
528 you need to create alternative tables when cross compiling, you will
529 have to do so "by hand".)
530
531
532 USING EBCDIC CODE
533
534 PCRE assumes by default that it will run in an environment where the
535 character code is ASCII (or Unicode, which is a superset of ASCII).
536 This is the case for most computer operating systems. PCRE can, how-
537 ever, be compiled to run in an EBCDIC environment by adding
538
539 --enable-ebcdic
540
541 to the configure command. This setting implies --enable-rebuild-charta-
542 bles. You should only use it if you know that you are in an EBCDIC
543 environment (for example, an IBM mainframe operating system). The
544 --enable-ebcdic option is incompatible with --enable-utf8.
545
546
547 PCREGREP OPTIONS FOR COMPRESSED FILE SUPPORT
548
549 By default, pcregrep reads all files as plain text. You can build it so
550 that it recognizes files whose names end in .gz or .bz2, and reads them
551 with libz or libbz2, respectively, by adding one or both of
552
553 --enable-pcregrep-libz
554 --enable-pcregrep-libbz2
555
556 to the configure command. These options naturally require that the rel-
557 evant libraries are installed on your system. Configuration will fail
558 if they are not.
559
560
561 PCREGREP BUFFER SIZE
562
563 pcregrep uses an internal buffer to hold a "window" on the file it is
564 scanning, in order to be able to output "before" and "after" lines when
565 it finds a match. The size of the buffer is controlled by a parameter
566 whose default value is 20K. The buffer itself is three times this size,
567 but because of the way it is used for holding "before" lines, the long-
568 est line that is guaranteed to be processable is the parameter size.
569 You can change the default parameter value by adding, for example,
570
571 --with-pcregrep-bufsize=50K
572
573 to the configure command. The caller of pcregrep can, however, override
574 this value by specifying a run-time option.
575
576
577 PCRETEST OPTION FOR LIBREADLINE SUPPORT
578
579 If you add
580
581 --enable-pcretest-libreadline
582
583 to the configure command, pcretest is linked with the libreadline
584 library, and when its input is from a terminal, it reads it using the
585 readline() function. This provides line-editing and history facilities.
586 Note that libreadline is GPL-licensed, so if you distribute a binary of
587 pcretest linked in this way, there may be licensing issues.
588
589 Setting this option causes the -lreadline option to be added to the
590 pcretest build. In many operating environments with a sytem-installed
591 libreadline this is sufficient. However, in some environments (e.g. if
592 an unmodified distribution version of readline is in use), some extra
593 configuration may be necessary. The INSTALL file for libreadline says
594 this:
595
596 "Readline uses the termcap functions, but does not link with the
597 termcap or curses library itself, allowing applications which link
598 with readline the to choose an appropriate library."
599
600 If your environment has not been set up so that an appropriate library
601 is automatically included, you may need to add something like
602
603 LIBS="-ncurses"
604
605 immediately before the configure command.
606
607
608 SEE ALSO
609
610 pcreapi(3), pcre_config(3).
611
612
613 AUTHOR
614
615 Philip Hazel
616 University Computing Service
617 Cambridge CB2 3QH, England.
618
619
620 REVISION
621
622 Last updated: 02 August 2011
623 Copyright (c) 1997-2011 University of Cambridge.
624 ------------------------------------------------------------------------------
625
626
627 PCREMATCHING(3) PCREMATCHING(3)
628
629
630 NAME
631 PCRE - Perl-compatible regular expressions
632
633
634 PCRE MATCHING ALGORITHMS
635
636 This document describes the two different algorithms that are available
637 in PCRE for matching a compiled regular expression against a given sub-
638 ject string. The "standard" algorithm is the one provided by the
639 pcre_exec() function. This works in the same was as Perl's matching
640 function, and provides a Perl-compatible matching operation.
641
642 An alternative algorithm is provided by the pcre_dfa_exec() function;
643 this operates in a different way, and is not Perl-compatible. It has
644 advantages and disadvantages compared with the standard algorithm, and
645 these are described below.
646
647 When there is only one possible way in which a given subject string can
648 match a pattern, the two algorithms give the same answer. A difference
649 arises, however, when there are multiple possibilities. For example, if
650 the pattern
651
652 ^<.*>
653
654 is matched against the string
655
656 <something> <something else> <something further>
657
658 there are three possible answers. The standard algorithm finds only one
659 of them, whereas the alternative algorithm finds all three.
660
661
662 REGULAR EXPRESSIONS AS TREES
663
664 The set of strings that are matched by a regular expression can be rep-
665 resented as a tree structure. An unlimited repetition in the pattern
666 makes the tree of infinite size, but it is still a tree. Matching the
667 pattern to a given subject string (from a given starting point) can be
668 thought of as a search of the tree. There are two ways to search a
669 tree: depth-first and breadth-first, and these correspond to the two
670 matching algorithms provided by PCRE.
671
672
673 THE STANDARD MATCHING ALGORITHM
674
675 In the terminology of Jeffrey Friedl's book "Mastering Regular Expres-
676 sions", the standard algorithm is an "NFA algorithm". It conducts a
677 depth-first search of the pattern tree. That is, it proceeds along a
678 single path through the tree, checking that the subject matches what is
679 required. When there is a mismatch, the algorithm tries any alterna-
680 tives at the current point, and if they all fail, it backs up to the
681 previous branch point in the tree, and tries the next alternative
682 branch at that level. This often involves backing up (moving to the
683 left) in the subject string as well. The order in which repetition
684 branches are tried is controlled by the greedy or ungreedy nature of
685 the quantifier.
686
687 If a leaf node is reached, a matching string has been found, and at
688 that point the algorithm stops. Thus, if there is more than one possi-
689 ble match, this algorithm returns the first one that it finds. Whether
690 this is the shortest, the longest, or some intermediate length depends
691 on the way the greedy and ungreedy repetition quantifiers are specified
692 in the pattern.
693
694 Because it ends up with a single path through the tree, it is rela-
695 tively straightforward for this algorithm to keep track of the sub-
696 strings that are matched by portions of the pattern in parentheses.
697 This provides support for capturing parentheses and back references.
698
699
700 THE ALTERNATIVE MATCHING ALGORITHM
701
702 This algorithm conducts a breadth-first search of the tree. Starting
703 from the first matching point in the subject, it scans the subject
704 string from left to right, once, character by character, and as it does
705 this, it remembers all the paths through the tree that represent valid
706 matches. In Friedl's terminology, this is a kind of "DFA algorithm",
707 though it is not implemented as a traditional finite state machine (it
708 keeps multiple states active simultaneously).
709
710 Although the general principle of this matching algorithm is that it
711 scans the subject string only once, without backtracking, there is one
712 exception: when a lookaround assertion is encountered, the characters
713 following or preceding the current point have to be independently
714 inspected.
715
716 The scan continues until either the end of the subject is reached, or
717 there are no more unterminated paths. At this point, terminated paths
718 represent the different matching possibilities (if there are none, the
719 match has failed). Thus, if there is more than one possible match,
720 this algorithm finds all of them, and in particular, it finds the long-
721 est. The matches are returned in decreasing order of length. There is
722 an option to stop the algorithm after the first match (which is neces-
723 sarily the shortest) is found.
724
725 Note that all the matches that are found start at the same point in the
726 subject. If the pattern
727
728 cat(er(pillar)?)?
729
730 is matched against the string "the caterpillar catchment", the result
731 will be the three strings "caterpillar", "cater", and "cat" that start
732 at the fifth character of the subject. The algorithm does not automati-
733 cally move on to find matches that start at later positions.
734
735 There are a number of features of PCRE regular expressions that are not
736 supported by the alternative matching algorithm. They are as follows:
737
738 1. Because the algorithm finds all possible matches, the greedy or
739 ungreedy nature of repetition quantifiers is not relevant. Greedy and
740 ungreedy quantifiers are treated in exactly the same way. However, pos-
741 sessive quantifiers can make a difference when what follows could also
742 match what is quantified, for example in a pattern like this:
743
744 ^a++\w!
745
746 This pattern matches "aaab!" but not "aaa!", which would be matched by
747 a non-possessive quantifier. Similarly, if an atomic group is present,
748 it is matched as if it were a standalone pattern at the current point,
749 and the longest match is then "locked in" for the rest of the overall
750 pattern.
751
752 2. When dealing with multiple paths through the tree simultaneously, it
753 is not straightforward to keep track of captured substrings for the
754 different matching possibilities, and PCRE's implementation of this
755 algorithm does not attempt to do this. This means that no captured sub-
756 strings are available.
757
758 3. Because no substrings are captured, back references within the pat-
759 tern are not supported, and cause errors if encountered.
760
761 4. For the same reason, conditional expressions that use a backrefer-
762 ence as the condition or test for a specific group recursion are not
763 supported.
764
765 5. Because many paths through the tree may be active, the \K escape
766 sequence, which resets the start of the match when encountered (but may
767 be on some paths and not on others), is not supported. It causes an
768 error if encountered.
769
770 6. Callouts are supported, but the value of the capture_top field is
771 always 1, and the value of the capture_last field is always -1.
772
773 7. The \C escape sequence, which (in the standard algorithm) matches a
774 single byte, even in UTF-8 mode, is not supported because the alterna-
775 tive algorithm moves through the subject string one character at a
776 time, for all active paths through the tree.
777
778 8. Except for (*FAIL), the backtracking control verbs such as (*PRUNE)
779 are not supported. (*FAIL) is supported, and behaves like a failing
780 negative assertion.
781
782
783 ADVANTAGES OF THE ALTERNATIVE ALGORITHM
784
785 Using the alternative matching algorithm provides the following advan-
786 tages:
787
788 1. All possible matches (at a single point in the subject) are automat-
789 ically found, and in particular, the longest match is found. To find
790 more than one match using the standard algorithm, you have to do kludgy
791 things with callouts.
792
793 2. Because the alternative algorithm scans the subject string just
794 once, and never needs to backtrack, it is possible to pass very long
795 subject strings to the matching function in several pieces, checking
796 for partial matching each time. Although it is possible to do multi-
797 segment matching using the standard algorithm (pcre_exec()), by retain-
798 ing partially matched substrings, it is more complicated. The pcrepar-
799 tial documentation gives details of partial matching and discusses
800 multi-segment matching.
801
802
803 DISADVANTAGES OF THE ALTERNATIVE ALGORITHM
804
805 The alternative algorithm suffers from a number of disadvantages:
806
807 1. It is substantially slower than the standard algorithm. This is
808 partly because it has to search for all possible matches, but is also
809 because it is less susceptible to optimization.
810
811 2. Capturing parentheses and back references are not supported.
812
813 3. Although atomic groups are supported, their use does not provide the
814 performance advantage that it does for the standard algorithm.
815
816
817 AUTHOR
818
819 Philip Hazel
820 University Computing Service
821 Cambridge CB2 3QH, England.
822
823
824 REVISION
825
826 Last updated: 17 November 2010
827 Copyright (c) 1997-2010 University of Cambridge.
828 ------------------------------------------------------------------------------
829
830
831 PCREAPI(3) PCREAPI(3)
832
833
834 NAME
835 PCRE - Perl-compatible regular expressions
836
837
838 PCRE NATIVE API
839
840 #include <pcre.h>
841
842 pcre *pcre_compile(const char *pattern, int options,
843 const char **errptr, int *erroffset,
844 const unsigned char *tableptr);
845
846 pcre *pcre_compile2(const char *pattern, int options,
847 int *errorcodeptr,
848 const char **errptr, int *erroffset,
849 const unsigned char *tableptr);
850
851 pcre_extra *pcre_study(const pcre *code, int options,
852 const char **errptr);
853
854 int pcre_exec(const pcre *code, const pcre_extra *extra,
855 const char *subject, int length, int startoffset,
856 int options, int *ovector, int ovecsize);
857
858 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
859 const char *subject, int length, int startoffset,
860 int options, int *ovector, int ovecsize,
861 int *workspace, int wscount);
862
863 int pcre_copy_named_substring(const pcre *code,
864 const char *subject, int *ovector,
865 int stringcount, const char *stringname,
866 char *buffer, int buffersize);
867
868 int pcre_copy_substring(const char *subject, int *ovector,
869 int stringcount, int stringnumber, char *buffer,
870 int buffersize);
871
872 int pcre_get_named_substring(const pcre *code,
873 const char *subject, int *ovector,
874 int stringcount, const char *stringname,
875 const char **stringptr);
876
877 int pcre_get_stringnumber(const pcre *code,
878 const char *name);
879
880 int pcre_get_stringtable_entries(const pcre *code,
881 const char *name, char **first, char **last);
882
883 int pcre_get_substring(const char *subject, int *ovector,
884 int stringcount, int stringnumber,
885 const char **stringptr);
886
887 int pcre_get_substring_list(const char *subject,
888 int *ovector, int stringcount, const char ***listptr);
889
890 void pcre_free_substring(const char *stringptr);
891
892 void pcre_free_substring_list(const char **stringptr);
893
894 const unsigned char *pcre_maketables(void);
895
896 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
897 int what, void *where);
898
899 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
900
901 int pcre_refcount(pcre *code, int adjust);
902
903 int pcre_config(int what, void *where);
904
905 char *pcre_version(void);
906
907 void *(*pcre_malloc)(size_t);
908
909 void (*pcre_free)(void *);
910
911 void *(*pcre_stack_malloc)(size_t);
912
913 void (*pcre_stack_free)(void *);
914
915 int (*pcre_callout)(pcre_callout_block *);
916
917
918 PCRE API OVERVIEW
919
920 PCRE has its own native API, which is described in this document. There
921 are also some wrapper functions that correspond to the POSIX regular
922 expression API. These are described in the pcreposix documentation.
923 Both of these APIs define a set of C function calls. A C++ wrapper is
924 distributed with PCRE. It is documented in the pcrecpp page.
925
926 The native API C function prototypes are defined in the header file
927 pcre.h, and on Unix systems the library itself is called libpcre. It
928 can normally be accessed by adding -lpcre to the command for linking an
929 application that uses PCRE. The header file defines the macros
930 PCRE_MAJOR and PCRE_MINOR to contain the major and minor release num-
931 bers for the library. Applications can use these to include support
932 for different releases of PCRE.
933
934 In a Windows environment, if you want to statically link an application
935 program against a non-dll pcre.a file, you must define PCRE_STATIC
936 before including pcre.h or pcrecpp.h, because otherwise the pcre_mal-
937 loc() and pcre_free() exported functions will be declared
938 __declspec(dllimport), with unwanted results.
939
940 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
941 pcre_exec() are used for compiling and matching regular expressions in
942 a Perl-compatible manner. A sample program that demonstrates the sim-
943 plest way of using them is provided in the file called pcredemo.c in
944 the PCRE source distribution. A listing of this program is given in the
945 pcredemo documentation, and the pcresample documentation describes how
946 to compile and run it.
947
948 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
949 ble, is also provided. This uses a different algorithm for the match-
950 ing. The alternative algorithm finds all possible matches (at a given
951 point in the subject), and scans the subject just once (unless there
952 are lookbehind assertions). However, this algorithm does not return
953 captured substrings. A description of the two matching algorithms and
954 their advantages and disadvantages is given in the pcrematching docu-
955 mentation.
956
957 In addition to the main compiling and matching functions, there are
958 convenience functions for extracting captured substrings from a subject
959 string that is matched by pcre_exec(). They are:
960
961 pcre_copy_substring()
962 pcre_copy_named_substring()
963 pcre_get_substring()
964 pcre_get_named_substring()
965 pcre_get_substring_list()
966 pcre_get_stringnumber()
967 pcre_get_stringtable_entries()
968
969 pcre_free_substring() and pcre_free_substring_list() are also provided,
970 to free the memory used for extracted strings.
971
972 The function pcre_maketables() is used to build a set of character
973 tables in the current locale for passing to pcre_compile(),
974 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
975 provided for specialist use. Most commonly, no special tables are
976 passed, in which case internal tables that are generated when PCRE is
977 built are used.
978
979 The function pcre_fullinfo() is used to find out information about a
980 compiled pattern; pcre_info() is an obsolete version that returns only
981 some of the available information, but is retained for backwards com-
982 patibility. The function pcre_version() returns a pointer to a string
983 containing the version of PCRE and its date of release.
984
985 The function pcre_refcount() maintains a reference count in a data
986 block containing a compiled pattern. This is provided for the benefit
987 of object-oriented applications.
988
989 The global variables pcre_malloc and pcre_free initially contain the
990 entry points of the standard malloc() and free() functions, respec-
991 tively. PCRE calls the memory management functions via these variables,
992 so a calling program can replace them if it wishes to intercept the
993 calls. This should be done before calling any PCRE functions.
994
995 The global variables pcre_stack_malloc and pcre_stack_free are also
996 indirections to memory management functions. These special functions
997 are used only when PCRE is compiled to use the heap for remembering
998 data, instead of recursive function calls, when running the pcre_exec()
999 function. See the pcrebuild documentation for details of how to do
1000 this. It is a non-standard way of building PCRE, for use in environ-
1001 ments that have limited stacks. Because of the greater use of memory
1002 management, it runs more slowly. Separate functions are provided so
1003 that special-purpose external code can be used for this case. When
1004 used, these functions are always called in a stack-like manner (last
1005 obtained, first freed), and always for memory blocks of the same size.
1006 There is a discussion about PCRE's stack usage in the pcrestack docu-
1007 mentation.
1008
1009 The global variable pcre_callout initially contains NULL. It can be set
1010 by the caller to a "callout" function, which PCRE will then call at
1011 specified points during a matching operation. Details are given in the
1012 pcrecallout documentation.
1013
1014
1015 NEWLINES
1016
1017 PCRE supports five different conventions for indicating line breaks in
1018 strings: a single CR (carriage return) character, a single LF (line-
1019 feed) character, the two-character sequence CRLF, any of the three pre-
1020 ceding, or any Unicode newline sequence. The Unicode newline sequences
1021 are the three just mentioned, plus the single characters VT (vertical
1022 tab, U+000B), FF (formfeed, U+000C), NEL (next line, U+0085), LS (line
1023 separator, U+2028), and PS (paragraph separator, U+2029).
1024
1025 Each of the first three conventions is used by at least one operating
1026 system as its standard newline sequence. When PCRE is built, a default
1027 can be specified. The default default is LF, which is the Unix stan-
1028 dard. When PCRE is run, the default can be overridden, either when a
1029 pattern is compiled, or when it is matched.
1030
1031 At compile time, the newline convention can be specified by the options
1032 argument of pcre_compile(), or it can be specified by special text at
1033 the start of the pattern itself; this overrides any other settings. See
1034 the pcrepattern page for details of the special character sequences.
1035
1036 In the PCRE documentation the word "newline" is used to mean "the char-
1037 acter or pair of characters that indicate a line break". The choice of
1038 newline convention affects the handling of the dot, circumflex, and
1039 dollar metacharacters, the handling of #-comments in /x mode, and, when
1040 CRLF is a recognized line ending sequence, the match position advance-
1041 ment for a non-anchored pattern. There is more detail about this in the
1042 section on pcre_exec() options below.
1043
1044 The choice of newline convention does not affect the interpretation of
1045 the \n or \r escape sequences, nor does it affect what \R matches,
1046 which is controlled in a similar way, but by separate options.
1047
1048
1049 MULTITHREADING
1050
1051 The PCRE functions can be used in multi-threading applications, with
1052 the proviso that the memory management functions pointed to by
1053 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
1054 callout function pointed to by pcre_callout, are shared by all threads.
1055
1056 The compiled form of a regular expression is not altered during match-
1057 ing, so the same compiled pattern can safely be used by several threads
1058 at once.
1059
1060
1061 SAVING PRECOMPILED PATTERNS FOR LATER USE
1062
1063 The compiled form of a regular expression can be saved and re-used at a
1064 later time, possibly by a different program, and even on a host other
1065 than the one on which it was compiled. Details are given in the
1066 pcreprecompile documentation. However, compiling a regular expression
1067 with one version of PCRE for use with a different version is not guar-
1068 anteed to work and may cause crashes.
1069
1070
1071 CHECKING BUILD-TIME OPTIONS
1072
1073 int pcre_config(int what, void *where);
1074
1075 The function pcre_config() makes it possible for a PCRE client to dis-
1076 cover which optional features have been compiled into the PCRE library.
1077 The pcrebuild documentation has more details about these optional fea-
1078 tures.
1079
1080 The first argument for pcre_config() is an integer, specifying which
1081 information is required; the second argument is a pointer to a variable
1082 into which the information is placed. The following information is
1083 available:
1084
1085 PCRE_CONFIG_UTF8
1086
1087 The output is an integer that is set to one if UTF-8 support is avail-
1088 able; otherwise it is set to zero.
1089
1090 PCRE_CONFIG_UNICODE_PROPERTIES
1091
1092 The output is an integer that is set to one if support for Unicode
1093 character properties is available; otherwise it is set to zero.
1094
1095 PCRE_CONFIG_NEWLINE
1096
1097 The output is an integer whose value specifies the default character
1098 sequence that is recognized as meaning "newline". The four values that
1099 are supported are: 10 for LF, 13 for CR, 3338 for CRLF, -2 for ANYCRLF,
1100 and -1 for ANY. Though they are derived from ASCII, the same values
1101 are returned in EBCDIC environments. The default should normally corre-
1102 spond to the standard sequence for your operating system.
1103
1104 PCRE_CONFIG_BSR
1105
1106 The output is an integer whose value indicates what character sequences
1107 the \R escape sequence matches by default. A value of 0 means that \R
1108 matches any Unicode line ending sequence; a value of 1 means that \R
1109 matches only CR, LF, or CRLF. The default can be overridden when a pat-
1110 tern is compiled or matched.
1111
1112 PCRE_CONFIG_LINK_SIZE
1113
1114 The output is an integer that contains the number of bytes used for
1115 internal linkage in compiled regular expressions. The value is 2, 3, or
1116 4. Larger values allow larger regular expressions to be compiled, at
1117 the expense of slower matching. The default value of 2 is sufficient
1118 for all but the most massive patterns, since it allows the compiled
1119 pattern to be up to 64K in size.
1120
1121 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
1122
1123 The output is an integer that contains the threshold above which the
1124 POSIX interface uses malloc() for output vectors. Further details are
1125 given in the pcreposix documentation.
1126
1127 PCRE_CONFIG_MATCH_LIMIT
1128
1129 The output is a long integer that gives the default limit for the num-
1130 ber of internal matching function calls in a pcre_exec() execution.
1131 Further details are given with pcre_exec() below.
1132
1133 PCRE_CONFIG_MATCH_LIMIT_RECURSION
1134
1135 The output is a long integer that gives the default limit for the depth
1136 of recursion when calling the internal matching function in a
1137 pcre_exec() execution. Further details are given with pcre_exec()
1138 below.
1139
1140 PCRE_CONFIG_STACKRECURSE
1141
1142 The output is an integer that is set to one if internal recursion when
1143 running pcre_exec() is implemented by recursive function calls that use
1144 the stack to remember their state. This is the usual way that PCRE is
1145 compiled. The output is zero if PCRE was compiled to use blocks of data
1146 on the heap instead of recursive function calls. In this case,
1147 pcre_stack_malloc and pcre_stack_free are called to manage memory
1148 blocks on the heap, thus avoiding the use of the stack.
1149
1150
1151 COMPILING A PATTERN
1152
1153 pcre *pcre_compile(const char *pattern, int options,
1154 const char **errptr, int *erroffset,
1155 const unsigned char *tableptr);
1156
1157 pcre *pcre_compile2(const char *pattern, int options,
1158 int *errorcodeptr,
1159 const char **errptr, int *erroffset,
1160 const unsigned char *tableptr);
1161
1162 Either of the functions pcre_compile() or pcre_compile2() can be called
1163 to compile a pattern into an internal form. The only difference between
1164 the two interfaces is that pcre_compile2() has an additional argument,
1165 errorcodeptr, via which a numerical error code can be returned. To
1166 avoid too much repetition, we refer just to pcre_compile() below, but
1167 the information applies equally to pcre_compile2().
1168
1169 The pattern is a C string terminated by a binary zero, and is passed in
1170 the pattern argument. A pointer to a single block of memory that is
1171 obtained via pcre_malloc is returned. This contains the compiled code
1172 and related data. The pcre type is defined for the returned block; this
1173 is a typedef for a structure whose contents are not externally defined.
1174 It is up to the caller to free the memory (via pcre_free) when it is no
1175 longer required.
1176
1177 Although the compiled code of a PCRE regex is relocatable, that is, it
1178 does not depend on memory location, the complete pcre data block is not
1179 fully relocatable, because it may contain a copy of the tableptr argu-
1180 ment, which is an address (see below).
1181
1182 The options argument contains various bit settings that affect the com-
1183 pilation. It should be zero if no options are required. The available
1184 options are described below. Some of them (in particular, those that
1185 are compatible with Perl, but some others as well) can also be set and
1186 unset from within the pattern (see the detailed description in the
1187 pcrepattern documentation). For those options that can be different in
1188 different parts of the pattern, the contents of the options argument
1189 specifies their settings at the start of compilation and execution. The
1190 PCRE_ANCHORED, PCRE_BSR_xxx, PCRE_NEWLINE_xxx, PCRE_NO_UTF8_CHECK, and
1191 PCRE_NO_START_OPT options can be set at the time of matching as well as
1192 at compile time.
1193
1194 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
1195 if compilation of a pattern fails, pcre_compile() returns NULL, and
1196 sets the variable pointed to by errptr to point to a textual error mes-
1197 sage. This is a static string that is part of the library. You must not
1198 try to free it. Normally, the offset from the start of the pattern to
1199 the byte that was being processed when the error was discovered is
1200 placed in the variable pointed to by erroffset, which must not be NULL
1201 (if it is, an immediate error is given). However, for an invalid UTF-8
1202 string, the offset is that of the first byte of the failing character.
1203 Also, some errors are not detected until checks are carried out when
1204 the whole pattern has been scanned; in these cases the offset passed
1205 back is the length of the pattern.
1206
1207 Note that the offset is in bytes, not characters, even in UTF-8 mode.
1208 It may sometimes point into the middle of a UTF-8 character.
1209
1210 If pcre_compile2() is used instead of pcre_compile(), and the error-
1211 codeptr argument is not NULL, a non-zero error code number is returned
1212 via this argument in the event of an error. This is in addition to the
1213 textual error message. Error codes and messages are listed below.
1214
1215 If the final argument, tableptr, is NULL, PCRE uses a default set of
1216 character tables that are built when PCRE is compiled, using the
1217 default C locale. Otherwise, tableptr must be an address that is the
1218 result of a call to pcre_maketables(). This value is stored with the
1219 compiled pattern, and used again by pcre_exec(), unless another table
1220 pointer is passed to it. For more discussion, see the section on locale
1221 support below.
1222
1223 This code fragment shows a typical straightforward call to pcre_com-
1224 pile():
1225
1226 pcre *re;
1227 const char *error;
1228 int erroffset;
1229 re = pcre_compile(
1230 "^A.*Z", /* the pattern */
1231 0, /* default options */
1232 &error, /* for error message */
1233 &erroffset, /* for error offset */
1234 NULL); /* use default character tables */
1235
1236 The following names for option bits are defined in the pcre.h header
1237 file:
1238
1239 PCRE_ANCHORED
1240
1241 If this bit is set, the pattern is forced to be "anchored", that is, it
1242 is constrained to match only at the first matching point in the string
1243 that is being searched (the "subject string"). This effect can also be
1244 achieved by appropriate constructs in the pattern itself, which is the
1245 only way to do it in Perl.
1246
1247 PCRE_AUTO_CALLOUT
1248
1249 If this bit is set, pcre_compile() automatically inserts callout items,
1250 all with number 255, before each pattern item. For discussion of the
1251 callout facility, see the pcrecallout documentation.
1252
1253 PCRE_BSR_ANYCRLF
1254 PCRE_BSR_UNICODE
1255
1256 These options (which are mutually exclusive) control what the \R escape
1257 sequence matches. The choice is either to match only CR, LF, or CRLF,
1258 or to match any Unicode newline sequence. The default is specified when
1259 PCRE is built. It can be overridden from within the pattern, or by set-
1260 ting an option when a compiled pattern is matched.
1261
1262 PCRE_CASELESS
1263
1264 If this bit is set, letters in the pattern match both upper and lower
1265 case letters. It is equivalent to Perl's /i option, and it can be
1266 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
1267 always understands the concept of case for characters whose values are
1268 less than 128, so caseless matching is always possible. For characters
1269 with higher values, the concept of case is supported if PCRE is com-
1270 piled with Unicode property support, but not otherwise. If you want to
1271 use caseless matching for characters 128 and above, you must ensure
1272 that PCRE is compiled with Unicode property support as well as with
1273 UTF-8 support.
1274
1275 PCRE_DOLLAR_ENDONLY
1276
1277 If this bit is set, a dollar metacharacter in the pattern matches only
1278 at the end of the subject string. Without this option, a dollar also
1279 matches immediately before a newline at the end of the string (but not
1280 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
1281 if PCRE_MULTILINE is set. There is no equivalent to this option in
1282 Perl, and no way to set it within a pattern.
1283
1284 PCRE_DOTALL
1285
1286 If this bit is set, a dot metacharacter in the pattern matches a char-
1287 acter of any value, including one that indicates a newline. However, it
1288 only ever matches one character, even if newlines are coded as CRLF.
1289 Without this option, a dot does not match when the current position is
1290 at a newline. This option is equivalent to Perl's /s option, and it can
1291 be changed within a pattern by a (?s) option setting. A negative class
1292 such as [^a] always matches newline characters, independent of the set-
1293 ting of this option.
1294
1295 PCRE_DUPNAMES
1296
1297 If this bit is set, names used to identify capturing subpatterns need
1298 not be unique. This can be helpful for certain types of pattern when it
1299 is known that only one instance of the named subpattern can ever be
1300 matched. There are more details of named subpatterns below; see also
1301 the pcrepattern documentation.
1302
1303 PCRE_EXTENDED
1304
1305 If this bit is set, whitespace data characters in the pattern are
1306 totally ignored except when escaped or inside a character class. White-
1307 space does not include the VT character (code 11). In addition, charac-
1308 ters between an unescaped # outside a character class and the next new-
1309 line, inclusive, are also ignored. This is equivalent to Perl's /x
1310 option, and it can be changed within a pattern by a (?x) option set-
1311 ting.
1312
1313 Which characters are interpreted as newlines is controlled by the
1314 options passed to pcre_compile() or by a special sequence at the start
1315 of the pattern, as described in the section entitled "Newline conven-
1316 tions" in the pcrepattern documentation. Note that the end of this type
1317 of comment is a literal newline sequence in the pattern; escape
1318 sequences that happen to represent a newline do not count.
1319
1320 This option makes it possible to include comments inside complicated
1321 patterns. Note, however, that this applies only to data characters.
1322 Whitespace characters may never appear within special character
1323 sequences in a pattern, for example within the sequence (?( that intro-
1324 duces a conditional subpattern.
1325
1326 PCRE_EXTRA
1327
1328 This option was invented in order to turn on additional functionality
1329 of PCRE that is incompatible with Perl, but it is currently of very
1330 little use. When set, any backslash in a pattern that is followed by a
1331 letter that has no special meaning causes an error, thus reserving
1332 these combinations for future expansion. By default, as in Perl, a
1333 backslash followed by a letter with no special meaning is treated as a
1334 literal. (Perl can, however, be persuaded to give an error for this, by
1335 running it with the -w option.) There are at present no other features
1336 controlled by this option. It can also be set by a (?X) option setting
1337 within a pattern.
1338
1339 PCRE_FIRSTLINE
1340
1341 If this option is set, an unanchored pattern is required to match
1342 before or at the first newline in the subject string, though the
1343 matched text may continue over the newline.
1344
1345 PCRE_JAVASCRIPT_COMPAT
1346
1347 If this option is set, PCRE's behaviour is changed in some ways so that
1348 it is compatible with JavaScript rather than Perl. The changes are as
1349 follows:
1350
1351 (1) A lone closing square bracket in a pattern causes a compile-time
1352 error, because this is illegal in JavaScript (by default it is treated
1353 as a data character). Thus, the pattern AB]CD becomes illegal when this
1354 option is set.
1355
1356 (2) At run time, a back reference to an unset subpattern group matches
1357 an empty string (by default this causes the current matching alterna-
1358 tive to fail). A pattern such as (\1)(a) succeeds when this option is
1359 set (assuming it can find an "a" in the subject), whereas it fails by
1360 default, for Perl compatibility.
1361
1362 PCRE_MULTILINE
1363
1364 By default, PCRE treats the subject string as consisting of a single
1365 line of characters (even if it actually contains newlines). The "start
1366 of line" metacharacter (^) matches only at the start of the string,
1367 while the "end of line" metacharacter ($) matches only at the end of
1368 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1369 is set). This is the same as Perl.
1370
1371 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1372 constructs match immediately following or immediately before internal
1373 newlines in the subject string, respectively, as well as at the very
1374 start and end. This is equivalent to Perl's /m option, and it can be
1375 changed within a pattern by a (?m) option setting. If there are no new-
1376 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1377 setting PCRE_MULTILINE has no effect.
1378
1379 PCRE_NEWLINE_CR
1380 PCRE_NEWLINE_LF
1381 PCRE_NEWLINE_CRLF
1382 PCRE_NEWLINE_ANYCRLF
1383 PCRE_NEWLINE_ANY
1384
1385 These options override the default newline definition that was chosen
1386 when PCRE was built. Setting the first or the second specifies that a
1387 newline is indicated by a single character (CR or LF, respectively).
1388 Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
1389 two-character CRLF sequence. Setting PCRE_NEWLINE_ANYCRLF specifies
1390 that any of the three preceding sequences should be recognized. Setting
1391 PCRE_NEWLINE_ANY specifies that any Unicode newline sequence should be
1392 recognized. The Unicode newline sequences are the three just mentioned,
1393 plus the single characters VT (vertical tab, U+000B), FF (formfeed,
1394 U+000C), NEL (next line, U+0085), LS (line separator, U+2028), and PS
1395 (paragraph separator, U+2029). The last two are recognized only in
1396 UTF-8 mode.
1397
1398 The newline setting in the options word uses three bits that are
1399 treated as a number, giving eight possibilities. Currently only six are
1400 used (default plus the five values above). This means that if you set
1401 more than one newline option, the combination may or may not be sensi-
1402 ble. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equivalent to
1403 PCRE_NEWLINE_CRLF, but other combinations may yield unused numbers and
1404 cause an error.
1405
1406 The only time that a line break in a pattern is specially recognized
1407 when compiling is when PCRE_EXTENDED is set. CR and LF are whitespace
1408 characters, and so are ignored in this mode. Also, an unescaped # out-
1409 side a character class indicates a comment that lasts until after the
1410 next line break sequence. In other circumstances, line break sequences
1411 in patterns are treated as literal data.
1412
1413 The newline option that is set at compile time becomes the default that
1414 is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1415
1416 PCRE_NO_AUTO_CAPTURE
1417
1418 If this option is set, it disables the use of numbered capturing paren-
1419 theses in the pattern. Any opening parenthesis that is not followed by
1420 ? behaves as if it were followed by ?: but named parentheses can still
1421 be used for capturing (and they acquire numbers in the usual way).
1422 There is no equivalent of this option in Perl.
1423
1424 NO_START_OPTIMIZE
1425
1426 This is an option that acts at matching time; that is, it is really an
1427 option for pcre_exec() or pcre_dfa_exec(). If it is set at compile
1428 time, it is remembered with the compiled pattern and assumed at match-
1429 ing time. For details see the discussion of PCRE_NO_START_OPTIMIZE
1430 below.
1431
1432 PCRE_UCP
1433
1434 This option changes the way PCRE processes \B, \b, \D, \d, \S, \s, \W,
1435 \w, and some of the POSIX character classes. By default, only ASCII
1436 characters are recognized, but if PCRE_UCP is set, Unicode properties
1437 are used instead to classify characters. More details are given in the
1438 section on generic character types in the pcrepattern page. If you set
1439 PCRE_UCP, matching one of the items it affects takes much longer. The
1440 option is available only if PCRE has been compiled with Unicode prop-
1441 erty support.
1442
1443 PCRE_UNGREEDY
1444
1445 This option inverts the "greediness" of the quantifiers so that they
1446 are not greedy by default, but become greedy if followed by "?". It is
1447 not compatible with Perl. It can also be set by a (?U) option setting
1448 within the pattern.
1449
1450 PCRE_UTF8
1451
1452 This option causes PCRE to regard both the pattern and the subject as
1453 strings of UTF-8 characters instead of single-byte character strings.
1454 However, it is available only when PCRE is built to include UTF-8 sup-
1455 port. If not, the use of this option provokes an error. Details of how
1456 this option changes the behaviour of PCRE are given in the section on
1457 UTF-8 support in the main pcre page.
1458
1459 PCRE_NO_UTF8_CHECK
1460
1461 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1462 automatically checked. There is a discussion about the validity of
1463 UTF-8 strings in the main pcre page. If an invalid UTF-8 sequence of
1464 bytes is found, pcre_compile() returns an error. If you already know
1465 that your pattern is valid, and you want to skip this check for perfor-
1466 mance reasons, you can set the PCRE_NO_UTF8_CHECK option. When it is
1467 set, the effect of passing an invalid UTF-8 string as a pattern is
1468 undefined. It may cause your program to crash. Note that this option
1469 can also be passed to pcre_exec() and pcre_dfa_exec(), to suppress the
1470 UTF-8 validity checking of subject strings.
1471
1472
1473 COMPILATION ERROR CODES
1474
1475 The following table lists the error codes than may be returned by
1476 pcre_compile2(), along with the error messages that may be returned by
1477 both compiling functions. As PCRE has developed, some error codes have
1478 fallen out of use. To avoid confusion, they have not been re-used.
1479
1480 0 no error
1481 1 \ at end of pattern
1482 2 \c at end of pattern
1483 3 unrecognized character follows \
1484 4 numbers out of order in {} quantifier
1485 5 number too big in {} quantifier
1486 6 missing terminating ] for character class
1487 7 invalid escape sequence in character class
1488 8 range out of order in character class
1489 9 nothing to repeat
1490 10 [this code is not in use]
1491 11 internal error: unexpected repeat
1492 12 unrecognized character after (? or (?-
1493 13 POSIX named classes are supported only within a class
1494 14 missing )
1495 15 reference to non-existent subpattern
1496 16 erroffset passed as NULL
1497 17 unknown option bit(s) set
1498 18 missing ) after comment
1499 19 [this code is not in use]
1500 20 regular expression is too large
1501 21 failed to get memory
1502 22 unmatched parentheses
1503 23 internal error: code overflow
1504 24 unrecognized character after (?<
1505 25 lookbehind assertion is not fixed length
1506 26 malformed number or name after (?(
1507 27 conditional group contains more than two branches
1508 28 assertion expected after (?(
1509 29 (?R or (?[+-]digits must be followed by )
1510 30 unknown POSIX class name
1511 31 POSIX collating elements are not supported
1512 32 this version of PCRE is not compiled with PCRE_UTF8 support
1513 33 [this code is not in use]
1514 34 character value in \x{...} sequence is too large
1515 35 invalid condition (?(0)
1516 36 \C not allowed in lookbehind assertion
1517 37 PCRE does not support \L, \l, \N, \U, or \u
1518 38 number after (?C is > 255
1519 39 closing ) for (?C expected
1520 40 recursive call could loop indefinitely
1521 41 unrecognized character after (?P
1522 42 syntax error in subpattern name (missing terminator)
1523 43 two named subpatterns have the same name
1524 44 invalid UTF-8 string
1525 45 support for \P, \p, and \X has not been compiled
1526 46 malformed \P or \p sequence
1527 47 unknown property name after \P or \p
1528 48 subpattern name is too long (maximum 32 characters)
1529 49 too many named subpatterns (maximum 10000)
1530 50 [this code is not in use]
1531 51 octal value is greater than \377 (not in UTF-8 mode)
1532 52 internal error: overran compiling workspace
1533 53 internal error: previously-checked referenced subpattern
1534 not found
1535 54 DEFINE group contains more than one branch
1536 55 repeating a DEFINE group is not allowed
1537 56 inconsistent NEWLINE options
1538 57 \g is not followed by a braced, angle-bracketed, or quoted
1539 name/number or by a plain number
1540 58 a numbered reference must not be zero
1541 59 an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)
1542 60 (*VERB) not recognized
1543 61 number is too big
1544 62 subpattern name expected
1545 63 digit expected after (?+
1546 64 ] is an invalid data character in JavaScript compatibility mode
1547 65 different names for subpatterns of the same number are
1548 not allowed
1549 66 (*MARK) must have an argument
1550 67 this version of PCRE is not compiled with PCRE_UCP support
1551
1552 The numbers 32 and 10000 in errors 48 and 49 are defaults; different
1553 values may be used if the limits were changed when PCRE was built.
1554
1555
1556 STUDYING A PATTERN
1557
1558 pcre_extra *pcre_study(const pcre *code, int options
1559 const char **errptr);
1560
1561 If a compiled pattern is going to be used several times, it is worth
1562 spending more time analyzing it in order to speed up the time taken for
1563 matching. The function pcre_study() takes a pointer to a compiled pat-
1564 tern as its first argument. If studying the pattern produces additional
1565 information that will help speed up matching, pcre_study() returns a
1566 pointer to a pcre_extra block, in which the study_data field points to
1567 the results of the study.
1568
1569 The returned value from pcre_study() can be passed directly to
1570 pcre_exec() or pcre_dfa_exec(). However, a pcre_extra block also con-
1571 tains other fields that can be set by the caller before the block is
1572 passed; these are described below in the section on matching a pattern.
1573
1574 If studying the pattern does not produce any useful information,
1575 pcre_study() returns NULL. In that circumstance, if the calling program
1576 wants to pass any of the other fields to pcre_exec() or
1577 pcre_dfa_exec(), it must set up its own pcre_extra block.
1578
1579 The second argument of pcre_study() contains option bits. At present,
1580 no options are defined, and this argument should always be zero.
1581
1582 The third argument for pcre_study() is a pointer for an error message.
1583 If studying succeeds (even if no data is returned), the variable it
1584 points to is set to NULL. Otherwise it is set to point to a textual
1585 error message. This is a static string that is part of the library. You
1586 must not try to free it. You should test the error pointer for NULL
1587 after calling pcre_study(), to be sure that it has run successfully.
1588
1589 This is a typical call to pcre_study():
1590
1591 pcre_extra *pe;
1592 pe = pcre_study(
1593 re, /* result of pcre_compile() */
1594 0, /* no options exist */
1595 &error); /* set to NULL or points to a message */
1596
1597 Studying a pattern does two things: first, a lower bound for the length
1598 of subject string that is needed to match the pattern is computed. This
1599 does not mean that there are any strings of that length that match, but
1600 it does guarantee that no shorter strings match. The value is used by
1601 pcre_exec() and pcre_dfa_exec() to avoid wasting time by trying to
1602 match strings that are shorter than the lower bound. You can find out
1603 the value in a calling program via the pcre_fullinfo() function.
1604
1605 Studying a pattern is also useful for non-anchored patterns that do not
1606 have a single fixed starting character. A bitmap of possible starting
1607 bytes is created. This speeds up finding a position in the subject at
1608 which to start matching.
1609
1610 The two optimizations just described can be disabled by setting the
1611 PCRE_NO_START_OPTIMIZE option when calling pcre_exec() or
1612 pcre_dfa_exec(). You might want to do this if your pattern contains
1613 callouts or (*MARK), and you want to make use of these facilities in
1614 cases where matching fails. See the discussion of PCRE_NO_START_OPTI-
1615 MIZE below.
1616
1617
1618 LOCALE SUPPORT
1619
1620 PCRE handles caseless matching, and determines whether characters are
1621 letters, digits, or whatever, by reference to a set of tables, indexed
1622 by character value. When running in UTF-8 mode, this applies only to
1623 characters with codes less than 128. By default, higher-valued codes
1624 never match escapes such as \w or \d, but they can be tested with \p if
1625 PCRE is built with Unicode character property support. Alternatively,
1626 the PCRE_UCP option can be set at compile time; this causes \w and
1627 friends to use Unicode property support instead of built-in tables. The
1628 use of locales with Unicode is discouraged. If you are handling charac-
1629 ters with codes greater than 128, you should either use UTF-8 and Uni-
1630 code, or use locales, but not try to mix the two.
1631
1632 PCRE contains an internal set of tables that are used when the final
1633 argument of pcre_compile() is NULL. These are sufficient for many
1634 applications. Normally, the internal tables recognize only ASCII char-
1635 acters. However, when PCRE is built, it is possible to cause the inter-
1636 nal tables to be rebuilt in the default "C" locale of the local system,
1637 which may cause them to be different.
1638
1639 The internal tables can always be overridden by tables supplied by the
1640 application that calls PCRE. These may be created in a different locale
1641 from the default. As more and more applications change to using Uni-
1642 code, the need for this locale support is expected to die away.
1643
1644 External tables are built by calling the pcre_maketables() function,
1645 which has no arguments, in the relevant locale. The result can then be
1646 passed to pcre_compile() or pcre_exec() as often as necessary. For
1647 example, to build and use tables that are appropriate for the French
1648 locale (where accented characters with values greater than 128 are
1649 treated as letters), the following code could be used:
1650
1651 setlocale(LC_CTYPE, "fr_FR");
1652 tables = pcre_maketables();
1653 re = pcre_compile(..., tables);
1654
1655 The locale name "fr_FR" is used on Linux and other Unix-like systems;
1656 if you are using Windows, the name for the French locale is "french".
1657
1658 When pcre_maketables() runs, the tables are built in memory that is
1659 obtained via pcre_malloc. It is the caller's responsibility to ensure
1660 that the memory containing the tables remains available for as long as
1661 it is needed.
1662
1663 The pointer that is passed to pcre_compile() is saved with the compiled
1664 pattern, and the same tables are used via this pointer by pcre_study()
1665 and normally also by pcre_exec(). Thus, by default, for any single pat-
1666 tern, compilation, studying and matching all happen in the same locale,
1667 but different patterns can be compiled in different locales.
1668
1669 It is possible to pass a table pointer or NULL (indicating the use of
1670 the internal tables) to pcre_exec(). Although not intended for this
1671 purpose, this facility could be used to match a pattern in a different
1672 locale from the one in which it was compiled. Passing table pointers at
1673 run time is discussed below in the section on matching a pattern.
1674
1675
1676 INFORMATION ABOUT A PATTERN
1677
1678 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1679 int what, void *where);
1680
1681 The pcre_fullinfo() function returns information about a compiled pat-
1682 tern. It replaces the obsolete pcre_info() function, which is neverthe-
1683 less retained for backwards compability (and is documented below).
1684
1685 The first argument for pcre_fullinfo() is a pointer to the compiled
1686 pattern. The second argument is the result of pcre_study(), or NULL if
1687 the pattern was not studied. The third argument specifies which piece
1688 of information is required, and the fourth argument is a pointer to a
1689 variable to receive the data. The yield of the function is zero for
1690 success, or one of the following negative numbers:
1691
1692 PCRE_ERROR_NULL the argument code was NULL
1693 the argument where was NULL
1694 PCRE_ERROR_BADMAGIC the "magic number" was not found
1695 PCRE_ERROR_BADOPTION the value of what was invalid
1696
1697 The "magic number" is placed at the start of each compiled pattern as
1698 an simple check against passing an arbitrary memory pointer. Here is a
1699 typical call of pcre_fullinfo(), to obtain the length of the compiled
1700 pattern:
1701
1702 int rc;
1703 size_t length;
1704 rc = pcre_fullinfo(
1705 re, /* result of pcre_compile() */
1706 pe, /* result of pcre_study(), or NULL */
1707 PCRE_INFO_SIZE, /* what is required */
1708 &length); /* where to put the data */
1709
1710 The possible values for the third argument are defined in pcre.h, and
1711 are as follows:
1712
1713 PCRE_INFO_BACKREFMAX
1714
1715 Return the number of the highest back reference in the pattern. The
1716 fourth argument should point to an int variable. Zero is returned if
1717 there are no back references.
1718
1719 PCRE_INFO_CAPTURECOUNT
1720
1721 Return the number of capturing subpatterns in the pattern. The fourth
1722 argument should point to an int variable.
1723
1724 PCRE_INFO_DEFAULT_TABLES
1725
1726 Return a pointer to the internal default character tables within PCRE.
1727 The fourth argument should point to an unsigned char * variable. This
1728 information call is provided for internal use by the pcre_study() func-
1729 tion. External callers can cause PCRE to use its internal tables by
1730 passing a NULL table pointer.
1731
1732 PCRE_INFO_FIRSTBYTE
1733
1734 Return information about the first byte of any matched string, for a
1735 non-anchored pattern. The fourth argument should point to an int vari-
1736 able. (This option used to be called PCRE_INFO_FIRSTCHAR; the old name
1737 is still recognized for backwards compatibility.)
1738
1739 If there is a fixed first byte, for example, from a pattern such as
1740 (cat|cow|coyote), its value is returned. Otherwise, if either
1741
1742 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
1743 branch starts with "^", or
1744
1745 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
1746 set (if it were set, the pattern would be anchored),
1747
1748 -1 is returned, indicating that the pattern matches only at the start
1749 of a subject string or after any newline within the string. Otherwise
1750 -2 is returned. For anchored patterns, -2 is returned.
1751
1752 PCRE_INFO_FIRSTTABLE
1753
1754 If the pattern was studied, and this resulted in the construction of a
1755 256-bit table indicating a fixed set of bytes for the first byte in any
1756 matching string, a pointer to the table is returned. Otherwise NULL is
1757 returned. The fourth argument should point to an unsigned char * vari-
1758 able.
1759
1760 PCRE_INFO_HASCRORLF
1761
1762 Return 1 if the pattern contains any explicit matches for CR or LF
1763 characters, otherwise 0. The fourth argument should point to an int
1764 variable. An explicit match is either a literal CR or LF character, or
1765 \r or \n.
1766
1767 PCRE_INFO_JCHANGED
1768
1769 Return 1 if the (?J) or (?-J) option setting is used in the pattern,
1770 otherwise 0. The fourth argument should point to an int variable. (?J)
1771 and (?-J) set and unset the local PCRE_DUPNAMES option, respectively.
1772
1773 PCRE_INFO_LASTLITERAL
1774
1775 Return the value of the rightmost literal byte that must exist in any
1776 matched string, other than at its start, if such a byte has been
1777 recorded. The fourth argument should point to an int variable. If there
1778 is no such byte, -1 is returned. For anchored patterns, a last literal
1779 byte is recorded only if it follows something of variable length. For
1780 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
1781 /^a\dz\d/ the returned value is -1.
1782
1783 PCRE_INFO_MINLENGTH
1784
1785 If the pattern was studied and a minimum length for matching subject
1786 strings was computed, its value is returned. Otherwise the returned
1787 value is -1. The value is a number of characters, not bytes (this may
1788 be relevant in UTF-8 mode). The fourth argument should point to an int
1789 variable. A non-negative value is a lower bound to the length of any
1790 matching string. There may not be any strings of that length that do
1791 actually match, but every string that does match is at least that long.
1792
1793 PCRE_INFO_NAMECOUNT
1794 PCRE_INFO_NAMEENTRYSIZE
1795 PCRE_INFO_NAMETABLE
1796
1797 PCRE supports the use of named as well as numbered capturing parenthe-
1798 ses. The names are just an additional way of identifying the parenthe-
1799 ses, which still acquire numbers. Several convenience functions such as
1800 pcre_get_named_substring() are provided for extracting captured sub-
1801 strings by name. It is also possible to extract the data directly, by
1802 first converting the name to a number in order to access the correct
1803 pointers in the output vector (described with pcre_exec() below). To do
1804 the conversion, you need to use the name-to-number map, which is
1805 described by these three values.
1806
1807 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
1808 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
1809 of each entry; both of these return an int value. The entry size
1810 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
1811 a pointer to the first entry of the table (a pointer to char). The
1812 first two bytes of each entry are the number of the capturing parenthe-
1813 sis, most significant byte first. The rest of the entry is the corre-
1814 sponding name, zero terminated.
1815
1816 The names are in alphabetical order. Duplicate names may appear if (?|
1817 is used to create multiple groups with the same number, as described in
1818 the section on duplicate subpattern numbers in the pcrepattern page.
1819 Duplicate names for subpatterns with different numbers are permitted
1820 only if PCRE_DUPNAMES is set. In all cases of duplicate names, they
1821 appear in the table in the order in which they were found in the pat-
1822 tern. In the absence of (?| this is the order of increasing number;
1823 when (?| is used this is not necessarily the case because later subpat-
1824 terns may have lower numbers.
1825
1826 As a simple example of the name/number table, consider the following
1827 pattern (assume PCRE_EXTENDED is set, so white space - including new-
1828 lines - is ignored):
1829
1830 (?<date> (?<year>(\d\d)?\d\d) -
1831 (?<month>\d\d) - (?<day>\d\d) )
1832
1833 There are four named subpatterns, so the table has four entries, and
1834 each entry in the table is eight bytes long. The table is as follows,
1835 with non-printing bytes shows in hexadecimal, and undefined bytes shown
1836 as ??:
1837
1838 00 01 d a t e 00 ??
1839 00 05 d a y 00 ?? ??
1840 00 04 m o n t h 00
1841 00 02 y e a r 00 ??
1842
1843 When writing code to extract data from named subpatterns using the
1844 name-to-number map, remember that the length of the entries is likely
1845 to be different for each compiled pattern.
1846
1847 PCRE_INFO_OKPARTIAL
1848
1849 Return 1 if the pattern can be used for partial matching with
1850 pcre_exec(), otherwise 0. The fourth argument should point to an int
1851 variable. From release 8.00, this always returns 1, because the
1852 restrictions that previously applied to partial matching have been
1853 lifted. The pcrepartial documentation gives details of partial match-
1854 ing.
1855
1856 PCRE_INFO_OPTIONS
1857
1858 Return a copy of the options with which the pattern was compiled. The
1859 fourth argument should point to an unsigned long int variable. These
1860 option bits are those specified in the call to pcre_compile(), modified
1861 by any top-level option settings at the start of the pattern itself. In
1862 other words, they are the options that will be in force when matching
1863 starts. For example, if the pattern /(?im)abc(?-i)d/ is compiled with
1864 the PCRE_EXTENDED option, the result is PCRE_CASELESS, PCRE_MULTILINE,
1865 and PCRE_EXTENDED.
1866
1867 A pattern is automatically anchored by PCRE if all of its top-level
1868 alternatives begin with one of the following:
1869
1870 ^ unless PCRE_MULTILINE is set
1871 \A always
1872 \G always
1873 .* if PCRE_DOTALL is set and there are no back
1874 references to the subpattern in which .* appears
1875
1876 For such patterns, the PCRE_ANCHORED bit is set in the options returned
1877 by pcre_fullinfo().
1878
1879 PCRE_INFO_SIZE
1880
1881 Return the size of the compiled pattern, that is, the value that was
1882 passed as the argument to pcre_malloc() when PCRE was getting memory in
1883 which to place the compiled data. The fourth argument should point to a
1884 size_t variable.
1885
1886 PCRE_INFO_STUDYSIZE
1887
1888 Return the size of the data block pointed to by the study_data field in
1889 a pcre_extra block. That is, it is the value that was passed to
1890 pcre_malloc() when PCRE was getting memory into which to place the data
1891 created by pcre_study(). If pcre_extra is NULL, or there is no study
1892 data, zero is returned. The fourth argument should point to a size_t
1893 variable.
1894
1895
1896 OBSOLETE INFO FUNCTION
1897
1898 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
1899
1900 The pcre_info() function is now obsolete because its interface is too
1901 restrictive to return all the available data about a compiled pattern.
1902 New programs should use pcre_fullinfo() instead. The yield of
1903 pcre_info() is the number of capturing subpatterns, or one of the fol-
1904 lowing negative numbers:
1905
1906 PCRE_ERROR_NULL the argument code was NULL
1907 PCRE_ERROR_BADMAGIC the "magic number" was not found
1908
1909 If the optptr argument is not NULL, a copy of the options with which
1910 the pattern was compiled is placed in the integer it points to (see
1911 PCRE_INFO_OPTIONS above).
1912
1913 If the pattern is not anchored and the firstcharptr argument is not
1914 NULL, it is used to pass back information about the first character of
1915 any matched string (see PCRE_INFO_FIRSTBYTE above).
1916
1917
1918 REFERENCE COUNTS
1919
1920 int pcre_refcount(pcre *code, int adjust);
1921
1922 The pcre_refcount() function is used to maintain a reference count in
1923 the data block that contains a compiled pattern. It is provided for the
1924 benefit of applications that operate in an object-oriented manner,
1925 where different parts of the application may be using the same compiled
1926 pattern, but you want to free the block when they are all done.
1927
1928 When a pattern is compiled, the reference count field is initialized to
1929 zero. It is changed only by calling this function, whose action is to
1930 add the adjust value (which may be positive or negative) to it. The
1931 yield of the function is the new value. However, the value of the count
1932 is constrained to lie between 0 and 65535, inclusive. If the new value
1933 is outside these limits, it is forced to the appropriate limit value.
1934
1935 Except when it is zero, the reference count is not correctly preserved
1936 if a pattern is compiled on one host and then transferred to a host
1937 whose byte-order is different. (This seems a highly unlikely scenario.)
1938
1939
1940 MATCHING A PATTERN: THE TRADITIONAL FUNCTION
1941
1942 int pcre_exec(const pcre *code, const pcre_extra *extra,
1943 const char *subject, int length, int startoffset,
1944 int options, int *ovector, int ovecsize);
1945
1946 The function pcre_exec() is called to match a subject string against a
1947 compiled pattern, which is passed in the code argument. If the pattern
1948 was studied, the result of the study should be passed in the extra
1949 argument. This function is the main matching facility of the library,
1950 and it operates in a Perl-like manner. For specialist use there is also
1951 an alternative matching function, which is described below in the sec-
1952 tion about the pcre_dfa_exec() function.
1953
1954 In most applications, the pattern will have been compiled (and option-
1955 ally studied) in the same process that calls pcre_exec(). However, it
1956 is possible to save compiled patterns and study data, and then use them
1957 later in different processes, possibly even on different hosts. For a
1958 discussion about this, see the pcreprecompile documentation.
1959
1960 Here is an example of a simple call to pcre_exec():
1961
1962 int rc;
1963 int ovector[30];
1964 rc = pcre_exec(
1965 re, /* result of pcre_compile() */
1966 NULL, /* we didn't study the pattern */
1967 "some string", /* the subject string */
1968 11, /* the length of the subject string */
1969 0, /* start at offset 0 in the subject */
1970 0, /* default options */
1971 ovector, /* vector of integers for substring information */
1972 30); /* number of elements (NOT size in bytes) */
1973
1974 Extra data for pcre_exec()
1975
1976 If the extra argument is not NULL, it must point to a pcre_extra data
1977 block. The pcre_study() function returns such a block (when it doesn't
1978 return NULL), but you can also create one for yourself, and pass addi-
1979 tional information in it. The pcre_extra block contains the following
1980 fields (not necessarily in this order):
1981
1982 unsigned long int flags;
1983 void *study_data;
1984 unsigned long int match_limit;
1985 unsigned long int match_limit_recursion;
1986 void *callout_data;
1987 const unsigned char *tables;
1988 unsigned char **mark;
1989
1990 The flags field is a bitmap that specifies which of the other fields
1991 are set. The flag bits are:
1992
1993 PCRE_EXTRA_STUDY_DATA
1994 PCRE_EXTRA_MATCH_LIMIT
1995 PCRE_EXTRA_MATCH_LIMIT_RECURSION
1996 PCRE_EXTRA_CALLOUT_DATA
1997 PCRE_EXTRA_TABLES
1998 PCRE_EXTRA_MARK
1999
2000 Other flag bits should be set to zero. The study_data field is set in
2001 the pcre_extra block that is returned by pcre_study(), together with
2002 the appropriate flag bit. You should not set this yourself, but you may
2003 add to the block by setting the other fields and their corresponding
2004 flag bits.
2005
2006 The match_limit field provides a means of preventing PCRE from using up
2007 a vast amount of resources when running patterns that are not going to
2008 match, but which have a very large number of possibilities in their
2009 search trees. The classic example is a pattern that uses nested unlim-
2010 ited repeats.
2011
2012 Internally, PCRE uses a function called match() which it calls repeat-
2013 edly (sometimes recursively). The limit set by match_limit is imposed
2014 on the number of times this function is called during a match, which
2015 has the effect of limiting the amount of backtracking that can take
2016 place. For patterns that are not anchored, the count restarts from zero
2017 for each position in the subject string.
2018
2019 The default value for the limit can be set when PCRE is built; the
2020 default default is 10 million, which handles all but the most extreme
2021 cases. You can override the default by suppling pcre_exec() with a
2022 pcre_extra block in which match_limit is set, and
2023 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
2024 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
2025
2026 The match_limit_recursion field is similar to match_limit, but instead
2027 of limiting the total number of times that match() is called, it limits
2028 the depth of recursion. The recursion depth is a smaller number than
2029 the total number of calls, because not all calls to match() are recur-
2030 sive. This limit is of use only if it is set smaller than match_limit.
2031
2032 Limiting the recursion depth limits the amount of stack that can be
2033 used, or, when PCRE has been compiled to use memory on the heap instead
2034 of the stack, the amount of heap memory that can be used.
2035
2036 The default value for match_limit_recursion can be set when PCRE is
2037 built; the default default is the same value as the default for
2038 match_limit. You can override the default by suppling pcre_exec() with
2039 a pcre_extra block in which match_limit_recursion is set, and
2040 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
2041 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
2042
2043 The callout_data field is used in conjunction with the "callout" fea-
2044 ture, and is described in the pcrecallout documentation.
2045
2046 The tables field is used to pass a character tables pointer to
2047 pcre_exec(); this overrides the value that is stored with the compiled
2048 pattern. A non-NULL value is stored with the compiled pattern only if
2049 custom tables were supplied to pcre_compile() via its tableptr argu-
2050 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
2051 PCRE's internal tables to be used. This facility is helpful when re-
2052 using patterns that have been saved after compiling with an external
2053 set of tables, because the external tables might be at a different
2054 address when pcre_exec() is called. See the pcreprecompile documenta-
2055 tion for a discussion of saving compiled patterns for later use.
2056
2057 If PCRE_EXTRA_MARK is set in the flags field, the mark field must be
2058 set to point to a char * variable. If the pattern contains any back-
2059 tracking control verbs such as (*MARK:NAME), and the execution ends up
2060 with a name to pass back, a pointer to the name string (zero termi-
2061 nated) is placed in the variable pointed to by the mark field. The
2062 names are within the compiled pattern; if you wish to retain such a
2063 name you must copy it before freeing the memory of a compiled pattern.
2064 If there is no name to pass back, the variable pointed to by the mark
2065 field set to NULL. For details of the backtracking control verbs, see
2066 the section entitled "Backtracking control" in the pcrepattern documen-
2067 tation.
2068
2069 Option bits for pcre_exec()
2070
2071 The unused bits of the options argument for pcre_exec() must be zero.
2072 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
2073 PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NOTEMPTY_ATSTART,
2074 PCRE_NO_START_OPTIMIZE, PCRE_NO_UTF8_CHECK, PCRE_PARTIAL_SOFT, and
2075 PCRE_PARTIAL_HARD.
2076
2077 PCRE_ANCHORED
2078
2079 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
2080 matching position. If a pattern was compiled with PCRE_ANCHORED, or
2081 turned out to be anchored by virtue of its contents, it cannot be made
2082 unachored at matching time.
2083
2084 PCRE_BSR_ANYCRLF
2085 PCRE_BSR_UNICODE
2086
2087 These options (which are mutually exclusive) control what the \R escape
2088 sequence matches. The choice is either to match only CR, LF, or CRLF,
2089 or to match any Unicode newline sequence. These options override the
2090 choice that was made or defaulted when the pattern was compiled.
2091
2092 PCRE_NEWLINE_CR
2093 PCRE_NEWLINE_LF
2094 PCRE_NEWLINE_CRLF
2095 PCRE_NEWLINE_ANYCRLF
2096 PCRE_NEWLINE_ANY
2097
2098 These options override the newline definition that was chosen or
2099 defaulted when the pattern was compiled. For details, see the descrip-
2100 tion of pcre_compile() above. During matching, the newline choice
2101 affects the behaviour of the dot, circumflex, and dollar metacharac-
2102 ters. It may also alter the way the match position is advanced after a
2103 match failure for an unanchored pattern.
2104
2105 When PCRE_NEWLINE_CRLF, PCRE_NEWLINE_ANYCRLF, or PCRE_NEWLINE_ANY is
2106 set, and a match attempt for an unanchored pattern fails when the cur-
2107 rent position is at a CRLF sequence, and the pattern contains no
2108 explicit matches for CR or LF characters, the match position is
2109 advanced by two characters instead of one, in other words, to after the
2110 CRLF.
2111
2112 The above rule is a compromise that makes the most common cases work as
2113 expected. For example, if the pattern is .+A (and the PCRE_DOTALL
2114 option is not set), it does not match the string "\r\nA" because, after
2115 failing at the start, it skips both the CR and the LF before retrying.
2116 However, the pattern [\r\n]A does match that string, because it con-
2117 tains an explicit CR or LF reference, and so advances only by one char-
2118 acter after the first failure.
2119
2120 An explicit match for CR of LF is either a literal appearance of one of
2121 those characters, or one of the \r or \n escape sequences. Implicit
2122 matches such as [^X] do not count, nor does \s (which includes CR and
2123 LF in the characters that it matches).
2124
2125 Notwithstanding the above, anomalous effects may still occur when CRLF
2126 is a valid newline sequence and explicit \r or \n escapes appear in the
2127 pattern.
2128
2129 PCRE_NOTBOL
2130
2131 This option specifies that first character of the subject string is not
2132 the beginning of a line, so the circumflex metacharacter should not
2133 match before it. Setting this without PCRE_MULTILINE (at compile time)
2134 causes circumflex never to match. This option affects only the behav-
2135 iour of the circumflex metacharacter. It does not affect \A.
2136
2137 PCRE_NOTEOL
2138
2139 This option specifies that the end of the subject string is not the end
2140 of a line, so the dollar metacharacter should not match it nor (except
2141 in multiline mode) a newline immediately before it. Setting this with-
2142 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
2143 option affects only the behaviour of the dollar metacharacter. It does
2144 not affect \Z or \z.
2145
2146 PCRE_NOTEMPTY
2147
2148 An empty string is not considered to be a valid match if this option is
2149 set. If there are alternatives in the pattern, they are tried. If all
2150 the alternatives match the empty string, the entire match fails. For
2151 example, if the pattern
2152
2153 a?b?
2154
2155 is applied to a string not beginning with "a" or "b", it matches an
2156 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
2157 match is not valid, so PCRE searches further into the string for occur-
2158 rences of "a" or "b".
2159
2160 PCRE_NOTEMPTY_ATSTART
2161
2162 This is like PCRE_NOTEMPTY, except that an empty string match that is
2163 not at the start of the subject is permitted. If the pattern is
2164 anchored, such a match can occur only if the pattern contains \K.
2165
2166 Perl has no direct equivalent of PCRE_NOTEMPTY or
2167 PCRE_NOTEMPTY_ATSTART, but it does make a special case of a pattern
2168 match of the empty string within its split() function, and when using
2169 the /g modifier. It is possible to emulate Perl's behaviour after
2170 matching a null string by first trying the match again at the same off-
2171 set with PCRE_NOTEMPTY_ATSTART and PCRE_ANCHORED, and then if that
2172 fails, by advancing the starting offset (see below) and trying an ordi-
2173 nary match again. There is some code that demonstrates how to do this
2174 in the pcredemo sample program. In the most general case, you have to
2175 check to see if the newline convention recognizes CRLF as a newline,
2176 and if so, and the current character is CR followed by LF, advance the
2177 starting offset by two characters instead of one.
2178
2179 PCRE_NO_START_OPTIMIZE
2180
2181 There are a number of optimizations that pcre_exec() uses at the start
2182 of a match, in order to speed up the process. For example, if it is
2183 known that an unanchored match must start with a specific character, it
2184 searches the subject for that character, and fails immediately if it
2185 cannot find it, without actually running the main matching function.
2186 This means that a special item such as (*COMMIT) at the start of a pat-
2187 tern is not considered until after a suitable starting point for the
2188 match has been found. When callouts or (*MARK) items are in use, these
2189 "start-up" optimizations can cause them to be skipped if the pattern is
2190 never actually used. The start-up optimizations are in effect a pre-
2191 scan of the subject that takes place before the pattern is run.
2192
2193 The PCRE_NO_START_OPTIMIZE option disables the start-up optimizations,
2194 possibly causing performance to suffer, but ensuring that in cases
2195 where the result is "no match", the callouts do occur, and that items
2196 such as (*COMMIT) and (*MARK) are considered at every possible starting
2197 position in the subject string. If PCRE_NO_START_OPTIMIZE is set at
2198 compile time, it cannot be unset at matching time.
2199
2200 Setting PCRE_NO_START_OPTIMIZE can change the outcome of a matching
2201 operation. Consider the pattern
2202
2203 (*COMMIT)ABC
2204
2205 When this is compiled, PCRE records the fact that a match must start
2206 with the character "A". Suppose the subject string is "DEFABC". The
2207 start-up optimization scans along the subject, finds "A" and runs the
2208 first match attempt from there. The (*COMMIT) item means that the pat-
2209 tern must match the current starting position, which in this case, it
2210 does. However, if the same match is run with PCRE_NO_START_OPTIMIZE
2211 set, the initial scan along the subject string does not happen. The
2212 first match attempt is run starting from "D" and when this fails,
2213 (*COMMIT) prevents any further matches being tried, so the overall
2214 result is "no match". If the pattern is studied, more start-up opti-
2215 mizations may be used. For example, a minimum length for the subject
2216 may be recorded. Consider the pattern
2217
2218 (*MARK:A)(X|Y)
2219
2220 The minimum length for a match is one character. If the subject is
2221 "ABC", there will be attempts to match "ABC", "BC", "C", and then
2222 finally an empty string. If the pattern is studied, the final attempt
2223 does not take place, because PCRE knows that the subject is too short,
2224 and so the (*MARK) is never encountered. In this case, studying the
2225 pattern does not affect the overall match result, which is still "no
2226 match", but it does affect the auxiliary information that is returned.
2227
2228 PCRE_NO_UTF8_CHECK
2229
2230 When PCRE_UTF8 is set at compile time, the validity of the subject as a
2231 UTF-8 string is automatically checked when pcre_exec() is subsequently
2232 called. The value of startoffset is also checked to ensure that it
2233 points to the start of a UTF-8 character. There is a discussion about
2234 the validity of UTF-8 strings in the section on UTF-8 support in the
2235 main pcre page. If an invalid UTF-8 sequence of bytes is found,
2236 pcre_exec() returns the error PCRE_ERROR_BADUTF8 or, if PCRE_PAR-
2237 TIAL_HARD is set and the problem is a truncated UTF-8 character at the
2238 end of the subject, PCRE_ERROR_SHORTUTF8. In both cases, information
2239 about the precise nature of the error may also be returned (see the
2240 descriptions of these errors in the section entitled Error return val-
2241 ues from pcre_exec() below). If startoffset contains a value that does
2242 not point to the start of a UTF-8 character (or to the end of the sub-
2243 ject), PCRE_ERROR_BADUTF8_OFFSET is returned.
2244
2245 If you already know that your subject is valid, and you want to skip
2246 these checks for performance reasons, you can set the
2247 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
2248 do this for the second and subsequent calls to pcre_exec() if you are
2249 making repeated calls to find all the matches in a single subject
2250 string. However, you should be sure that the value of startoffset
2251 points to the start of a UTF-8 character (or the end of the subject).
2252 When PCRE_NO_UTF8_CHECK is set, the effect of passing an invalid UTF-8
2253 string as a subject or an invalid value of startoffset is undefined.
2254 Your program may crash.
2255
2256 PCRE_PARTIAL_HARD
2257 PCRE_PARTIAL_SOFT
2258
2259 These options turn on the partial matching feature. For backwards com-
2260 patibility, PCRE_PARTIAL is a synonym for PCRE_PARTIAL_SOFT. A partial
2261 match occurs if the end of the subject string is reached successfully,
2262 but there are not enough subject characters to complete the match. If
2263 this happens when PCRE_PARTIAL_SOFT (but not PCRE_PARTIAL_HARD) is set,
2264 matching continues by testing any remaining alternatives. Only if no
2265 complete match can be found is PCRE_ERROR_PARTIAL returned instead of
2266 PCRE_ERROR_NOMATCH. In other words, PCRE_PARTIAL_SOFT says that the
2267 caller is prepared to handle a partial match, but only if no complete
2268 match can be found.
2269
2270 If PCRE_PARTIAL_HARD is set, it overrides PCRE_PARTIAL_SOFT. In this
2271 case, if a partial match is found, pcre_exec() immediately returns
2272 PCRE_ERROR_PARTIAL, without considering any other alternatives. In
2273 other words, when PCRE_PARTIAL_HARD is set, a partial match is consid-
2274 ered to be more important that an alternative complete match.
2275
2276 In both cases, the portion of the string that was inspected when the
2277 partial match was found is set as the first matching string. There is a
2278 more detailed discussion of partial and multi-segment matching, with
2279 examples, in the pcrepartial documentation.
2280
2281 The string to be matched by pcre_exec()
2282
2283 The subject string is passed to pcre_exec() as a pointer in subject, a
2284 length (in bytes) in length, and a starting byte offset in startoffset.
2285 If this is negative or greater than the length of the subject,
2286 pcre_exec() returns PCRE_ERROR_BADOFFSET. When the starting offset is
2287 zero, the search for a match starts at the beginning of the subject,
2288 and this is by far the most common case. In UTF-8 mode, the byte offset
2289 must point to the start of a UTF-8 character (or the end of the sub-
2290 ject). Unlike the pattern string, the subject may contain binary zero
2291 bytes.
2292
2293 A non-zero starting offset is useful when searching for another match
2294 in the same subject by calling pcre_exec() again after a previous suc-
2295 cess. Setting startoffset differs from just passing over a shortened
2296 string and setting PCRE_NOTBOL in the case of a pattern that begins
2297 with any kind of lookbehind. For example, consider the pattern
2298
2299 \Biss\B
2300
2301 which finds occurrences of "iss" in the middle of words. (\B matches
2302 only if the current position in the subject is not a word boundary.)
2303 When applied to the string "Mississipi" the first call to pcre_exec()
2304 finds the first occurrence. If pcre_exec() is called again with just
2305 the remainder of the subject, namely "issipi", it does not match,
2306 because \B is always false at the start of the subject, which is deemed
2307 to be a word boundary. However, if pcre_exec() is passed the entire
2308 string again, but with startoffset set to 4, it finds the second occur-
2309 rence of "iss" because it is able to look behind the starting point to
2310 discover that it is preceded by a letter.
2311
2312 Finding all the matches in a subject is tricky when the pattern can
2313 match an empty string. It is possible to emulate Perl's /g behaviour by
2314 first trying the match again at the same offset, with the
2315 PCRE_NOTEMPTY_ATSTART and PCRE_ANCHORED options, and then if that
2316 fails, advancing the starting offset and trying an ordinary match
2317 again. There is some code that demonstrates how to do this in the pcre-
2318 demo sample program. In the most general case, you have to check to see
2319 if the newline convention recognizes CRLF as a newline, and if so, and
2320 the current character is CR followed by LF, advance the starting offset
2321 by two characters instead of one.
2322
2323 If a non-zero starting offset is passed when the pattern is anchored,
2324 one attempt to match at the given offset is made. This can only succeed
2325 if the pattern does not require the match to be at the start of the
2326 subject.
2327
2328 How pcre_exec() returns captured substrings
2329
2330 In general, a pattern matches a certain portion of the subject, and in
2331 addition, further substrings from the subject may be picked out by
2332 parts of the pattern. Following the usage in Jeffrey Friedl's book,
2333 this is called "capturing" in what follows, and the phrase "capturing
2334 subpattern" is used for a fragment of a pattern that picks out a sub-
2335 string. PCRE supports several other kinds of parenthesized subpattern
2336 that do not cause substrings to be captured.
2337
2338 Captured substrings are returned to the caller via a vector of integers
2339 whose address is passed in ovector. The number of elements in the vec-
2340 tor is passed in ovecsize, which must be a non-negative number. Note:
2341 this argument is NOT the size of ovector in bytes.
2342
2343 The first two-thirds of the vector is used to pass back captured sub-
2344 strings, each substring using a pair of integers. The remaining third
2345 of the vector is used as workspace by pcre_exec() while matching cap-
2346 turing subpatterns, and is not available for passing back information.
2347 The number passed in ovecsize should always be a multiple of three. If
2348 it is not, it is rounded down.
2349
2350 When a match is successful, information about captured substrings is
2351 returned in pairs of integers, starting at the beginning of ovector,
2352 and continuing up to two-thirds of its length at the most. The first
2353 element of each pair is set to the byte offset of the first character
2354 in a substring, and the second is set to the byte offset of the first
2355 character after the end of a substring. Note: these values are always
2356 byte offsets, even in UTF-8 mode. They are not character counts.
2357
2358 The first pair of integers, ovector[0] and ovector[1], identify the
2359 portion of the subject string matched by the entire pattern. The next
2360 pair is used for the first capturing subpattern, and so on. The value
2361 returned by pcre_exec() is one more than the highest numbered pair that
2362 has been set. For example, if two substrings have been captured, the
2363 returned value is 3. If there are no capturing subpatterns, the return
2364 value from a successful match is 1, indicating that just the first pair
2365 of offsets has been set.
2366
2367 If a capturing subpattern is matched repeatedly, it is the last portion
2368 of the string that it matched that is returned.
2369
2370 If the vector is too small to hold all the captured substring offsets,
2371 it is used as far as possible (up to two-thirds of its length), and the
2372 function returns a value of zero. If the substring offsets are not of
2373 interest, pcre_exec() may be called with ovector passed as NULL and
2374 ovecsize as zero. However, if the pattern contains back references and
2375 the ovector is not big enough to remember the related substrings, PCRE
2376 has to get additional memory for use during matching. Thus it is usu-
2377 ally advisable to supply an ovector.
2378
2379 The pcre_fullinfo() function can be used to find out how many capturing
2380 subpatterns there are in a compiled pattern. The smallest size for
2381 ovector that will allow for n captured substrings, in addition to the
2382 offsets of the substring matched by the whole pattern, is (n+1)*3.
2383
2384 It is possible for capturing subpattern number n+1 to match some part
2385 of the subject when subpattern n has not been used at all. For example,
2386 if the string "abc" is matched against the pattern (a|(z))(bc) the
2387 return from the function is 4, and subpatterns 1 and 3 are matched, but
2388 2 is not. When this happens, both values in the offset pairs corre-
2389 sponding to unused subpatterns are set to -1.
2390
2391 Offset values that correspond to unused subpatterns at the end of the
2392 expression are also set to -1. For example, if the string "abc" is
2393 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
2394 matched. The return from the function is 2, because the highest used
2395 capturing subpattern number is 1, and the offsets for for the second
2396 and third capturing subpatterns (assuming the vector is large enough,
2397 of course) are set to -1.
2398
2399 Note: Elements of ovector that do not correspond to capturing parenthe-
2400 ses in the pattern are never changed. That is, if a pattern contains n
2401 capturing parentheses, no more than ovector[0] to ovector[2n+1] are set
2402 by pcre_exec(). The other elements retain whatever values they previ-
2403 ously had.
2404
2405 Some convenience functions are provided for extracting the captured
2406 substrings as separate strings. These are described below.
2407
2408 Error return values from pcre_exec()
2409
2410 If pcre_exec() fails, it returns a negative number. The following are
2411 defined in the header file:
2412
2413 PCRE_ERROR_NOMATCH (-1)
2414
2415 The subject string did not match the pattern.
2416
2417 PCRE_ERROR_NULL (-2)
2418
2419 Either code or subject was passed as NULL, or ovector was NULL and
2420 ovecsize was not zero.
2421
2422 PCRE_ERROR_BADOPTION (-3)
2423
2424 An unrecognized bit was set in the options argument.
2425
2426 PCRE_ERROR_BADMAGIC (-4)
2427
2428 PCRE stores a 4-byte "magic number" at the start of the compiled code,
2429 to catch the case when it is passed a junk pointer and to detect when a
2430 pattern that was compiled in an environment of one endianness is run in
2431 an environment with the other endianness. This is the error that PCRE
2432 gives when the magic number is not present.
2433
2434 PCRE_ERROR_UNKNOWN_OPCODE (-5)
2435
2436 While running the pattern match, an unknown item was encountered in the
2437 compiled pattern. This error could be caused by a bug in PCRE or by
2438 overwriting of the compiled pattern.
2439
2440 PCRE_ERROR_NOMEMORY (-6)
2441
2442 If a pattern contains back references, but the ovector that is passed
2443 to pcre_exec() is not big enough to remember the referenced substrings,
2444 PCRE gets a block of memory at the start of matching to use for this
2445 purpose. If the call via pcre_malloc() fails, this error is given. The
2446 memory is automatically freed at the end of matching.
2447
2448 This error is also given if pcre_stack_malloc() fails in pcre_exec().
2449 This can happen only when PCRE has been compiled with --disable-stack-
2450 for-recursion.
2451
2452 PCRE_ERROR_NOSUBSTRING (-7)
2453
2454 This error is used by the pcre_copy_substring(), pcre_get_substring(),
2455 and pcre_get_substring_list() functions (see below). It is never
2456 returned by pcre_exec().
2457
2458 PCRE_ERROR_MATCHLIMIT (-8)
2459
2460 The backtracking limit, as specified by the match_limit field in a
2461 pcre_extra structure (or defaulted) was reached. See the description
2462 above.
2463
2464 PCRE_ERROR_CALLOUT (-9)
2465
2466 This error is never generated by pcre_exec() itself. It is provided for
2467 use by callout functions that want to yield a distinctive error code.
2468 See the pcrecallout documentation for details.
2469
2470 PCRE_ERROR_BADUTF8 (-10)
2471
2472 A string that contains an invalid UTF-8 byte sequence was passed as a
2473 subject, and the PCRE_NO_UTF8_CHECK option was not set. If the size of
2474 the output vector (ovecsize) is at least 2, the byte offset to the
2475 start of the the invalid UTF-8 character is placed in the first ele-
2476 ment, and a reason code is placed in the second element. The reason
2477 codes are listed in the following section. For backward compatibility,
2478 if PCRE_PARTIAL_HARD is set and the problem is a truncated UTF-8 char-
2479 acter at the end of the subject (reason codes 1 to 5),
2480 PCRE_ERROR_SHORTUTF8 is returned instead of PCRE_ERROR_BADUTF8.
2481
2482 PCRE_ERROR_BADUTF8_OFFSET (-11)
2483
2484 The UTF-8 byte sequence that was passed as a subject was checked and
2485 found to be valid (the PCRE_NO_UTF8_CHECK option was not set), but the
2486 value of startoffset did not point to the beginning of a UTF-8 charac-
2487 ter or the end of the subject.
2488
2489 PCRE_ERROR_PARTIAL (-12)
2490
2491 The subject string did not match, but it did match partially. See the
2492 pcrepartial documentation for details of partial matching.
2493
2494 PCRE_ERROR_BADPARTIAL (-13)
2495
2496 This code is no longer in use. It was formerly returned when the
2497 PCRE_PARTIAL option was used with a compiled pattern containing items
2498 that were not supported for partial matching. From release 8.00
2499 onwards, there are no restrictions on partial matching.
2500
2501 PCRE_ERROR_INTERNAL (-14)
2502
2503 An unexpected internal error has occurred. This error could be caused
2504 by a bug in PCRE or by overwriting of the compiled pattern.
2505
2506 PCRE_ERROR_BADCOUNT (-15)
2507
2508 This error is given if the value of the ovecsize argument is negative.
2509
2510 PCRE_ERROR_RECURSIONLIMIT (-21)
2511
2512 The internal recursion limit, as specified by the match_limit_recursion
2513 field in a pcre_extra structure (or defaulted) was reached. See the
2514 description above.
2515
2516 PCRE_ERROR_BADNEWLINE (-23)
2517
2518 An invalid combination of PCRE_NEWLINE_xxx options was given.
2519
2520 PCRE_ERROR_BADOFFSET (-24)
2521
2522 The value of startoffset was negative or greater than the length of the
2523 subject, that is, the value in length.
2524
2525 PCRE_ERROR_SHORTUTF8 (-25)
2526
2527 This error is returned instead of PCRE_ERROR_BADUTF8 when the subject
2528 string ends with a truncated UTF-8 character and the PCRE_PARTIAL_HARD
2529 option is set. Information about the failure is returned as for
2530 PCRE_ERROR_BADUTF8. It is in fact sufficient to detect this case, but
2531 this special error code for PCRE_PARTIAL_HARD precedes the implementa-
2532 tion of returned information; it is retained for backwards compatibil-
2533 ity.
2534
2535 PCRE_ERROR_RECURSELOOP (-26)
2536
2537 This error is returned when pcre_exec() detects a recursion loop within
2538 the pattern. Specifically, it means that either the whole pattern or a
2539 subpattern has been called recursively for the second time at the same
2540 position in the subject string. Some simple patterns that might do this
2541 are detected and faulted at compile time, but more complicated cases,
2542 in particular mutual recursions between two different subpatterns, can-
2543 not be detected until run time.
2544
2545 Error numbers -16 to -20 and -22 are not used by pcre_exec().
2546
2547 Reason codes for invalid UTF-8 strings
2548
2549 When pcre_exec() returns either PCRE_ERROR_BADUTF8 or PCRE_ERROR_SHORT-
2550 UTF8, and the size of the output vector (ovecsize) is at least 2, the
2551 offset of the start of the invalid UTF-8 character is placed in the
2552 first output vector element (ovector[0]) and a reason code is placed in
2553 the second element (ovector[1]). The reason codes are given names in
2554 the pcre.h header file:
2555
2556 PCRE_UTF8_ERR1
2557 PCRE_UTF8_ERR2
2558 PCRE_UTF8_ERR3
2559 PCRE_UTF8_ERR4
2560 PCRE_UTF8_ERR5
2561
2562 The string ends with a truncated UTF-8 character; the code specifies
2563 how many bytes are missing (1 to 5). Although RFC 3629 restricts UTF-8
2564 characters to be no longer than 4 bytes, the encoding scheme (origi-
2565 nally defined by RFC 2279) allows for up to 6 bytes, and this is
2566 checked first; hence the possibility of 4 or 5 missing bytes.
2567
2568 PCRE_UTF8_ERR6
2569 PCRE_UTF8_ERR7
2570 PCRE_UTF8_ERR8
2571 PCRE_UTF8_ERR9
2572 PCRE_UTF8_ERR10
2573
2574 The two most significant bits of the 2nd, 3rd, 4th, 5th, or 6th byte of
2575 the character do not have the binary value 0b10 (that is, either the
2576 most significant bit is 0, or the next bit is 1).
2577
2578 PCRE_UTF8_ERR11
2579 PCRE_UTF8_ERR12
2580
2581 A character that is valid by the RFC 2279 rules is either 5 or 6 bytes
2582 long; these code points are excluded by RFC 3629.
2583
2584 PCRE_UTF8_ERR13
2585
2586 A 4-byte character has a value greater than 0x10fff; these code points
2587 are excluded by RFC 3629.
2588
2589 PCRE_UTF8_ERR14
2590
2591 A 3-byte character has a value in the range 0xd800 to 0xdfff; this
2592 range of code points are reserved by RFC 3629 for use with UTF-16, and
2593 so are excluded from UTF-8.
2594
2595 PCRE_UTF8_ERR15
2596 PCRE_UTF8_ERR16
2597 PCRE_UTF8_ERR17
2598 PCRE_UTF8_ERR18
2599 PCRE_UTF8_ERR19
2600
2601 A 2-, 3-, 4-, 5-, or 6-byte character is "overlong", that is, it codes
2602 for a value that can be represented by fewer bytes, which is invalid.
2603 For example, the two bytes 0xc0, 0xae give the value 0x2e, whose cor-
2604 rect coding uses just one byte.
2605
2606 PCRE_UTF8_ERR20
2607
2608 The two most significant bits of the first byte of a character have the
2609 binary value 0b10 (that is, the most significant bit is 1 and the sec-
2610 ond is 0). Such a byte can only validly occur as the second or subse-
2611 quent byte of a multi-byte character.
2612
2613 PCRE_UTF8_ERR21
2614
2615 The first byte of a character has the value 0xfe or 0xff. These values
2616 can never occur in a valid UTF-8 string.
2617
2618
2619 EXTRACTING CAPTURED SUBSTRINGS BY NUMBER
2620
2621 int pcre_copy_substring(const char *subject, int *ovector,
2622 int stringcount, int stringnumber, char *buffer,
2623 int buffersize);
2624
2625 int pcre_get_substring(const char *subject, int *ovector,
2626 int stringcount, int stringnumber,
2627 const char **stringptr);
2628
2629 int pcre_get_substring_list(const char *subject,
2630 int *ovector, int stringcount, const char ***listptr);
2631
2632 Captured substrings can be accessed directly by using the offsets
2633 returned by pcre_exec() in ovector. For convenience, the functions
2634 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
2635 string_list() are provided for extracting captured substrings as new,
2636 separate, zero-terminated strings. These functions identify substrings
2637 by number. The next section describes functions for extracting named
2638 substrings.
2639
2640 A substring that contains a binary zero is correctly extracted and has
2641 a further zero added on the end, but the result is not, of course, a C
2642 string. However, you can process such a string by referring to the
2643 length that is returned by pcre_copy_substring() and pcre_get_sub-
2644 string(). Unfortunately, the interface to pcre_get_substring_list() is
2645 not adequate for handling strings containing binary zeros, because the
2646 end of the final string is not independently indicated.
2647
2648 The first three arguments are the same for all three of these func-
2649 tions: subject is the subject string that has just been successfully
2650 matched, ovector is a pointer to the vector of integer offsets that was
2651 passed to pcre_exec(), and stringcount is the number of substrings that
2652 were captured by the match, including the substring that matched the
2653 entire regular expression. This is the value returned by pcre_exec() if
2654 it is greater than zero. If pcre_exec() returned zero, indicating that
2655 it ran out of space in ovector, the value passed as stringcount should
2656 be the number of elements in the vector divided by three.
2657
2658 The functions pcre_copy_substring() and pcre_get_substring() extract a
2659 single substring, whose number is given as stringnumber. A value of
2660 zero extracts the substring that matched the entire pattern, whereas
2661 higher values extract the captured substrings. For pcre_copy_sub-
2662 string(), the string is placed in buffer, whose length is given by
2663 buffersize, while for pcre_get_substring() a new block of memory is
2664 obtained via pcre_malloc, and its address is returned via stringptr.
2665 The yield of the function is the length of the string, not including
2666 the terminating zero, or one of these error codes:
2667
2668 PCRE_ERROR_NOMEMORY (-6)
2669
2670 The buffer was too small for pcre_copy_substring(), or the attempt to
2671 get memory failed for pcre_get_substring().
2672
2673 PCRE_ERROR_NOSUBSTRING (-7)
2674
2675 There is no substring whose number is stringnumber.
2676
2677 The pcre_get_substring_list() function extracts all available sub-
2678 strings and builds a list of pointers to them. All this is done in a
2679 single block of memory that is obtained via pcre_malloc. The address of
2680 the memory block is returned via listptr, which is also the start of
2681 the list of string pointers. The end of the list is marked by a NULL
2682 pointer. The yield of the function is zero if all went well, or the
2683 error code
2684
2685 PCRE_ERROR_NOMEMORY (-6)
2686
2687 if the attempt to get the memory block failed.
2688
2689 When any of these functions encounter a substring that is unset, which
2690 can happen when capturing subpattern number n+1 matches some part of
2691 the subject, but subpattern n has not been used at all, they return an
2692 empty string. This can be distinguished from a genuine zero-length sub-
2693 string by inspecting the appropriate offset in ovector, which is nega-
2694 tive for unset substrings.
2695
2696 The two convenience functions pcre_free_substring() and pcre_free_sub-
2697 string_list() can be used to free the memory returned by a previous
2698 call of pcre_get_substring() or pcre_get_substring_list(), respec-
2699 tively. They do nothing more than call the function pointed to by
2700 pcre_free, which of course could be called directly from a C program.
2701 However, PCRE is used in some situations where it is linked via a spe-
2702 cial interface to another programming language that cannot use
2703 pcre_free directly; it is for these cases that the functions are pro-
2704 vided.
2705
2706
2707 EXTRACTING CAPTURED SUBSTRINGS BY NAME
2708
2709 int pcre_get_stringnumber(const pcre *code,
2710 const char *name);
2711
2712 int pcre_copy_named_substring(const pcre *code,
2713 const char *subject, int *ovector,
2714 int stringcount, const char *stringname,
2715 char *buffer, int buffersize);
2716
2717 int pcre_get_named_substring(const pcre *code,
2718 const char *subject, int *ovector,
2719 int stringcount, const char *stringname,
2720 const char **stringptr);
2721
2722 To extract a substring by name, you first have to find associated num-
2723 ber. For example, for this pattern
2724
2725 (a+)b(?<xxx>\d+)...
2726
2727 the number of the subpattern called "xxx" is 2. If the name is known to
2728 be unique (PCRE_DUPNAMES was not set), you can find the number from the
2729 name by calling pcre_get_stringnumber(). The first argument is the com-
2730 piled pattern, and the second is the name. The yield of the function is
2731 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
2732 subpattern of that name.
2733
2734 Given the number, you can extract the substring directly, or use one of
2735 the functions described in the previous section. For convenience, there
2736 are also two functions that do the whole job.
2737
2738 Most of the arguments of pcre_copy_named_substring() and
2739 pcre_get_named_substring() are the same as those for the similarly
2740 named functions that extract by number. As these are described in the
2741 previous section, they are not re-described here. There are just two
2742 differences:
2743
2744 First, instead of a substring number, a substring name is given. Sec-
2745 ond, there is an extra argument, given at the start, which is a pointer
2746 to the compiled pattern. This is needed in order to gain access to the
2747 name-to-number translation table.
2748
2749 These functions call pcre_get_stringnumber(), and if it succeeds, they
2750 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
2751 ate. NOTE: If PCRE_DUPNAMES is set and there are duplicate names, the
2752 behaviour may not be what you want (see the next section).
2753
2754 Warning: If the pattern uses the (?| feature to set up multiple subpat-
2755 terns with the same number, as described in the section on duplicate
2756 subpattern numbers in the pcrepattern page, you cannot use names to
2757 distinguish the different subpatterns, because names are not included
2758 in the compiled code. The matching process uses only numbers. For this
2759 reason, the use of different names for subpatterns of the same number
2760 causes an error at compile time.
2761
2762
2763 DUPLICATE SUBPATTERN NAMES
2764
2765 int pcre_get_stringtable_entries(const pcre *code,
2766 const char *name, char **first, char **last);
2767
2768 When a pattern is compiled with the PCRE_DUPNAMES option, names for
2769 subpatterns are not required to be unique. (Duplicate names are always
2770 allowed for subpatterns with the same number, created by using the (?|
2771 feature. Indeed, if such subpatterns are named, they are required to
2772 use the same names.)
2773
2774 Normally, patterns with duplicate names are such that in any one match,
2775 only one of the named subpatterns participates. An example is shown in
2776 the pcrepattern documentation.
2777
2778 When duplicates are present, pcre_copy_named_substring() and
2779 pcre_get_named_substring() return the first substring corresponding to
2780 the given name that is set. If none are set, PCRE_ERROR_NOSUBSTRING
2781 (-7) is returned; no data is returned. The pcre_get_stringnumber()
2782 function returns one of the numbers that are associated with the name,
2783 but it is not defined which it is.
2784
2785 If you want to get full details of all captured substrings for a given
2786 name, you must use the pcre_get_stringtable_entries() function. The
2787 first argument is the compiled pattern, and the second is the name. The
2788 third and fourth are pointers to variables which are updated by the
2789 function. After it has run, they point to the first and last entries in
2790 the name-to-number table for the given name. The function itself
2791 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
2792 there are none. The format of the table is described above in the sec-
2793 tion entitled Information about a pattern above. Given all the rele-
2794 vant entries for the name, you can extract each of their numbers, and
2795 hence the captured data, if any.
2796
2797
2798 FINDING ALL POSSIBLE MATCHES
2799
2800 The traditional matching function uses a similar algorithm to Perl,
2801 which stops when it finds the first match, starting at a given point in
2802 the subject. If you want to find all possible matches, or the longest
2803 possible match, consider using the alternative matching function (see
2804 below) instead. If you cannot use the alternative function, but still
2805 need to find all possible matches, you can kludge it up by making use
2806 of the callout facility, which is described in the pcrecallout documen-
2807 tation.
2808
2809 What you have to do is to insert a callout right at the end of the pat-
2810 tern. When your callout function is called, extract and save the cur-
2811 rent matched substring. Then return 1, which forces pcre_exec() to
2812 backtrack and try other alternatives. Ultimately, when it runs out of
2813 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
2814
2815
2816 MATCHING A PATTERN: THE ALTERNATIVE FUNCTION
2817
2818 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
2819 const char *subject, int length, int startoffset,
2820 int options, int *ovector, int ovecsize,
2821 int *workspace, int wscount);
2822
2823 The function pcre_dfa_exec() is called to match a subject string
2824 against a compiled pattern, using a matching algorithm that scans the
2825 subject string just once, and does not backtrack. This has different
2826 characteristics to the normal algorithm, and is not compatible with
2827 Perl. Some of the features of PCRE patterns are not supported. Never-
2828 theless, there are times when this kind of matching can be useful. For
2829 a discussion of the two matching algorithms, and a list of features
2830 that pcre_dfa_exec() does not support, see the pcrematching documenta-
2831 tion.
2832
2833 The arguments for the pcre_dfa_exec() function are the same as for
2834 pcre_exec(), plus two extras. The ovector argument is used in a differ-
2835 ent way, and this is described below. The other common arguments are
2836 used in the same way as for pcre_exec(), so their description is not
2837 repeated here.
2838
2839 The two additional arguments provide workspace for the function. The
2840 workspace vector should contain at least 20 elements. It is used for
2841 keeping track of multiple paths through the pattern tree. More
2842 workspace will be needed for patterns and subjects where there are a
2843 lot of potential matches.
2844
2845 Here is an example of a simple call to pcre_dfa_exec():
2846
2847 int rc;
2848 int ovector[10];
2849 int wspace[20];
2850 rc = pcre_dfa_exec(
2851 re, /* result of pcre_compile() */
2852 NULL, /* we didn't study the pattern */
2853 "some string", /* the subject string */
2854 11, /* the length of the subject string */
2855 0, /* start at offset 0 in the subject */
2856 0, /* default options */
2857 ovector, /* vector of integers for substring information */
2858 10, /* number of elements (NOT size in bytes) */
2859 wspace, /* working space vector */
2860 20); /* number of elements (NOT size in bytes) */
2861
2862 Option bits for pcre_dfa_exec()
2863
2864 The unused bits of the options argument for pcre_dfa_exec() must be
2865 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
2866 LINE_xxx, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY,
2867 PCRE_NOTEMPTY_ATSTART, PCRE_NO_UTF8_CHECK, PCRE_BSR_ANYCRLF,
2868 PCRE_BSR_UNICODE, PCRE_NO_START_OPTIMIZE, PCRE_PARTIAL_HARD, PCRE_PAR-
2869 TIAL_SOFT, PCRE_DFA_SHORTEST, and PCRE_DFA_RESTART. All but the last
2870 four of these are exactly the same as for pcre_exec(), so their
2871 description is not repeated here.
2872
2873 PCRE_PARTIAL_HARD
2874 PCRE_PARTIAL_SOFT
2875
2876 These have the same general effect as they do for pcre_exec(), but the
2877 details are slightly different. When PCRE_PARTIAL_HARD is set for
2878 pcre_dfa_exec(), it returns PCRE_ERROR_PARTIAL if the end of the sub-
2879 ject is reached and there is still at least one matching possibility
2880 that requires additional characters. This happens even if some complete
2881 matches have also been found. When PCRE_PARTIAL_SOFT is set, the return
2882 code PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end
2883 of the subject is reached, there have been no complete matches, but
2884 there is still at least one matching possibility. The portion of the
2885 string that was inspected when the longest partial match was found is
2886 set as the first matching string in both cases. There is a more
2887 detailed discussion of partial and multi-segment matching, with exam-
2888 ples, in the pcrepartial documentation.
2889
2890 PCRE_DFA_SHORTEST
2891
2892 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
2893 stop as soon as it has found one match. Because of the way the alterna-
2894 tive algorithm works, this is necessarily the shortest possible match
2895 at the first possible matching point in the subject string.
2896
2897 PCRE_DFA_RESTART
2898
2899 When pcre_dfa_exec() returns a partial match, it is possible to call it
2900 again, with additional subject characters, and have it continue with
2901 the same match. The PCRE_DFA_RESTART option requests this action; when
2902 it is set, the workspace and wscount options must reference the same
2903 vector as before because data about the match so far is left in them
2904 after a partial match. There is more discussion of this facility in the
2905 pcrepartial documentation.
2906
2907 Successful returns from pcre_dfa_exec()
2908
2909 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
2910 string in the subject. Note, however, that all the matches from one run
2911 of the function start at the same point in the subject. The shorter
2912 matches are all initial substrings of the longer matches. For example,
2913 if the pattern
2914
2915 <.*>
2916
2917 is matched against the string
2918
2919 This is <something> <something else> <something further> no more
2920
2921 the three matched strings are
2922
2923 <something>
2924 <something> <something else>
2925 <something> <something else> <something further>
2926
2927 On success, the yield of the function is a number greater than zero,
2928 which is the number of matched substrings. The substrings themselves
2929 are returned in ovector. Each string uses two elements; the first is
2930 the offset to the start, and the second is the offset to the end. In
2931 fact, all the strings have the same start offset. (Space could have
2932 been saved by giving this only once, but it was decided to retain some
2933 compatibility with the way pcre_exec() returns data, even though the
2934 meaning of the strings is different.)
2935
2936 The strings are returned in reverse order of length; that is, the long-
2937 est matching string is given first. If there were too many matches to
2938 fit into ovector, the yield of the function is zero, and the vector is
2939 filled with the longest matches.
2940
2941 Error returns from pcre_dfa_exec()
2942
2943 The pcre_dfa_exec() function returns a negative number when it fails.
2944 Many of the errors are the same as for pcre_exec(), and these are
2945 described above. There are in addition the following errors that are
2946 specific to pcre_dfa_exec():
2947
2948 PCRE_ERROR_DFA_UITEM (-16)
2949
2950 This return is given if pcre_dfa_exec() encounters an item in the pat-
2951 tern that it does not support, for instance, the use of \C or a back
2952 reference.
2953
2954 PCRE_ERROR_DFA_UCOND (-17)
2955
2956 This return is given if pcre_dfa_exec() encounters a condition item
2957 that uses a back reference for the condition, or a test for recursion
2958 in a specific group. These are not supported.
2959
2960 PCRE_ERROR_DFA_UMLIMIT (-18)
2961
2962 This return is given if pcre_dfa_exec() is called with an extra block
2963 that contains a setting of the match_limit field. This is not supported
2964 (it is meaningless).
2965
2966 PCRE_ERROR_DFA_WSSIZE (-19)
2967
2968 This return is given if pcre_dfa_exec() runs out of space in the
2969 workspace vector.
2970
2971 PCRE_ERROR_DFA_RECURSE (-20)
2972
2973 When a recursive subpattern is processed, the matching function calls
2974 itself recursively, using private vectors for ovector and workspace.
2975 This error is given if the output vector is not large enough. This
2976 should be extremely rare, as a vector of size 1000 is used.
2977
2978
2979 SEE ALSO
2980
2981 pcrebuild(3), pcrecallout(3), pcrecpp(3)(3), pcrematching(3), pcrepar-
2982 tial(3), pcreposix(3), pcreprecompile(3), pcresample(3), pcrestack(3).
2983
2984
2985 AUTHOR
2986
2987 Philip Hazel
2988 University Computing Service
2989 Cambridge CB2 3QH, England.
2990
2991
2992 REVISION
2993
2994 Last updated: 28 July 2011
2995 Copyright (c) 1997-2011 University of Cambridge.
2996 ------------------------------------------------------------------------------
2997
2998
2999 PCRECALLOUT(3) PCRECALLOUT(3)
3000
3001
3002 NAME
3003 PCRE - Perl-compatible regular expressions
3004
3005
3006 PCRE CALLOUTS
3007
3008 int (*pcre_callout)(pcre_callout_block *);
3009
3010 PCRE provides a feature called "callout", which is a means of temporar-
3011 ily passing control to the caller of PCRE in the middle of pattern
3012 matching. The caller of PCRE provides an external function by putting
3013 its entry point in the global variable pcre_callout. By default, this
3014 variable contains NULL, which disables all calling out.
3015
3016 Within a regular expression, (?C) indicates the points at which the
3017 external function is to be called. Different callout points can be
3018 identified by putting a number less than 256 after the letter C. The
3019 default value is zero. For example, this pattern has two callout
3020 points:
3021
3022 (?C1)abc(?C2)def
3023
3024 If the PCRE_AUTO_CALLOUT option bit is set when pcre_compile() or
3025 pcre_compile2() is called, PCRE automatically inserts callouts, all
3026 with number 255, before each item in the pattern. For example, if
3027 PCRE_AUTO_CALLOUT is used with the pattern
3028
3029 A(\d{2}|--)
3030
3031 it is processed as if it were
3032
3033 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
3034
3035 Notice that there is a callout before and after each parenthesis and
3036 alternation bar. Automatic callouts can be used for tracking the
3037 progress of pattern matching. The pcretest command has an option that
3038 sets automatic callouts; when it is used, the output indicates how the
3039 pattern is matched. This is useful information when you are trying to
3040 optimize the performance of a particular pattern.
3041
3042
3043 MISSING CALLOUTS
3044
3045 You should be aware that, because of optimizations in the way PCRE
3046 matches patterns by default, callouts sometimes do not happen. For
3047 example, if the pattern is
3048
3049 ab(?C4)cd
3050
3051 PCRE knows that any matching string must contain the letter "d". If the
3052 subject string is "abyz", the lack of "d" means that matching doesn't
3053 ever start, and the callout is never reached. However, with "abyd",
3054 though the result is still no match, the callout is obeyed.
3055
3056 If the pattern is studied, PCRE knows the minimum length of a matching
3057 string, and will immediately give a "no match" return without actually
3058 running a match if the subject is not long enough, or, for unanchored
3059 patterns, if it has been scanned far enough.
3060
3061 You can disable these optimizations by passing the PCRE_NO_START_OPTI-
3062 MIZE option to pcre_compile(), pcre_exec(), or pcre_dfa_exec(), or by
3063 starting the pattern with (*NO_START_OPT). This slows down the matching
3064 process, but does ensure that callouts such as the example above are
3065 obeyed.
3066
3067
3068 THE CALLOUT INTERFACE
3069
3070 During matching, when PCRE reaches a callout point, the external func-
3071 tion defined by pcre_callout is called (if it is set). This applies to
3072 both the pcre_exec() and the pcre_dfa_exec() matching functions. The
3073 only argument to the callout function is a pointer to a pcre_callout
3074 block. This structure contains the following fields:
3075
3076 int version;
3077 int callout_number;
3078 int *offset_vector;
3079 const char *subject;
3080 int subject_length;
3081 int start_match;
3082 int current_position;
3083 int capture_top;
3084 int capture_last;
3085 void *callout_data;
3086 int pattern_position;
3087 int next_item_length;
3088 const unsigned char *mark;
3089
3090 The version field is an integer containing the version number of the
3091 block format. The initial version was 0; the current version is 2. The
3092 version number will change again in future if additional fields are
3093 added, but the intention is never to remove any of the existing fields.
3094
3095 The callout_number field contains the number of the callout, as com-
3096 piled into the pattern (that is, the number after ?C for manual call-
3097 outs, and 255 for automatically generated callouts).
3098
3099 The offset_vector field is a pointer to the vector of offsets that was
3100 passed by the caller to pcre_exec() or pcre_dfa_exec(). When
3101 pcre_exec() is used, the contents can be inspected in order to extract
3102 substrings that have been matched so far, in the same way as for
3103 extracting substrings after a match has completed. For pcre_dfa_exec()
3104 this field is not useful.
3105
3106 The subject and subject_length fields contain copies of the values that
3107 were passed to pcre_exec().
3108
3109 The start_match field normally contains the offset within the subject
3110 at which the current match attempt started. However, if the escape
3111 sequence \K has been encountered, this value is changed to reflect the
3112 modified starting point. If the pattern is not anchored, the callout
3113 function may be called several times from the same point in the pattern
3114 for different starting points in the subject.
3115
3116 The current_position field contains the offset within the subject of
3117 the current match pointer.
3118
3119 When the pcre_exec() function is used, the capture_top field contains
3120 one more than the number of the highest numbered captured substring so
3121 far. If no substrings have been captured, the value of capture_top is
3122 one. This is always the case when pcre_dfa_exec() is used, because it
3123 does not support captured substrings.
3124
3125 The capture_last field contains the number of the most recently cap-
3126 tured substring. If no substrings have been captured, its value is -1.
3127 This is always the case when pcre_dfa_exec() is used.
3128
3129 The callout_data field contains a value that is passed to pcre_exec()
3130 or pcre_dfa_exec() specifically so that it can be passed back in call-
3131 outs. It is passed in the pcre_callout field of the pcre_extra data
3132 structure. If no such data was passed, the value of callout_data in a
3133 pcre_callout block is NULL. There is a description of the pcre_extra
3134 structure in the pcreapi documentation.
3135
3136 The pattern_position field is present from version 1 of the pcre_call-
3137 out structure. It contains the offset to the next item to be matched in
3138 the pattern string.
3139
3140 The next_item_length field is present from version 1 of the pcre_call-
3141 out structure. It contains the length of the next item to be matched in
3142 the pattern string. When the callout immediately precedes an alterna-
3143 tion bar, a closing parenthesis, or the end of the pattern, the length
3144 is zero. When the callout precedes an opening parenthesis, the length
3145 is that of the entire subpattern.
3146
3147 The pattern_position and next_item_length fields are intended to help
3148 in distinguishing between different automatic callouts, which all have
3149 the same callout number. However, they are set for all callouts.
3150
3151 The mark field is present from version 2 of the pcre_callout structure.
3152 In callouts from pcre_exec() it contains a pointer to the zero-termi-
3153 nated name of the most recently passed (*MARK) item in the match, or
3154 NULL if there are no (*MARK)s in the current matching path. In callouts
3155 from pcre_dfa_exec() this field always contains NULL.
3156
3157
3158 RETURN VALUES
3159
3160 The external callout function returns an integer to PCRE. If the value
3161 is zero, matching proceeds as normal. If the value is greater than
3162 zero, matching fails at the current point, but the testing of other
3163 matching possibilities goes ahead, just as if a lookahead assertion had
3164 failed. If the value is less than zero, the match is abandoned, and
3165 pcre_exec() or pcre_dfa_exec() returns the negative value.
3166
3167 Negative values should normally be chosen from the set of
3168 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
3169 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
3170 reserved for use by callout functions; it will never be used by PCRE
3171 itself.
3172
3173
3174 AUTHOR
3175
3176 Philip Hazel
3177 University Computing Service
3178 Cambridge CB2 3QH, England.
3179
3180
3181 REVISION
3182
3183 Last updated: 31 July 2011
3184 Copyright (c) 1997-2011 University of Cambridge.
3185 ------------------------------------------------------------------------------
3186
3187
3188 PCRECOMPAT(3) PCRECOMPAT(3)
3189
3190
3191 NAME
3192 PCRE - Perl-compatible regular expressions
3193
3194
3195 DIFFERENCES BETWEEN PCRE AND PERL
3196
3197 This document describes the differences in the ways that PCRE and Perl
3198 handle regular expressions. The differences described here are with
3199 respect to Perl versions 5.10 and above.
3200
3201 1. PCRE has only a subset of Perl's UTF-8 and Unicode support. Details
3202 of what it does have are given in the section on UTF-8 support in the
3203 main pcre page.
3204
3205 2. PCRE allows repeat quantifiers only on parenthesized assertions, but
3206 they do not mean what you might think. For example, (?!a){3} does not
3207 assert that the next three characters are not "a". It just asserts that
3208 the next character is not "a" three times (in principle: PCRE optimizes
3209 this to run the assertion just once). Perl allows repeat quantifiers on
3210 other assertions such as \b, but these do not seem to have any use.
3211
3212 3. Capturing subpatterns that occur inside negative lookahead asser-
3213 tions are counted, but their entries in the offsets vector are never
3214 set. Perl sets its numerical variables from any such patterns that are
3215 matched before the assertion fails to match something (thereby succeed-
3216 ing), but only if the negative lookahead assertion contains just one
3217 branch.
3218
3219 4. Though binary zero characters are supported in the subject string,
3220 they are not allowed in a pattern string because it is passed as a nor-
3221 mal C string, terminated by zero. The escape sequence \0 can be used in
3222 the pattern to represent a binary zero.
3223
3224 5. The following Perl escape sequences are not supported: \l, \u, \L,
3225 \U, and \N when followed by a character name or Unicode value. (\N on
3226 its own, matching a non-newline character, is supported.) In fact these
3227 are implemented by Perl's general string-handling and are not part of
3228 its pattern matching engine. If any of these are encountered by PCRE,
3229 an error is generated.
3230
3231 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
3232 is built with Unicode character property support. The properties that
3233 can be tested with \p and \P are limited to the general category prop-
3234 erties such as Lu and Nd, script names such as Greek or Han, and the
3235 derived properties Any and L&. PCRE does support the Cs (surrogate)
3236 property, which Perl does not; the Perl documentation says "Because
3237 Perl hides the need for the user to understand the internal representa-
3238 tion of Unicode characters, there is no need to implement the somewhat
3239 messy concept of surrogates."
3240
3241 7. PCRE implements a simpler version of \X than Perl, which changed to
3242 make \X match what Unicode calls an "extended grapheme cluster". This
3243 is more complicated than an extended Unicode sequence, which is what
3244 PCRE matches.
3245
3246 8. PCRE does support the \Q...\E escape for quoting substrings. Charac-
3247 ters in between are treated as literals. This is slightly different
3248 from Perl in that $ and @ are also handled as literals inside the
3249 quotes. In Perl, they cause variable interpolation (but of course PCRE
3250 does not have variables). Note the following examples:
3251
3252 Pattern PCRE matches Perl matches
3253
3254 \Qabc$xyz\E abc$xyz abc followed by the
3255 contents of $xyz
3256 \Qabc\$xyz\E abc\$xyz abc\$xyz
3257 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3258
3259 The \Q...\E sequence is recognized both inside and outside character
3260 classes.
3261
3262 9. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
3263 constructions. However, there is support for recursive patterns. This
3264 is not available in Perl 5.8, but it is in Perl 5.10. Also, the PCRE
3265 "callout" feature allows an external function to be called during pat-
3266 tern matching. See the pcrecallout documentation for details.
3267
3268 10. Subpatterns that are called recursively or as "subroutines" are
3269 always treated as atomic groups in PCRE. This is like Python, but
3270 unlike Perl. There is a discussion of an example that explains this in
3271 more detail in the section on recursion differences from Perl in the
3272 pcrepattern page.
3273
3274 11. There are some differences that are concerned with the settings of
3275 captured strings when part of a pattern is repeated. For example,
3276 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
3277 unset, but in PCRE it is set to "b".
3278
3279 12. PCRE's handling of duplicate subpattern numbers and duplicate sub-
3280 pattern names is not as general as Perl's. This is a consequence of the
3281 fact the PCRE works internally just with numbers, using an external ta-
3282 ble to translate between numbers and names. In particular, a pattern
3283 such as (?|(?<a>A)|(?<b)B), where the two capturing parentheses have
3284 the same number but different names, is not supported, and causes an
3285 error at compile time. If it were allowed, it would not be possible to
3286 distinguish which parentheses matched, because both names map to cap-
3287 turing subpattern number 1. To avoid this confusing situation, an error
3288 is given at compile time.
3289
3290 13. Perl recognizes comments in some places that PCRE does not, for
3291 example, between the ( and ? at the start of a subpattern. If the /x
3292 modifier is set, Perl allows whitespace between ( and ? but PCRE never
3293 does, even if the PCRE_EXTENDED option is set.
3294
3295 14. PCRE provides some extensions to the Perl regular expression facil-
3296 ities. Perl 5.10 includes new features that are not in earlier ver-
3297 sions of Perl, some of which (such as named parentheses) have been in
3298 PCRE for some time. This list is with respect to Perl 5.10:
3299
3300 (a) Although lookbehind assertions in PCRE must match fixed length
3301 strings, each alternative branch of a lookbehind assertion can match a
3302 different length of string. Perl requires them all to have the same
3303 length.
3304
3305 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
3306 meta-character matches only at the very end of the string.
3307
3308 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
3309 cial meaning is faulted. Otherwise, like Perl, the backslash is quietly
3310 ignored. (Perl can be made to issue a warning.)
3311
3312 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
3313 fiers is inverted, that is, by default they are not greedy, but if fol-
3314 lowed by a question mark they are.
3315
3316 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
3317 tried only at the first matching position in the subject string.
3318
3319 (f) The PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NOTEMPTY_ATSTART,
3320 and PCRE_NO_AUTO_CAPTURE options for pcre_exec() have no Perl equiva-
3321 lents.
3322
3323 (g) The \R escape sequence can be restricted to match only CR, LF, or
3324 CRLF by the PCRE_BSR_ANYCRLF option.
3325
3326 (h) The callout facility is PCRE-specific.
3327
3328 (i) The partial matching facility is PCRE-specific.
3329
3330 (j) Patterns compiled by PCRE can be saved and re-used at a later time,
3331 even on different hosts that have the other endianness.
3332
3333 (k) The alternative matching function (pcre_dfa_exec()) matches in a
3334 different way and is not Perl-compatible.
3335
3336 (l) PCRE recognizes some special sequences such as (*CR) at the start
3337 of a pattern that set overall options that cannot be changed within the
3338 pattern.
3339
3340
3341 AUTHOR
3342
3343 Philip Hazel
3344 University Computing Service
3345 Cambridge CB2 3QH, England.
3346
3347
3348 REVISION
3349
3350 Last updated: 24 July 2011
3351 Copyright (c) 1997-2011 University of Cambridge.
3352 ------------------------------------------------------------------------------
3353
3354
3355 PCREPATTERN(3) PCREPATTERN(3)
3356
3357
3358 NAME
3359 PCRE - Perl-compatible regular expressions
3360
3361
3362 PCRE REGULAR EXPRESSION DETAILS
3363
3364 The syntax and semantics of the regular expressions that are supported
3365 by PCRE are described in detail below. There is a quick-reference syn-
3366 tax summary in the pcresyntax page. PCRE tries to match Perl syntax and
3367 semantics as closely as it can. PCRE also supports some alternative
3368 regular expression syntax (which does not conflict with the Perl syn-
3369 tax) in order to provide some compatibility with regular expressions in
3370 Python, .NET, and Oniguruma.
3371
3372 Perl's regular expressions are described in its own documentation, and
3373 regular expressions in general are covered in a number of books, some
3374 of which have copious examples. Jeffrey Friedl's "Mastering Regular
3375 Expressions", published by O'Reilly, covers regular expressions in
3376 great detail. This description of PCRE's regular expressions is
3377 intended as reference material.
3378
3379 The original operation of PCRE was on strings of one-byte characters.
3380 However, there is now also support for UTF-8 character strings. To use
3381 this, PCRE must be built to include UTF-8 support, and you must call
3382 pcre_compile() or pcre_compile2() with the PCRE_UTF8 option. There is
3383 also a special sequence that can be given at the start of a pattern:
3384
3385 (*UTF8)
3386
3387 Starting a pattern with this sequence is equivalent to setting the
3388 PCRE_UTF8 option. This feature is not Perl-compatible. How setting
3389 UTF-8 mode affects pattern matching is mentioned in several places
3390 below. There is also a summary of UTF-8 features in the section on
3391 UTF-8 support in the main pcre page.
3392
3393 Another special sequence that may appear at the start of a pattern or
3394 in combination with (*UTF8) is:
3395
3396 (*UCP)
3397
3398 This has the same effect as setting the PCRE_UCP option: it causes
3399 sequences such as \d and \w to use Unicode properties to determine
3400 character types, instead of recognizing only characters with codes less
3401 than 128 via a lookup table.
3402
3403 If a pattern starts with (*NO_START_OPT), it has the same effect as
3404 setting the PCRE_NO_START_OPTIMIZE option either at compile or matching
3405 time. There are also some more of these special sequences that are con-
3406 cerned with the handling of newlines; they are described below.
3407
3408 The remainder of this document discusses the patterns that are sup-
3409 ported by PCRE when its main matching function, pcre_exec(), is used.
3410 From release 6.0, PCRE offers a second matching function,
3411 pcre_dfa_exec(), which matches using a different algorithm that is not
3412 Perl-compatible. Some of the features discussed below are not available
3413 when pcre_dfa_exec() is used. The advantages and disadvantages of the
3414 alternative function, and how it differs from the normal function, are
3415 discussed in the pcrematching page.
3416
3417
3418 NEWLINE CONVENTIONS
3419
3420 PCRE supports five different conventions for indicating line breaks in
3421 strings: a single CR (carriage return) character, a single LF (line-
3422 feed) character, the two-character sequence CRLF, any of the three pre-
3423 ceding, or any Unicode newline sequence. The pcreapi page has further
3424 discussion about newlines, and shows how to set the newline convention
3425 in the options arguments for the compiling and matching functions.
3426
3427 It is also possible to specify a newline convention by starting a pat-
3428 tern string with one of the following five sequences:
3429
3430 (*CR) carriage return
3431 (*LF) linefeed
3432 (*CRLF) carriage return, followed by linefeed
3433 (*ANYCRLF) any of the three above
3434 (*ANY) all Unicode newline sequences
3435
3436 These override the default and the options given to pcre_compile() or
3437 pcre_compile2(). For example, on a Unix system where LF is the default
3438 newline sequence, the pattern
3439
3440 (*CR)a.b
3441
3442 changes the convention to CR. That pattern matches "a\nb" because LF is
3443 no longer a newline. Note that these special settings, which are not
3444 Perl-compatible, are recognized only at the very start of a pattern,
3445 and that they must be in upper case. If more than one of them is
3446 present, the last one is used.
3447
3448 The newline convention affects the interpretation of the dot metachar-
3449 acter when PCRE_DOTALL is not set, and also the behaviour of \N. How-
3450 ever, it does not affect what the \R escape sequence matches. By
3451 default, this is any Unicode newline sequence, for Perl compatibility.
3452 However, this can be changed; see the description of \R in the section
3453 entitled "Newline sequences" below. A change of \R setting can be com-
3454 bined with a change of newline convention.
3455
3456
3457 CHARACTERS AND METACHARACTERS
3458
3459 A regular expression is a pattern that is matched against a subject
3460 string from left to right. Most characters stand for themselves in a
3461 pattern, and match the corresponding characters in the subject. As a
3462 trivial example, the pattern
3463
3464 The quick brown fox
3465
3466 matches a portion of a subject string that is identical to itself. When
3467 caseless matching is specified (the PCRE_CASELESS option), letters are
3468 matched independently of case. In UTF-8 mode, PCRE always understands
3469 the concept of case for characters whose values are less than 128, so
3470 caseless matching is always possible. For characters with higher val-
3471 ues, the concept of case is supported if PCRE is compiled with Unicode
3472 property support, but not otherwise. If you want to use caseless
3473 matching for characters 128 and above, you must ensure that PCRE is
3474 compiled with Unicode property support as well as with UTF-8 support.
3475
3476 The power of regular expressions comes from the ability to include
3477 alternatives and repetitions in the pattern. These are encoded in the
3478 pattern by the use of metacharacters, which do not stand for themselves
3479 but instead are interpreted in some special way.
3480
3481 There are two different sets of metacharacters: those that are recog-
3482 nized anywhere in the pattern except within square brackets, and those
3483 that are recognized within square brackets. Outside square brackets,
3484 the metacharacters are as follows:
3485
3486 \ general escape character with several uses
3487 ^ assert start of string (or line, in multiline mode)
3488 $ assert end of string (or line, in multiline mode)
3489 . match any character except newline (by default)
3490 [ start character class definition
3491 | start of alternative branch
3492 ( start subpattern
3493 ) end subpattern
3494 ? extends the meaning of (
3495 also 0 or 1 quantifier
3496 also quantifier minimizer
3497 * 0 or more quantifier
3498 + 1 or more quantifier
3499 also "possessive quantifier"
3500 { start min/max quantifier
3501
3502 Part of a pattern that is in square brackets is called a "character
3503 class". In a character class the only metacharacters are:
3504
3505 \ general escape character
3506 ^ negate the class, but only if the first character
3507 - indicates character range
3508 [ POSIX character class (only if followed by POSIX
3509 syntax)
3510 ] terminates the character class
3511
3512 The following sections describe the use of each of the metacharacters.
3513
3514
3515 BACKSLASH
3516
3517 The backslash character has several uses. Firstly, if it is followed by
3518 a character that is not a number or a letter, it takes away any special
3519 meaning that character may have. This use of backslash as an escape
3520 character applies both inside and outside character classes.
3521
3522 For example, if you want to match a * character, you write \* in the
3523 pattern. This escaping action applies whether or not the following
3524 character would otherwise be interpreted as a metacharacter, so it is
3525 always safe to precede a non-alphanumeric with backslash to specify
3526 that it stands for itself. In particular, if you want to match a back-
3527 slash, you write \\.
3528
3529 In UTF-8 mode, only ASCII numbers and letters have any special meaning
3530 after a backslash. All other characters (in particular, those whose
3531 codepoints are greater than 127) are treated as literals.
3532
3533 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
3534 the pattern (other than in a character class) and characters between a
3535 # outside a character class and the next newline are ignored. An escap-
3536 ing backslash can be used to include a whitespace or # character as
3537 part of the pattern.
3538
3539 If you want to remove the special meaning from a sequence of charac-
3540 ters, you can do so by putting them between \Q and \E. This is differ-
3541 ent from Perl in that $ and @ are handled as literals in \Q...\E
3542 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
3543 tion. Note the following examples:
3544
3545 Pattern PCRE matches Perl matches
3546
3547 \Qabc$xyz\E abc$xyz abc followed by the
3548 contents of $xyz
3549 \Qabc\$xyz\E abc\$xyz abc\$xyz
3550 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3551
3552 The \Q...\E sequence is recognized both inside and outside character
3553 classes. An isolated \E that is not preceded by \Q is ignored. If \Q
3554 is not followed by \E later in the pattern, the literal interpretation
3555 continues to the end of the pattern (that is, \E is assumed at the
3556 end). If the isolated \Q is inside a character class, this causes an
3557 error, because the character class is not terminated.
3558
3559 Non-printing characters
3560
3561 A second use of backslash provides a way of encoding non-printing char-
3562 acters in patterns in a visible manner. There is no restriction on the
3563 appearance of non-printing characters, apart from the binary zero that
3564 terminates a pattern, but when a pattern is being prepared by text
3565 editing, it is often easier to use one of the following escape
3566 sequences than the binary character it represents:
3567
3568 \a alarm, that is, the BEL character (hex 07)
3569 \cx "control-x", where x is any ASCII character
3570 \e escape (hex 1B)
3571 \f formfeed (hex 0C)
3572 \n linefeed (hex 0A)
3573 \r carriage return (hex 0D)
3574 \t tab (hex 09)
3575 \ddd character with octal code ddd, or back reference
3576 \xhh character with hex code hh
3577 \x{hhh..} character with hex code hhh..
3578
3579 The precise effect of \cx is as follows: if x is a lower case letter,
3580 it is converted to upper case. Then bit 6 of the character (hex 40) is
3581 inverted. Thus \cz becomes hex 1A (z is 7A), but \c{ becomes hex 3B ({
3582 is 7B), while \c; becomes hex 7B (; is 3B). If the byte following \c
3583 has a value greater than 127, a compile-time error occurs. This locks
3584 out non-ASCII characters in both byte mode and UTF-8 mode. (When PCRE
3585 is compiled in EBCDIC mode, all byte values are valid. A lower case
3586 letter is converted to upper case, and then the 0xc0 bits are flipped.)
3587
3588 After \x, from zero to two hexadecimal digits are read (letters can be
3589 in upper or lower case). Any number of hexadecimal digits may appear
3590 between \x{ and }, but the value of the character code must be less
3591 than 256 in non-UTF-8 mode, and less than 2**31 in UTF-8 mode. That is,
3592 the maximum value in hexadecimal is 7FFFFFFF. Note that this is bigger
3593 than the largest Unicode code point, which is 10FFFF.
3594
3595 If characters other than hexadecimal digits appear between \x{ and },
3596 or if there is no terminating }, this form of escape is not recognized.
3597 Instead, the initial \x will be interpreted as a basic hexadecimal
3598 escape, with no following digits, giving a character whose value is
3599 zero.
3600
3601 Characters whose value is less than 256 can be defined by either of the
3602 two syntaxes for \x. There is no difference in the way they are han-
3603 dled. For example, \xdc is exactly the same as \x{dc}.
3604
3605 After \0 up to two further octal digits are read. If there are fewer
3606 than two digits, just those that are present are used. Thus the
3607 sequence \0\x\07 specifies two binary zeros followed by a BEL character
3608 (code value 7). Make sure you supply two digits after the initial zero
3609 if the pattern character that follows is itself an octal digit.
3610
3611 The handling of a backslash followed by a digit other than 0 is compli-
3612 cated. Outside a character class, PCRE reads it and any following dig-
3613 its as a decimal number. If the number is less than 10, or if there
3614 have been at least that many previous capturing left parentheses in the
3615 expression, the entire sequence is taken as a back reference. A
3616 description of how this works is given later, following the discussion
3617 of parenthesized subpatterns.
3618
3619 Inside a character class, or if the decimal number is greater than 9
3620 and there have not been that many capturing subpatterns, PCRE re-reads
3621 up to three octal digits following the backslash, and uses them to gen-
3622 erate a data character. Any subsequent digits stand for themselves. In
3623 non-UTF-8 mode, the value of a character specified in octal must be
3624 less than \400. In UTF-8 mode, values up to \777 are permitted. For
3625 example:
3626
3627 \040 is another way of writing a space
3628 \40 is the same, provided there are fewer than 40
3629 previous capturing subpatterns
3630 \7 is always a back reference
3631 \11 might be a back reference, or another way of
3632 writing a tab
3633 \011 is always a tab
3634 \0113 is a tab followed by the character "3"
3635 \113 might be a back reference, otherwise the
3636 character with octal code 113
3637 \377 might be a back reference, otherwise
3638 the byte consisting entirely of 1 bits
3639 \81 is either a back reference, or a binary zero
3640 followed by the two characters "8" and "1"
3641
3642 Note that octal values of 100 or greater must not be introduced by a
3643 leading zero, because no more than three octal digits are ever read.
3644
3645 All the sequences that define a single character value can be used both
3646 inside and outside character classes. In addition, inside a character
3647 class, the sequence \b is interpreted as the backspace character (hex
3648 08). The sequences \B, \N, \R, and \X are not special inside a charac-
3649 ter class. Like any other unrecognized escape sequences, they are
3650 treated as the literal characters "B", "N", "R", and "X" by default,
3651 but cause an error if the PCRE_EXTRA option is set. Outside a character
3652 class, these sequences have different meanings.
3653
3654 Absolute and relative back references
3655
3656 The sequence \g followed by an unsigned or a negative number, option-
3657 ally enclosed in braces, is an absolute or relative back reference. A
3658 named back reference can be coded as \g{name}. Back references are dis-
3659 cussed later, following the discussion of parenthesized subpatterns.
3660
3661 Absolute and relative subroutine calls
3662
3663 For compatibility with Oniguruma, the non-Perl syntax \g followed by a
3664 name or a number enclosed either in angle brackets or single quotes, is
3665 an alternative syntax for referencing a subpattern as a "subroutine".
3666 Details are discussed later. Note that \g{...} (Perl syntax) and
3667 \g<...> (Oniguruma syntax) are not synonymous. The former is a back
3668 reference; the latter is a subroutine call.
3669
3670 Generic character types
3671
3672 Another use of backslash is for specifying generic character types:
3673
3674 \d any decimal digit
3675 \D any character that is not a decimal digit
3676 \h any horizontal whitespace character
3677 \H any character that is not a horizontal whitespace character
3678 \s any whitespace character
3679 \S any character that is not a whitespace character
3680 \v any vertical whitespace character
3681 \V any character that is not a vertical whitespace character
3682 \w any "word" character
3683 \W any "non-word" character
3684
3685 There is also the single sequence \N, which matches a non-newline char-
3686 acter. This is the same as the "." metacharacter when PCRE_DOTALL is
3687 not set.
3688
3689 Each pair of lower and upper case escape sequences partitions the com-
3690 plete set of characters into two disjoint sets. Any given character
3691 matches one, and only one, of each pair. The sequences can appear both
3692 inside and outside character classes. They each match one character of
3693 the appropriate type. If the current matching point is at the end of
3694 the subject string, all of them fail, because there is no character to
3695 match.
3696
3697 For compatibility with Perl, \s does not match the VT character (code
3698 11). This makes it different from the the POSIX "space" class. The \s
3699 characters are HT (9), LF (10), FF (12), CR (13), and space (32). If
3700 "use locale;" is included in a Perl script, \s may match the VT charac-
3701 ter. In PCRE, it never does.
3702
3703 A "word" character is an underscore or any character that is a letter
3704 or digit. By default, the definition of letters and digits is con-
3705 trolled by PCRE's low-valued character tables, and may vary if locale-
3706 specific matching is taking place (see "Locale support" in the pcreapi
3707 page). For example, in a French locale such as "fr_FR" in Unix-like
3708 systems, or "french" in Windows, some character codes greater than 128
3709 are used for accented letters, and these are then matched by \w. The
3710 use of locales with Unicode is discouraged.
3711
3712 By default, in UTF-8 mode, characters with values greater than 128
3713 never match \d, \s, or \w, and always match \D, \S, and \W. These
3714 sequences retain their original meanings from before UTF-8 support was
3715 available, mainly for efficiency reasons. However, if PCRE is compiled
3716 with Unicode property support, and the PCRE_UCP option is set, the be-
3717 haviour is changed so that Unicode properties are used to determine
3718 character types, as follows:
3719
3720 \d any character that \p{Nd} matches (decimal digit)
3721 \s any character that \p{Z} matches, plus HT, LF, FF, CR
3722 \w any character that \p{L} or \p{N} matches, plus underscore
3723
3724 The upper case escapes match the inverse sets of characters. Note that
3725 \d matches only decimal digits, whereas \w matches any Unicode digit,
3726 as well as any Unicode letter, and underscore. Note also that PCRE_UCP
3727 affects \b, and \B because they are defined in terms of \w and \W.
3728 Matching these sequences is noticeably slower when PCRE_UCP is set.
3729
3730 The sequences \h, \H, \v, and \V are features that were added to Perl
3731 at release 5.10. In contrast to the other sequences, which match only
3732 ASCII characters by default, these always match certain high-valued
3733 codepoints in UTF-8 mode, whether or not PCRE_UCP is set. The horizon-
3734 tal space characters are:
3735
3736 U+0009 Horizontal tab
3737 U+0020 Space
3738 U+00A0 Non-break space
3739 U+1680 Ogham space mark
3740 U+180E Mongolian vowel separator
3741 U+2000 En quad
3742 U+2001 Em quad
3743 U+2002 En space
3744 U+2003 Em space
3745 U+2004 Three-per-em space
3746 U+2005 Four-per-em space
3747 U+2006 Six-per-em space
3748 U+2007 Figure space
3749 U+2008 Punctuation space
3750 U+2009 Thin space
3751 U+200A Hair space
3752 U+202F Narrow no-break space
3753 U+205F Medium mathematical space
3754 U+3000 Ideographic space
3755
3756 The vertical space characters are:
3757
3758 U+000A Linefeed
3759 U+000B Vertical tab
3760 U+000C Formfeed
3761 U+000D Carriage return
3762 U+0085 Next line
3763 U+2028 Line separator
3764 U+2029 Paragraph separator
3765
3766 Newline sequences
3767
3768 Outside a character class, by default, the escape sequence \R matches
3769 any Unicode newline sequence. In non-UTF-8 mode \R is equivalent to the
3770 following:
3771
3772 (?>\r\n|\n|\x0b|\f|\r|\x85)
3773
3774 This is an example of an "atomic group", details of which are given
3775 below. This particular group matches either the two-character sequence
3776 CR followed by LF, or one of the single characters LF (linefeed,
3777 U+000A), VT (vertical tab, U+000B), FF (formfeed, U+000C), CR (carriage
3778 return, U+000D), or NEL (next line, U+0085). The two-character sequence
3779 is treated as a single unit that cannot be split.
3780
3781 In UTF-8 mode, two additional characters whose codepoints are greater
3782 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
3783 rator, U+2029). Unicode character property support is not needed for
3784 these characters to be recognized.
3785
3786 It is possible to restrict \R to match only CR, LF, or CRLF (instead of
3787 the complete set of Unicode line endings) by setting the option
3788 PCRE_BSR_ANYCRLF either at compile time or when the pattern is matched.
3789 (BSR is an abbrevation for "backslash R".) This can be made the default
3790 when PCRE is built; if this is the case, the other behaviour can be
3791 requested via the PCRE_BSR_UNICODE option. It is also possible to
3792 specify these settings by starting a pattern string with one of the
3793 following sequences:
3794
3795 (*BSR_ANYCRLF) CR, LF, or CRLF only
3796 (*BSR_UNICODE) any Unicode newline sequence
3797
3798 These override the default and the options given to pcre_compile() or
3799 pcre_compile2(), but they can be overridden by options given to
3800 pcre_exec() or pcre_dfa_exec(). Note that these special settings, which
3801 are not Perl-compatible, are recognized only at the very start of a
3802 pattern, and that they must be in upper case. If more than one of them
3803 is present, the last one is used. They can be combined with a change of
3804 newline convention; for example, a pattern can start with:
3805
3806 (*ANY)(*BSR_ANYCRLF)
3807
3808 They can also be combined with the (*UTF8) or (*UCP) special sequences.
3809 Inside a character class, \R is treated as an unrecognized escape
3810 sequence, and so matches the letter "R" by default, but causes an error
3811 if PCRE_EXTRA is set.
3812
3813 Unicode character properties
3814
3815 When PCRE is built with Unicode character property support, three addi-
3816 tional escape sequences that match characters with specific properties
3817 are available. When not in UTF-8 mode, these sequences are of course
3818 limited to testing characters whose codepoints are less than 256, but
3819 they do work in this mode. The extra escape sequences are:
3820
3821 \p{xx} a character with the xx property
3822 \P{xx} a character without the xx property
3823 \X an extended Unicode sequence
3824
3825 The property names represented by xx above are limited to the Unicode
3826 script names, the general category properties, "Any", which matches any
3827 character (including newline), and some special PCRE properties
3828 (described in the next section). Other Perl properties such as "InMu-
3829 sicalSymbols" are not currently supported by PCRE. Note that \P{Any}
3830 does not match any characters, so always causes a match failure.
3831
3832 Sets of Unicode characters are defined as belonging to certain scripts.
3833 A character from one of these sets can be matched using a script name.
3834 For example:
3835
3836 \p{Greek}
3837 \P{Han}
3838
3839 Those that are not part of an identified script are lumped together as
3840 "Common". The current list of scripts is:
3841
3842 Arabic, Armenian, Avestan, Balinese, Bamum, Bengali, Bopomofo, Braille,
3843 Buginese, Buhid, Canadian_Aboriginal, Carian, Cham, Cherokee, Common,
3844 Coptic, Cuneiform, Cypriot, Cyrillic, Deseret, Devanagari, Egyp-
3845 tian_Hieroglyphs, Ethiopic, Georgian, Glagolitic, Gothic, Greek,
3846 Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hiragana, Impe-
3847 rial_Aramaic, Inherited, Inscriptional_Pahlavi, Inscriptional_Parthian,
3848 Javanese, Kaithi, Kannada, Katakana, Kayah_Li, Kharoshthi, Khmer, Lao,
3849 Latin, Lepcha, Limbu, Linear_B, Lisu, Lycian, Lydian, Malayalam,
3850 Meetei_Mayek, Mongolian, Myanmar, New_Tai_Lue, Nko, Ogham, Old_Italic,
3851 Old_Persian, Old_South_Arabian, Old_Turkic, Ol_Chiki, Oriya, Osmanya,
3852 Phags_Pa, Phoenician, Rejang, Runic, Samaritan, Saurashtra, Shavian,
3853 Sinhala, Sundanese, Syloti_Nagri, Syriac, Tagalog, Tagbanwa, Tai_Le,
3854 Tai_Tham, Tai_Viet, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh,
3855 Ugaritic, Vai, Yi.
3856
3857 Each character has exactly one Unicode general category property, spec-
3858 ified by a two-letter abbreviation. For compatibility with Perl, nega-
3859 tion can be specified by including a circumflex between the opening
3860 brace and the property name. For example, \p{^Lu} is the same as
3861 \P{Lu}.
3862
3863 If only one letter is specified with \p or \P, it includes all the gen-
3864 eral category properties that start with that letter. In this case, in
3865 the absence of negation, the curly brackets in the escape sequence are
3866 optional; these two examples have the same effect:
3867
3868 \p{L}
3869 \pL
3870
3871 The following general category property codes are supported:
3872
3873 C Other
3874 Cc Control
3875 Cf Format
3876 Cn Unassigned
3877 Co Private use
3878 Cs Surrogate
3879
3880 L Letter
3881 Ll Lower case letter
3882 Lm Modifier letter
3883 Lo Other letter
3884 Lt Title case letter
3885 Lu Upper case letter
3886
3887 M Mark
3888 Mc Spacing mark
3889 Me Enclosing mark
3890 Mn Non-spacing mark
3891
3892 N Number
3893 Nd Decimal number
3894 Nl Letter number
3895 No Other number
3896
3897 P Punctuation
3898 Pc Connector punctuation
3899 Pd Dash punctuation
3900 Pe Close punctuation
3901 Pf Final punctuation
3902 Pi Initial punctuation
3903 Po Other punctuation
3904 Ps Open punctuation
3905
3906 S Symbol
3907 Sc Currency symbol
3908 Sk Modifier symbol
3909 Sm Mathematical symbol
3910 So Other symbol
3911
3912 Z Separator
3913 Zl Line separator
3914 Zp Paragraph separator
3915 Zs Space separator
3916
3917 The special property L& is also supported: it matches a character that
3918 has the Lu, Ll, or Lt property, in other words, a letter that is not
3919 classified as a modifier or "other".
3920
3921 The Cs (Surrogate) property applies only to characters in the range
3922 U+D800 to U+DFFF. Such characters are not valid in UTF-8 strings (see
3923 RFC 3629) and so cannot be tested by PCRE, unless UTF-8 validity check-
3924 ing has been turned off (see the discussion of PCRE_NO_UTF8_CHECK in
3925 the pcreapi page). Perl does not support the Cs property.
3926
3927 The long synonyms for property names that Perl supports (such as
3928 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
3929 any of these properties with "Is".
3930
3931 No character that is in the Unicode table has the Cn (unassigned) prop-
3932 erty. Instead, this property is assumed for any code point that is not
3933 in the Unicode table.
3934
3935 Specifying caseless matching does not affect these escape sequences.
3936 For example, \p{Lu} always matches only upper case letters.
3937
3938 The \X escape matches any number of Unicode characters that form an
3939 extended Unicode sequence. \X is equivalent to
3940
3941 (?>\PM\pM*)
3942
3943 That is, it matches a character without the "mark" property, followed
3944 by zero or more characters with the "mark" property, and treats the
3945 sequence as an atomic group (see below). Characters with the "mark"
3946 property are typically accents that affect the preceding character.
3947 None of them have codepoints less than 256, so in non-UTF-8 mode \X
3948 matches any one character.
3949
3950 Note that recent versions of Perl have changed \X to match what Unicode
3951 calls an "extended grapheme cluster", which has a more complicated def-
3952 inition.
3953
3954 Matching characters by Unicode property is not fast, because PCRE has
3955 to search a structure that contains data for over fifteen thousand
3956 characters. That is why the traditional escape sequences such as \d and
3957 \w do not use Unicode properties in PCRE by default, though you can
3958 make them do so by setting the PCRE_UCP option for pcre_compile() or by
3959 starting the pattern with (*UCP).
3960
3961 PCRE's additional properties
3962
3963 As well as the standard Unicode properties described in the previous
3964 section, PCRE supports four more that make it possible to convert tra-
3965 ditional escape sequences such as \w and \s and POSIX character classes
3966 to use Unicode properties. PCRE uses these non-standard, non-Perl prop-
3967 erties internally when PCRE_UCP is set. They are:
3968
3969 Xan Any alphanumeric character
3970 Xps Any POSIX space character
3971 Xsp Any Perl space character
3972 Xwd Any Perl "word" character
3973
3974 Xan matches characters that have either the L (letter) or the N (num-
3975 ber) property. Xps matches the characters tab, linefeed, vertical tab,
3976 formfeed, or carriage return, and any other character that has the Z
3977 (separator) property. Xsp is the same as Xps, except that vertical tab
3978 is excluded. Xwd matches the same characters as Xan, plus underscore.
3979
3980 Resetting the match start
3981
3982 The escape sequence \K causes any previously matched characters not to
3983 be included in the final matched sequence. For example, the pattern:
3984
3985 foo\Kbar
3986
3987 matches "foobar", but reports that it has matched "bar". This feature
3988 is similar to a lookbehind assertion (described below). However, in
3989 this case, the part of the subject before the real match does not have
3990 to be of fixed length, as lookbehind assertions do. The use of \K does
3991 not interfere with the setting of captured substrings. For example,
3992 when the pattern
3993
3994 (foo)\Kbar
3995
3996 matches "foobar", the first substring is still set to "foo".
3997
3998 Perl documents that the use of \K within assertions is "not well
3999 defined". In PCRE, \K is acted upon when it occurs inside positive
4000 assertions, but is ignored in negative assertions.
4001
4002 Simple assertions
4003
4004 The final use of backslash is for certain simple assertions. An asser-
4005 tion specifies a condition that has to be met at a particular point in
4006 a match, without consuming any characters from the subject string. The
4007 use of subpatterns for more complicated assertions is described below.
4008 The backslashed assertions are:
4009
4010 \b matches at a word boundary
4011 \B matches when not at a word boundary
4012 \A matches at the start of the subject
4013 \Z matches at the end of the subject
4014 also matches before a newline at the end of the subject
4015 \z matches only at the end of the subject
4016 \G matches at the first matching position in the subject
4017
4018 Inside a character class, \b has a different meaning; it matches the
4019 backspace character. If any other of these assertions appears in a
4020 character class, by default it matches the corresponding literal char-
4021 acter (for example, \B matches the letter B). However, if the
4022 PCRE_EXTRA option is set, an "invalid escape sequence" error is gener-
4023 ated instead.
4024
4025 A word boundary is a position in the subject string where the current
4026 character and the previous character do not both match \w or \W (i.e.
4027 one matches \w and the other matches \W), or the start or end of the
4028 string if the first or last character matches \w, respectively. In
4029 UTF-8 mode, the meanings of \w and \W can be changed by setting the
4030 PCRE_UCP option. When this is done, it also affects \b and \B. Neither
4031 PCRE nor Perl has a separate "start of word" or "end of word" metase-
4032 quence. However, whatever follows \b normally determines which it is.
4033 For example, the fragment \ba matches "a" at the start of a word.
4034
4035 The \A, \Z, and \z assertions differ from the traditional circumflex
4036 and dollar (described in the next section) in that they only ever match
4037 at the very start and end of the subject string, whatever options are
4038 set. Thus, they are independent of multiline mode. These three asser-
4039 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
4040 affect only the behaviour of the circumflex and dollar metacharacters.
4041 However, if the startoffset argument of pcre_exec() is non-zero, indi-
4042 cating that matching is to start at a point other than the beginning of
4043 the subject, \A can never match. The difference between \Z and \z is
4044 that \Z matches before a newline at the end of the string as well as at
4045 the very end, whereas \z matches only at the end.
4046
4047 The \G assertion is true only when the current matching position is at
4048 the start point of the match, as specified by the startoffset argument
4049 of pcre_exec(). It differs from \A when the value of startoffset is
4050 non-zero. By calling pcre_exec() multiple times with appropriate argu-
4051 ments, you can mimic Perl's /g option, and it is in this kind of imple-
4052 mentation where \G can be useful.
4053
4054 Note, however, that PCRE's interpretation of \G, as the start of the
4055 current match, is subtly different from Perl's, which defines it as the
4056 end of the previous match. In Perl, these can be different when the
4057 previously matched string was empty. Because PCRE does just one match
4058 at a time, it cannot reproduce this behaviour.
4059
4060 If all the alternatives of a pattern begin with \G, the expression is
4061 anchored to the starting match position, and the "anchored" flag is set
4062 in the compiled regular expression.
4063
4064
4065 CIRCUMFLEX AND DOLLAR
4066
4067 Outside a character class, in the default matching mode, the circumflex
4068 character is an assertion that is true only if the current matching
4069 point is at the start of the subject string. If the startoffset argu-
4070 ment of pcre_exec() is non-zero, circumflex can never match if the
4071 PCRE_MULTILINE option is unset. Inside a character class, circumflex
4072 has an entirely different meaning (see below).
4073
4074 Circumflex need not be the first character of the pattern if a number
4075 of alternatives are involved, but it should be the first thing in each
4076 alternative in which it appears if the pattern is ever to match that
4077 branch. If all possible alternatives start with a circumflex, that is,
4078 if the pattern is constrained to match only at the start of the sub-
4079 ject, it is said to be an "anchored" pattern. (There are also other
4080 constructs that can cause a pattern to be anchored.)
4081
4082 A dollar character is an assertion that is true only if the current
4083 matching point is at the end of the subject string, or immediately
4084 before a newline at the end of the string (by default). Dollar need not
4085 be the last character of the pattern if a number of alternatives are
4086 involved, but it should be the last item in any branch in which it
4087 appears. Dollar has no special meaning in a character class.
4088
4089 The meaning of dollar can be changed so that it matches only at the
4090 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
4091 compile time. This does not affect the \Z assertion.
4092
4093 The meanings of the circumflex and dollar characters are changed if the
4094 PCRE_MULTILINE option is set. When this is the case, a circumflex
4095 matches immediately after internal newlines as well as at the start of
4096 the subject string. It does not match after a newline that ends the
4097 string. A dollar matches before any newlines in the string, as well as
4098 at the very end, when PCRE_MULTILINE is set. When newline is specified
4099 as the two-character sequence CRLF, isolated CR and LF characters do
4100 not indicate newlines.
4101
4102 For example, the pattern /^abc$/ matches the subject string "def\nabc"
4103 (where \n represents a newline) in multiline mode, but not otherwise.
4104 Consequently, patterns that are anchored in single line mode because
4105 all branches start with ^ are not anchored in multiline mode, and a
4106 match for circumflex is possible when the startoffset argument of
4107 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
4108 PCRE_MULTILINE is set.
4109
4110 Note that the sequences \A, \Z, and \z can be used to match the start
4111 and end of the subject in both modes, and if all branches of a pattern
4112 start with \A it is always anchored, whether or not PCRE_MULTILINE is
4113 set.
4114
4115
4116 FULL STOP (PERIOD, DOT) AND \N
4117
4118 Outside a character class, a dot in the pattern matches any one charac-
4119 ter in the subject string except (by default) a character that signi-
4120 fies the end of a line. In UTF-8 mode, the matched character may be
4121 more than one byte long.
4122
4123 When a line ending is defined as a single character, dot never matches
4124 that character; when the two-character sequence CRLF is used, dot does
4125 not match CR if it is immediately followed by LF, but otherwise it
4126 matches all characters (including isolated CRs and LFs). When any Uni-
4127 code line endings are being recognized, dot does not match CR or LF or
4128 any of the other line ending characters.
4129
4130 The behaviour of dot with regard to newlines can be changed. If the
4131 PCRE_DOTALL option is set, a dot matches any one character, without
4132 exception. If the two-character sequence CRLF is present in the subject
4133 string, it takes two dots to match it.
4134
4135 The handling of dot is entirely independent of the handling of circum-
4136 flex and dollar, the only relationship being that they both involve
4137 newlines. Dot has no special meaning in a character class.
4138
4139 The escape sequence \N behaves like a dot, except that it is not
4140 affected by the PCRE_DOTALL option. In other words, it matches any
4141 character except one that signifies the end of a line.
4142
4143
4144 MATCHING A SINGLE BYTE
4145
4146 Outside a character class, the escape sequence \C matches any one byte,
4147 both in and out of UTF-8 mode. Unlike a dot, it always matches any
4148 line-ending characters. The feature is provided in Perl in order to
4149 match individual bytes in UTF-8 mode. Because it breaks up UTF-8 char-
4150 acters into individual bytes, the rest of the string may start with a
4151 malformed UTF-8 character. For this reason, the \C escape sequence is
4152 best avoided.
4153
4154 PCRE does not allow \C to appear in lookbehind assertions (described
4155 below), because in UTF-8 mode this would make it impossible to calcu-
4156 late the length of the lookbehind.
4157
4158
4159 SQUARE BRACKETS AND CHARACTER CLASSES
4160
4161 An opening square bracket introduces a character class, terminated by a
4162 closing square bracket. A closing square bracket on its own is not spe-
4163 cial by default. However, if the PCRE_JAVASCRIPT_COMPAT option is set,
4164 a lone closing square bracket causes a compile-time error. If a closing
4165 square bracket is required as a member of the class, it should be the
4166 first data character in the class (after an initial circumflex, if
4167 present) or escaped with a backslash.
4168
4169 A character class matches a single character in the subject. In UTF-8
4170 mode, the character may be more than one byte long. A matched character
4171 must be in the set of characters defined by the class, unless the first
4172 character in the class definition is a circumflex, in which case the
4173 subject character must not be in the set defined by the class. If a
4174 circumflex is actually required as a member of the class, ensure it is
4175 not the first character, or escape it with a backslash.
4176
4177 For example, the character class [aeiou] matches any lower case vowel,
4178 while [^aeiou] matches any character that is not a lower case vowel.
4179 Note that a circumflex is just a convenient notation for specifying the
4180 characters that are in the class by enumerating those that are not. A
4181 class that starts with a circumflex is not an assertion; it still con-
4182 sumes a character from the subject string, and therefore it fails if
4183 the current pointer is at the end of the string.
4184
4185 In UTF-8 mode, characters with values greater than 255 can be included
4186 in a class as a literal string of bytes, or by using the \x{ escaping
4187 mechanism.
4188
4189 When caseless matching is set, any letters in a class represent both
4190 their upper case and lower case versions, so for example, a caseless
4191 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
4192 match "A", whereas a caseful version would. In UTF-8 mode, PCRE always
4193 understands the concept of case for characters whose values are less
4194 than 128, so caseless matching is always possible. For characters with
4195 higher values, the concept of case is supported if PCRE is compiled
4196 with Unicode property support, but not otherwise. If you want to use
4197 caseless matching in UTF8-mode for characters 128 and above, you must
4198 ensure that PCRE is compiled with Unicode property support as well as
4199 with UTF-8 support.
4200
4201 Characters that might indicate line breaks are never treated in any
4202 special way when matching character classes, whatever line-ending
4203 sequence is in use, and whatever setting of the PCRE_DOTALL and
4204 PCRE_MULTILINE options is used. A class such as [^a] always matches one
4205 of these characters.
4206
4207 The minus (hyphen) character can be used to specify a range of charac-
4208 ters in a character class. For example, [d-m] matches any letter
4209 between d and m, inclusive. If a minus character is required in a
4210 class, it must be escaped with a backslash or appear in a position
4211 where it cannot be interpreted as indicating a range, typically as the
4212 first or last character in the class.
4213
4214 It is not possible to have the literal character "]" as the end charac-
4215 ter of a range. A pattern such as [W-]46] is interpreted as a class of
4216 two characters ("W" and "-") followed by a literal string "46]", so it
4217 would match "W46]" or "-46]". However, if the "]" is escaped with a
4218 backslash it is interpreted as the end of range, so [W-\]46] is inter-
4219 preted as a class containing a range followed by two other characters.
4220 The octal or hexadecimal representation of "]" can also be used to end
4221 a range.
4222
4223 Ranges operate in the collating sequence of character values. They can
4224 also be used for characters specified numerically, for example
4225 [\000-\037]. In UTF-8 mode, ranges can include characters whose values
4226 are greater than 255, for example [\x{100}-\x{2ff}].
4227
4228 If a range that includes letters is used when caseless matching is set,
4229 it matches the letters in either case. For example, [W-c] is equivalent
4230 to [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if
4231 character tables for a French locale are in use, [\xc8-\xcb] matches
4232 accented E characters in both cases. In UTF-8 mode, PCRE supports the
4233 concept of case for characters with values greater than 128 only when
4234 it is compiled with Unicode property support.
4235
4236 The character escape sequences \d, \D, \h, \H, \p, \P, \s, \S, \v, \V,
4237 \w, and \W may appear in a character class, and add the characters that
4238 they match to the class. For example, [\dABCDEF] matches any hexadeci-
4239 mal digit. In UTF-8 mode, the PCRE_UCP option affects the meanings of
4240 \d, \s, \w and their upper case partners, just as it does when they
4241 appear outside a character class, as described in the section entitled
4242 "Generic character types" above. The escape sequence \b has a different
4243 meaning inside a character class; it matches the backspace character.
4244 The sequences \B, \N, \R, and \X are not special inside a character
4245 class. Like any other unrecognized escape sequences, they are treated
4246 as the literal characters "B", "N", "R", and "X" by default, but cause
4247 an error if the PCRE_EXTRA option is set.
4248
4249 A circumflex can conveniently be used with the upper case character
4250 types to specify a more restricted set of characters than the matching
4251 lower case type. For example, the class [^\W_] matches any letter or
4252 digit, but not underscore, whereas [\w] includes underscore. A positive
4253 character class should be read as "something OR something OR ..." and a
4254 negative class as "NOT something AND NOT something AND NOT ...".
4255
4256 The only metacharacters that are recognized in character classes are
4257 backslash, hyphen (only where it can be interpreted as specifying a
4258 range), circumflex (only at the start), opening square bracket (only
4259 when it can be interpreted as introducing a POSIX class name - see the
4260 next section), and the terminating closing square bracket. However,
4261 escaping other non-alphanumeric characters does no harm.
4262
4263
4264 POSIX CHARACTER CLASSES
4265
4266 Perl supports the POSIX notation for character classes. This uses names
4267 enclosed by [: and :] within the enclosing square brackets. PCRE also
4268 supports this notation. For example,
4269
4270 [01[:alpha:]%]
4271
4272 matches "0", "1", any alphabetic character, or "%". The supported class
4273 names are:
4274
4275 alnum letters and digits
4276 alpha letters
4277 ascii character codes 0 - 127
4278 blank space or tab only
4279 cntrl control characters
4280 digit decimal digits (same as \d)
4281 graph printing characters, excluding space
4282 lower lower case letters
4283 print printing characters, including space
4284 punct printing characters, excluding letters and digits and space
4285 space white space (not quite the same as \s)
4286 upper upper case letters
4287 word "word" characters (same as \w)
4288 xdigit hexadecimal digits
4289
4290 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
4291 and space (32). Notice that this list includes the VT character (code
4292 11). This makes "space" different to \s, which does not include VT (for
4293 Perl compatibility).
4294
4295 The name "word" is a Perl extension, and "blank" is a GNU extension
4296 from Perl 5.8. Another Perl extension is negation, which is indicated
4297 by a ^ character after the colon. For example,
4298
4299 [12[:^digit:]]
4300
4301 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
4302 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
4303 these are not supported, and an error is given if they are encountered.
4304
4305 By default, in UTF-8 mode, characters with values greater than 128 do
4306 not match any of the POSIX character classes. However, if the PCRE_UCP
4307 option is passed to pcre_compile(), some of the classes are changed so
4308 that Unicode character properties are used. This is achieved by replac-
4309 ing the POSIX classes by other sequences, as follows:
4310
4311 [:alnum:] becomes \p{Xan}
4312 [:alpha:] becomes \p{L}
4313 [:blank:] becomes \h
4314 [:digit:] becomes \p{Nd}
4315 [:lower:] becomes \p{Ll}
4316 [:space:] becomes \p{Xps}
4317 [:upper:] becomes \p{Lu}
4318 [:word:] becomes \p{Xwd}
4319
4320 Negated versions, such as [:^alpha:] use \P instead of \p. The other
4321 POSIX classes are unchanged, and match only characters with code points
4322 less than 128.
4323
4324
4325 VERTICAL BAR
4326
4327 Vertical bar characters are used to separate alternative patterns. For
4328 example, the pattern
4329
4330 gilbert|sullivan
4331
4332 matches either "gilbert" or "sullivan". Any number of alternatives may
4333 appear, and an empty alternative is permitted (matching the empty
4334 string). The matching process tries each alternative in turn, from left
4335 to right, and the first one that succeeds is used. If the alternatives
4336 are within a subpattern (defined below), "succeeds" means matching the
4337 rest of the main pattern as well as the alternative in the subpattern.
4338
4339
4340 INTERNAL OPTION SETTING
4341
4342 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
4343 PCRE_EXTENDED options (which are Perl-compatible) can be changed from
4344 within the pattern by a sequence of Perl option letters enclosed
4345 between "(?" and ")". The option letters are
4346
4347 i for PCRE_CASELESS
4348 m for PCRE_MULTILINE
4349 s for PCRE_DOTALL
4350 x for PCRE_EXTENDED
4351
4352 For example, (?im) sets caseless, multiline matching. It is also possi-
4353 ble to unset these options by preceding the letter with a hyphen, and a
4354 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
4355 LESS and PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED,
4356 is also permitted. If a letter appears both before and after the
4357 hyphen, the option is unset.
4358
4359 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA
4360 can be changed in the same way as the Perl-compatible options by using
4361 the characters J, U and X respectively.
4362
4363 When one of these option changes occurs at top level (that is, not
4364 inside subpattern parentheses), the change applies to the remainder of
4365 the pattern that follows. If the change is placed right at the start of
4366 a pattern, PCRE extracts it into the global options (and it will there-
4367 fore show up in data extracted by the pcre_fullinfo() function).
4368
4369 An option change within a subpattern (see below for a description of
4370 subpatterns) affects only that part of the subpattern that follows it,
4371 so
4372
4373 (a(?i)b)c
4374
4375 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
4376 used). By this means, options can be made to have different settings
4377 in different parts of the pattern. Any changes made in one alternative
4378 do carry on into subsequent branches within the same subpattern. For
4379 example,
4380
4381 (a(?i)b|c)
4382
4383 matches "ab", "aB", "c", and "C", even though when matching "C" the
4384 first branch is abandoned before the option setting. This is because
4385 the effects of option settings happen at compile time. There would be
4386 some very weird behaviour otherwise.
4387
4388 Note: There are other PCRE-specific options that can be set by the
4389 application when the compile or match functions are called. In some
4390 cases the pattern can contain special leading sequences such as (*CRLF)
4391 to override what the application has set or what has been defaulted.
4392 Details are given in the section entitled "Newline sequences" above.
4393 There are also the (*UTF8) and (*UCP) leading sequences that can be
4394 used to set UTF-8 and Unicode property modes; they are equivalent to
4395 setting the PCRE_UTF8 and the PCRE_UCP options, respectively.
4396
4397
4398 SUBPATTERNS
4399
4400 Subpatterns are delimited by parentheses (round brackets), which can be
4401 nested. Turning part of a pattern into a subpattern does two things:
4402
4403 1. It localizes a set of alternatives. For example, the pattern
4404
4405 cat(aract|erpillar|)
4406
4407 matches "cataract", "caterpillar", or "cat". Without the parentheses,
4408 it would match "cataract", "erpillar" or an empty string.
4409
4410 2. It sets up the subpattern as a capturing subpattern. This means
4411 that, when the whole pattern matches, that portion of the subject
4412 string that matched the subpattern is passed back to the caller via the
4413 ovector argument of pcre_exec(). Opening parentheses are counted from
4414 left to right (starting from 1) to obtain numbers for the capturing
4415 subpatterns. For example, if the string "the red king" is matched
4416 against the pattern
4417
4418 the ((red|white) (king|queen))
4419
4420 the captured substrings are "red king", "red", and "king", and are num-
4421 bered 1, 2, and 3, respectively.
4422
4423 The fact that plain parentheses fulfil two functions is not always
4424 helpful. There are often times when a grouping subpattern is required
4425 without a capturing requirement. If an opening parenthesis is followed
4426 by a question mark and a colon, the subpattern does not do any captur-
4427 ing, and is not counted when computing the number of any subsequent
4428 capturing subpatterns. For example, if the string "the white queen" is
4429 matched against the pattern
4430
4431 the ((?:red|white) (king|queen))
4432
4433 the captured substrings are "white queen" and "queen", and are numbered
4434 1 and 2. The maximum number of capturing subpatterns is 65535.
4435
4436 As a convenient shorthand, if any option settings are required at the
4437 start of a non-capturing subpattern, the option letters may appear
4438 between the "?" and the ":". Thus the two patterns
4439
4440 (?i:saturday|sunday)
4441 (?:(?i)saturday|sunday)
4442
4443 match exactly the same set of strings. Because alternative branches are
4444 tried from left to right, and options are not reset until the end of
4445 the subpattern is reached, an option setting in one branch does affect
4446 subsequent branches, so the above patterns match "SUNDAY" as well as
4447 "Saturday".
4448
4449
4450 DUPLICATE SUBPATTERN NUMBERS
4451
4452 Perl 5.10 introduced a feature whereby each alternative in a subpattern
4453 uses the same numbers for its capturing parentheses. Such a subpattern
4454 starts with (?| and is itself a non-capturing subpattern. For example,
4455 consider this pattern:
4456
4457 (?|(Sat)ur|(Sun))day
4458
4459 Because the two alternatives are inside a (?| group, both sets of cap-
4460 turing parentheses are numbered one. Thus, when the pattern matches,
4461 you can look at captured substring number one, whichever alternative
4462 matched. This construct is useful when you want to capture part, but
4463 not all, of one of a number of alternatives. Inside a (?| group, paren-
4464 theses are numbered as usual, but the number is reset at the start of
4465 each branch. The numbers of any capturing parentheses that follow the
4466 subpattern start after the highest number used in any branch. The fol-
4467 lowing example is taken from the Perl documentation. The numbers under-
4468 neath show in which buffer the captured content will be stored.
4469
4470 # before ---------------branch-reset----------- after
4471 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
4472 # 1 2 2 3 2 3 4
4473
4474 A back reference to a numbered subpattern uses the most recent value
4475 that is set for that number by any subpattern. The following pattern
4476 matches "abcabc" or "defdef":
4477
4478 /(?|(abc)|(def))\1/
4479
4480 In contrast, a recursive or "subroutine" call to a numbered subpattern
4481 always refers to the first one in the pattern with the given number.
4482 The following pattern matches "abcabc" or "defabc":
4483
4484 /(?|(abc)|(def))(?1)/
4485
4486 If a condition test for a subpattern's having matched refers to a non-
4487 unique number, the test is true if any of the subpatterns of that num-
4488 ber have matched.
4489
4490 An alternative approach to using this "branch reset" feature is to use
4491 duplicate named subpatterns, as described in the next section.
4492
4493
4494 NAMED SUBPATTERNS
4495
4496 Identifying capturing parentheses by number is simple, but it can be
4497 very hard to keep track of the numbers in complicated regular expres-
4498 sions. Furthermore, if an expression is modified, the numbers may
4499 change. To help with this difficulty, PCRE supports the naming of sub-
4500 patterns. This feature was not added to Perl until release 5.10. Python
4501 had the feature earlier, and PCRE introduced it at release 4.0, using
4502 the Python syntax. PCRE now supports both the Perl and the Python syn-
4503 tax. Perl allows identically numbered subpatterns to have different
4504 names, but PCRE does not.
4505
4506 In PCRE, a subpattern can be named in one of three ways: (?<name>...)
4507 or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
4508 to capturing parentheses from other parts of the pattern, such as back
4509 references, recursion, and conditions, can be made by name as well as
4510 by number.
4511
4512 Names consist of up to 32 alphanumeric characters and underscores.
4513 Named capturing parentheses are still allocated numbers as well as
4514 names, exactly as if the names were not present. The PCRE API provides
4515 function calls for extracting the name-to-number translation table from
4516 a compiled pattern. There is also a convenience function for extracting
4517 a captured substring by name.
4518
4519 By default, a name must be unique within a pattern, but it is possible
4520 to relax this constraint by setting the PCRE_DUPNAMES option at compile
4521 time. (Duplicate names are also always permitted for subpatterns with
4522 the same number, set up as described in the previous section.) Dupli-
4523 cate names can be useful for patterns where only one instance of the
4524 named parentheses can match. Suppose you want to match the name of a
4525 weekday, either as a 3-letter abbreviation or as the full name, and in
4526 both cases you want to extract the abbreviation. This pattern (ignoring
4527 the line breaks) does the job:
4528
4529 (?<DN>Mon|Fri|Sun)(?:day)?|
4530 (?<DN>Tue)(?:sday)?|
4531 (?<DN>Wed)(?:nesday)?|
4532 (?<DN>Thu)(?:rsday)?|
4533 (?<DN>Sat)(?:urday)?
4534
4535 There are five capturing substrings, but only one is ever set after a
4536 match. (An alternative way of solving this problem is to use a "branch
4537 reset" subpattern, as described in the previous section.)
4538
4539 The convenience function for extracting the data by name returns the
4540 substring for the first (and in this example, the only) subpattern of
4541 that name that matched. This saves searching to find which numbered
4542 subpattern it was.
4543
4544 If you make a back reference to a non-unique named subpattern from
4545 elsewhere in the pattern, the one that corresponds to the first occur-
4546 rence of the name is used. In the absence of duplicate numbers (see the
4547 previous section) this is the one with the lowest number. If you use a
4548 named reference in a condition test (see the section about conditions
4549 below), either to check whether a subpattern has matched, or to check
4550 for recursion, all subpatterns with the same name are tested. If the
4551 condition is true for any one of them, the overall condition is true.
4552 This is the same behaviour as testing by number. For further details of
4553 the interfaces for handling named subpatterns, see the pcreapi documen-
4554 tation.
4555
4556 Warning: You cannot use different names to distinguish between two sub-
4557 patterns with the same number because PCRE uses only the numbers when
4558 matching. For this reason, an error is given at compile time if differ-
4559 ent names are given to subpatterns with the same number. However, you
4560 can give the same name to subpatterns with the same number, even when
4561 PCRE_DUPNAMES is not set.
4562
4563
4564 REPETITION
4565
4566 Repetition is specified by quantifiers, which can follow any of the
4567 following items:
4568
4569 a literal data character
4570 the dot metacharacter
4571 the \C escape sequence
4572 the \X escape sequence (in UTF-8 mode with Unicode properties)
4573 the \R escape sequence
4574 an escape such as \d or \pL that matches a single character
4575 a character class
4576 a back reference (see next section)
4577 a parenthesized subpattern (including assertions)
4578 a recursive or "subroutine" call to a subpattern
4579
4580 The general repetition quantifier specifies a minimum and maximum num-
4581 ber of permitted matches, by giving the two numbers in curly brackets
4582 (braces), separated by a comma. The numbers must be less than 65536,
4583 and the first must be less than or equal to the second. For example:
4584
4585 z{2,4}
4586
4587 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
4588 special character. If the second number is omitted, but the comma is
4589 present, there is no upper limit; if the second number and the comma
4590 are both omitted, the quantifier specifies an exact number of required
4591 matches. Thus
4592
4593 [aeiou]{3,}
4594
4595 matches at least 3 successive vowels, but may match many more, while
4596
4597 \d{8}
4598
4599 matches exactly 8 digits. An opening curly bracket that appears in a
4600 position where a quantifier is not allowed, or one that does not match
4601 the syntax of a quantifier, is taken as a literal character. For exam-
4602 ple, {,6} is not a quantifier, but a literal string of four characters.
4603
4604 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to
4605 individual bytes. Thus, for example, \x{100}{2} matches two UTF-8 char-
4606 acters, each of which is represented by a two-byte sequence. Similarly,
4607 when Unicode property support is available, \X{3} matches three Unicode
4608 extended sequences, each of which may be several bytes long (and they
4609 may be of different lengths).
4610
4611 The quantifier {0} is permitted, causing the expression to behave as if
4612 the previous item and the quantifier were not present. This may be use-
4613 ful for subpatterns that are referenced as subroutines from elsewhere
4614 in the pattern (but see also the section entitled "Defining subpatterns
4615 for use by reference only" below). Items other than subpatterns that
4616 have a {0} quantifier are omitted from the compiled pattern.
4617
4618 For convenience, the three most common quantifiers have single-charac-
4619 ter abbreviations:
4620
4621 * is equivalent to {0,}
4622 + is equivalent to {1,}
4623 ? is equivalent to {0,1}
4624
4625 It is possible to construct infinite loops by following a subpattern
4626 that can match no characters with a quantifier that has no upper limit,
4627 for example:
4628
4629 (a?)*
4630
4631 Earlier versions of Perl and PCRE used to give an error at compile time
4632 for such patterns. However, because there are cases where this can be
4633 useful, such patterns are now accepted, but if any repetition of the
4634 subpattern does in fact match no characters, the loop is forcibly bro-
4635 ken.
4636
4637 By default, the quantifiers are "greedy", that is, they match as much
4638 as possible (up to the maximum number of permitted times), without
4639 causing the rest of the pattern to fail. The classic example of where
4640 this gives problems is in trying to match comments in C programs. These
4641 appear between /* and */ and within the comment, individual * and /
4642 characters may appear. An attempt to match C comments by applying the
4643 pattern
4644
4645 /\*.*\*/
4646
4647 to the string
4648
4649 /* first comment */ not comment /* second comment */
4650
4651 fails, because it matches the entire string owing to the greediness of
4652 the .* item.
4653
4654 However, if a quantifier is followed by a question mark, it ceases to
4655 be greedy, and instead matches the minimum number of times possible, so
4656 the pattern
4657
4658 /\*.*?\*/
4659
4660 does the right thing with the C comments. The meaning of the various
4661 quantifiers is not otherwise changed, just the preferred number of
4662 matches. Do not confuse this use of question mark with its use as a
4663 quantifier in its own right. Because it has two uses, it can sometimes
4664 appear doubled, as in
4665
4666 \d??\d
4667
4668 which matches one digit by preference, but can match two if that is the
4669 only way the rest of the pattern matches.
4670
4671 If the PCRE_UNGREEDY option is set (an option that is not available in
4672 Perl), the quantifiers are not greedy by default, but individual ones
4673 can be made greedy by following them with a question mark. In other
4674 words, it inverts the default behaviour.
4675
4676 When a parenthesized subpattern is quantified with a minimum repeat
4677 count that is greater than 1 or with a limited maximum, more memory is
4678 required for the compiled pattern, in proportion to the size of the
4679 minimum or maximum.
4680
4681 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
4682 alent to Perl's /s) is set, thus allowing the dot to match newlines,
4683 the pattern is implicitly anchored, because whatever follows will be
4684 tried against every character position in the subject string, so there
4685 is no point in retrying the overall match at any position after the
4686 first. PCRE normally treats such a pattern as though it were preceded
4687 by \A.
4688
4689 In cases where it is known that the subject string contains no new-
4690 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
4691 mization, or alternatively using ^ to indicate anchoring explicitly.
4692
4693 However, there is one situation where the optimization cannot be used.
4694 When .* is inside capturing parentheses that are the subject of a back
4695 reference elsewhere in the pattern, a match at the start may fail where
4696 a later one succeeds. Consider, for example:
4697
4698 (.*)abc\1
4699
4700 If the subject is "xyz123abc123" the match point is the fourth charac-
4701 ter. For this reason, such a pattern is not implicitly anchored.
4702
4703 When a capturing subpattern is repeated, the value captured is the sub-
4704 string that matched the final iteration. For example, after
4705
4706 (tweedle[dume]{3}\s*)+
4707
4708 has matched "tweedledum tweedledee" the value of the captured substring
4709 is "tweedledee". However, if there are nested capturing subpatterns,
4710 the corresponding captured values may have been set in previous itera-
4711 tions. For example, after
4712
4713 /(a|(b))+/
4714
4715 matches "aba" the value of the second captured substring is "b".
4716
4717
4718 ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS
4719
4720 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
4721 repetition, failure of what follows normally causes the repeated item
4722 to be re-evaluated to see if a different number of repeats allows the
4723 rest of the pattern to match. Sometimes it is useful to prevent this,
4724 either to change the nature of the match, or to cause it fail earlier
4725 than it otherwise might, when the author of the pattern knows there is
4726 no point in carrying on.
4727
4728 Consider, for example, the pattern \d+foo when applied to the subject
4729 line
4730
4731 123456bar
4732
4733 After matching all 6 digits and then failing to match "foo", the normal
4734 action of the matcher is to try again with only 5 digits matching the
4735 \d+ item, and then with 4, and so on, before ultimately failing.
4736 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
4737 the means for specifying that once a subpattern has matched, it is not
4738 to be re-evaluated in this way.
4739
4740 If we use atomic grouping for the previous example, the matcher gives
4741 up immediately on failing to match "foo" the first time. The notation
4742 is a kind of special parenthesis, starting with (?> as in this example:
4743
4744 (?>\d+)foo
4745
4746 This kind of parenthesis "locks up" the part of the pattern it con-
4747 tains once it has matched, and a failure further into the pattern is
4748 prevented from backtracking into it. Backtracking past it to previous
4749 items, however, works as normal.
4750
4751 An alternative description is that a subpattern of this type matches
4752 the string of characters that an identical standalone pattern would
4753 match, if anchored at the current point in the subject string.
4754
4755 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
4756 such as the above example can be thought of as a maximizing repeat that
4757 must swallow everything it can. So, while both \d+ and \d+? are pre-
4758 pared to adjust the number of digits they match in order to make the
4759 rest of the pattern match, (?>\d+) can only match an entire sequence of
4760 digits.
4761
4762 Atomic groups in general can of course contain arbitrarily complicated
4763 subpatterns, and can be nested. However, when the subpattern for an
4764 atomic group is just a single repeated item, as in the example above, a
4765 simpler notation, called a "possessive quantifier" can be used. This
4766 consists of an additional + character following a quantifier. Using
4767 this notation, the previous example can be rewritten as
4768
4769 \d++foo
4770
4771 Note that a possessive quantifier can be used with an entire group, for
4772 example:
4773
4774 (abc|xyz){2,3}+
4775
4776 Possessive quantifiers are always greedy; the setting of the
4777 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
4778 simpler forms of atomic group. However, there is no difference in the
4779 meaning of a possessive quantifier and the equivalent atomic group,
4780 though there may be a performance difference; possessive quantifiers
4781 should be slightly faster.
4782
4783 The possessive quantifier syntax is an extension to the Perl 5.8 syn-
4784 tax. Jeffrey Friedl originated the idea (and the name) in the first
4785 edition of his book. Mike McCloskey liked it, so implemented it when he
4786 built Sun's Java package, and PCRE copied it from there. It ultimately
4787 found its way into Perl at release 5.10.
4788
4789 PCRE has an optimization that automatically "possessifies" certain sim-
4790 ple pattern constructs. For example, the sequence A+B is treated as
4791 A++B because there is no point in backtracking into a sequence of A's
4792 when B must follow.
4793
4794 When a pattern contains an unlimited repeat inside a subpattern that
4795 can itself be repeated an unlimited number of times, the use of an
4796 atomic group is the only way to avoid some failing matches taking a
4797 very long time indeed. The pattern
4798
4799 (\D+|<\d+>)*[!?]
4800
4801 matches an unlimited number of substrings that either consist of non-
4802 digits, or digits enclosed in <>, followed by either ! or ?. When it
4803 matches, it runs quickly. However, if it is applied to
4804
4805 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
4806
4807 it takes a long time before reporting failure. This is because the
4808 string can be divided between the internal \D+ repeat and the external
4809 * repeat in a large number of ways, and all have to be tried. (The
4810 example uses [!?] rather than a single character at the end, because
4811 both PCRE and Perl have an optimization that allows for fast failure
4812 when a single character is used. They remember the last single charac-
4813 ter that is required for a match, and fail early if it is not present
4814 in the string.) If the pattern is changed so that it uses an atomic
4815 group, like this:
4816
4817 ((?>\D+)|<\d+>)*[!?]
4818
4819 sequences of non-digits cannot be broken, and failure happens quickly.
4820
4821
4822 BACK REFERENCES
4823
4824 Outside a character class, a backslash followed by a digit greater than
4825 0 (and possibly further digits) is a back reference to a capturing sub-
4826 pattern earlier (that is, to its left) in the pattern, provided there
4827 have been that many previous capturing left parentheses.
4828
4829 However, if the decimal number following the backslash is less than 10,
4830 it is always taken as a back reference, and causes an error only if
4831 there are not that many capturing left parentheses in the entire pat-
4832 tern. In other words, the parentheses that are referenced need not be
4833 to the left of the reference for numbers less than 10. A "forward back
4834 reference" of this type can make sense when a repetition is involved
4835 and the subpattern to the right has participated in an earlier itera-
4836 tion.
4837
4838 It is not possible to have a numerical "forward back reference" to a
4839 subpattern whose number is 10 or more using this syntax because a
4840 sequence such as \50 is interpreted as a character defined in octal.
4841 See the subsection entitled "Non-printing characters" above for further
4842 details of the handling of digits following a backslash. There is no
4843 such problem when named parentheses are used. A back reference to any
4844 subpattern is possible using named parentheses (see below).
4845
4846 Another way of avoiding the ambiguity inherent in the use of digits
4847 following a backslash is to use the \g escape sequence. This escape
4848 must be followed by an unsigned number or a negative number, optionally
4849 enclosed in braces. These examples are all identical:
4850
4851 (ring), \1
4852 (ring), \g1
4853 (ring), \g{1}
4854
4855 An unsigned number specifies an absolute reference without the ambigu-
4856 ity that is present in the older syntax. It is also useful when literal
4857 digits follow the reference. A negative number is a relative reference.
4858 Consider this example:
4859
4860 (abc(def)ghi)\g{-1}
4861
4862 The sequence \g{-1} is a reference to the most recently started captur-
4863 ing subpattern before \g, that is, is it equivalent to \2 in this exam-
4864 ple. Similarly, \g{-2} would be equivalent to \1. The use of relative
4865 references can be helpful in long patterns, and also in patterns that
4866 are created by joining together fragments that contain references
4867 within themselves.
4868
4869 A back reference matches whatever actually matched the capturing sub-
4870 pattern in the current subject string, rather than anything matching
4871 the subpattern itself (see "Subpatterns as subroutines" below for a way
4872 of doing that). So the pattern
4873
4874 (sens|respons)e and \1ibility
4875
4876 matches "sense and sensibility" and "response and responsibility", but
4877 not "sense and responsibility". If caseful matching is in force at the
4878 time of the back reference, the case of letters is relevant. For exam-
4879 ple,
4880
4881 ((?i)rah)\s+\1
4882
4883 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
4884 original capturing subpattern is matched caselessly.
4885
4886 There are several different ways of writing back references to named
4887 subpatterns. The .NET syntax \k{name} and the Perl syntax \k<name> or
4888 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's
4889 unified back reference syntax, in which \g can be used for both numeric
4890 and named references, is also supported. We could rewrite the above
4891 example in any of the following ways:
4892
4893 (?<p1>(?i)rah)\s+\k<p1>
4894 (?'p1'(?i)rah)\s+\k{p1}
4895 (?P<p1>(?i)rah)\s+(?P=p1)
4896 (?<p1>(?i)rah)\s+\g{p1}
4897
4898 A subpattern that is referenced by name may appear in the pattern
4899 before or after the reference.
4900
4901 There may be more than one back reference to the same subpattern. If a
4902 subpattern has not actually been used in a particular match, any back
4903 references to it always fail by default. For example, the pattern
4904
4905 (a|(bc))\2
4906
4907 always fails if it starts to match "a" rather than "bc". However, if
4908 the PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back refer-
4909 ence to an unset value matches an empty string.
4910
4911 Because there may be many capturing parentheses in a pattern, all dig-
4912 its following a backslash are taken as part of a potential back refer-
4913 ence number. If the pattern continues with a digit character, some
4914 delimiter must be used to terminate the back reference. If the
4915 PCRE_EXTENDED option is set, this can be whitespace. Otherwise, the \g{
4916 syntax or an empty comment (see "Comments" below) can be used.
4917
4918 Recursive back references
4919
4920 A back reference that occurs inside the parentheses to which it refers
4921 fails when the subpattern is first used, so, for example, (a\1) never
4922 matches. However, such references can be useful inside repeated sub-
4923 patterns. For example, the pattern
4924
4925 (a|b\1)+
4926
4927 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
4928 ation of the subpattern, the back reference matches the character
4929 string corresponding to the previous iteration. In order for this to
4930 work, the pattern must be such that the first iteration does not need
4931 to match the back reference. This can be done using alternation, as in
4932 the example above, or by a quantifier with a minimum of zero.
4933
4934 Back references of this type cause the group that they reference to be
4935 treated as an atomic group. Once the whole group has been matched, a
4936 subsequent matching failure cannot cause backtracking into the middle
4937 of the group.
4938
4939
4940 ASSERTIONS
4941
4942 An assertion is a test on the characters following or preceding the
4943 current matching point that does not actually consume any characters.
4944 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
4945 described above.
4946
4947 More complicated assertions are coded as subpatterns. There are two
4948 kinds: those that look ahead of the current position in the subject
4949 string, and those that look behind it. An assertion subpattern is
4950 matched in the normal way, except that it does not cause the current
4951 matching position to be changed.
4952
4953 Assertion subpatterns are not capturing subpatterns. If such an asser-
4954 tion contains capturing subpatterns within it, these are counted for
4955 the purposes of numbering the capturing subpatterns in the whole pat-
4956 tern. However, substring capturing is carried out only for positive
4957 assertions, because it does not make sense for negative assertions.
4958
4959 For compatibility with Perl, assertion subpatterns may be repeated;
4960 though it makes no sense to assert the same thing several times, the
4961 side effect of capturing parentheses may occasionally be useful. In
4962 practice, there only three cases:
4963
4964 (1) If the quantifier is {0}, the assertion is never obeyed during
4965 matching. However, it may contain internal capturing parenthesized
4966 groups that are called from elsewhere via the subroutine mechanism.
4967
4968 (2) If quantifier is {0,n} where n is greater than zero, it is treated
4969 as if it were {0,1}. At run time, the rest of the pattern match is
4970 tried with and without the assertion, the order depending on the greed-
4971 iness of the quantifier.
4972
4973 (3) If the minimum repetition is greater than zero, the quantifier is
4974 ignored. The assertion is obeyed just once when encountered during
4975 matching.
4976
4977 Lookahead assertions
4978
4979 Lookahead assertions start with (?= for positive assertions and (?! for
4980 negative assertions. For example,
4981
4982 \w+(?=;)
4983
4984 matches a word followed by a semicolon, but does not include the semi-
4985 colon in the match, and
4986
4987 foo(?!bar)
4988
4989 matches any occurrence of "foo" that is not followed by "bar". Note
4990 that the apparently similar pattern
4991
4992 (?!foo)bar
4993
4994 does not find an occurrence of "bar" that is preceded by something
4995 other than "foo"; it finds any occurrence of "bar" whatsoever, because
4996 the assertion (?!foo) is always true when the next three characters are
4997 "bar". A lookbehind assertion is needed to achieve the other effect.
4998
4999 If you want to force a matching failure at some point in a pattern, the
5000 most convenient way to do it is with (?!) because an empty string
5001 always matches, so an assertion that requires there not to be an empty
5002 string must always fail. The backtracking control verb (*FAIL) or (*F)
5003 is a synonym for (?!).
5004
5005 Lookbehind assertions
5006
5007 Lookbehind assertions start with (?<= for positive assertions and (?<!
5008 for negative assertions. For example,
5009
5010 (?<!foo)bar
5011
5012 does find an occurrence of "bar" that is not preceded by "foo". The
5013 contents of a lookbehind assertion are restricted such that all the
5014 strings it matches must have a fixed length. However, if there are sev-
5015 eral top-level alternatives, they do not all have to have the same
5016 fixed length. Thus
5017
5018 (?<=bullock|donkey)
5019
5020 is permitted, but
5021
5022 (?<!dogs?|cats?)
5023
5024 causes an error at compile time. Branches that match different length
5025 strings are permitted only at the top level of a lookbehind assertion.
5026 This is an extension compared with Perl, which requires all branches to
5027 match the same length of string. An assertion such as
5028
5029 (?<=ab(c|de))
5030
5031 is not permitted, because its single top-level branch can match two
5032 different lengths, but it is acceptable to PCRE if rewritten to use two
5033 top-level branches:
5034
5035 (?<=abc|abde)
5036
5037 In some cases, the escape sequence \K (see above) can be used instead
5038 of a lookbehind assertion to get round the fixed-length restriction.
5039
5040 The implementation of lookbehind assertions is, for each alternative,
5041 to temporarily move the current position back by the fixed length and
5042 then try to match. If there are insufficient characters before the cur-
5043 rent position, the assertion fails.
5044
5045 PCRE does not allow the \C escape (which matches a single byte in UTF-8
5046 mode) to appear in lookbehind assertions, because it makes it impossi-
5047 ble to calculate the length of the lookbehind. The \X and \R escapes,
5048 which can match different numbers of bytes, are also not permitted.
5049
5050 "Subroutine" calls (see below) such as (?2) or (?&X) are permitted in
5051 lookbehinds, as long as the subpattern matches a fixed-length string.
5052 Recursion, however, is not supported.
5053
5054 Possessive quantifiers can be used in conjunction with lookbehind
5055 assertions to specify efficient matching of fixed-length strings at the
5056 end of subject strings. Consider a simple pattern such as
5057
5058 abcd$
5059
5060 when applied to a long string that does not match. Because matching
5061 proceeds from left to right, PCRE will look for each "a" in the subject
5062 and then see if what follows matches the rest of the pattern. If the
5063 pattern is specified as
5064
5065 ^.*abcd$
5066
5067 the initial .* matches the entire string at first, but when this fails
5068 (because there is no following "a"), it backtracks to match all but the
5069 last character, then all but the last two characters, and so on. Once
5070 again the search for "a" covers the entire string, from right to left,
5071 so we are no better off. However, if the pattern is written as
5072
5073 ^.*+(?<=abcd)
5074
5075 there can be no backtracking for the .*+ item; it can match only the
5076 entire string. The subsequent lookbehind assertion does a single test
5077 on the last four characters. If it fails, the match fails immediately.
5078 For long strings, this approach makes a significant difference to the
5079 processing time.
5080
5081 Using multiple assertions
5082
5083 Several assertions (of any sort) may occur in succession. For example,
5084
5085 (?<=\d{3})(?<!999)foo
5086
5087 matches "foo" preceded by three digits that are not "999". Notice that
5088 each of the assertions is applied independently at the same point in
5089 the subject string. First there is a check that the previous three
5090 characters are all digits, and then there is a check that the same
5091 three characters are not "999". This pattern does not match "foo" pre-
5092 ceded by six characters, the first of which are digits and the last
5093 three of which are not "999". For example, it doesn't match "123abc-
5094 foo". A pattern to do that is
5095
5096 (?<=\d{3}...)(?<!999)foo
5097
5098 This time the first assertion looks at the preceding six characters,
5099 checking that the first three are digits, and then the second assertion
5100 checks that the preceding three characters are not "999".
5101
5102 Assertions can be nested in any combination. For example,
5103
5104 (?<=(?<!foo)bar)baz
5105
5106 matches an occurrence of "baz" that is preceded by "bar" which in turn
5107 is not preceded by "foo", while
5108
5109 (?<=\d{3}(?!999)...)foo
5110
5111 is another pattern that matches "foo" preceded by three digits and any
5112 three characters that are not "999".
5113
5114
5115 CONDITIONAL SUBPATTERNS
5116
5117 It is possible to cause the matching process to obey a subpattern con-
5118 ditionally or to choose between two alternative subpatterns, depending
5119 on the result of an assertion, or whether a specific capturing subpat-
5120 tern has already been matched. The two possible forms of conditional
5121 subpattern are:
5122
5123 (?(condition)yes-pattern)
5124 (?(condition)yes-pattern|no-pattern)
5125
5126 If the condition is satisfied, the yes-pattern is used; otherwise the
5127 no-pattern (if present) is used. If there are more than two alterna-
5128 tives in the subpattern, a compile-time error occurs. Each of the two
5129 alternatives may itself contain nested subpatterns of any form, includ-
5130 ing conditional subpatterns; the restriction to two alternatives
5131 applies only at the level of the condition. This pattern fragment is an
5132 example where the alternatives are complex:
5133
5134 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
5135
5136
5137 There are four kinds of condition: references to subpatterns, refer-
5138 ences to recursion, a pseudo-condition called DEFINE, and assertions.
5139
5140 Checking for a used subpattern by number
5141
5142 If the text between the parentheses consists of a sequence of digits,
5143 the condition is true if a capturing subpattern of that number has pre-
5144 viously matched. If there is more than one capturing subpattern with
5145 the same number (see the earlier section about duplicate subpattern
5146 numbers), the condition is true if any of them have matched. An alter-
5147 native notation is to precede the digits with a plus or minus sign. In
5148 this case, the subpattern number is relative rather than absolute. The
5149 most recently opened parentheses can be referenced by (?(-1), the next
5150 most recent by (?(-2), and so on. Inside loops it can also make sense
5151 to refer to subsequent groups. The next parentheses to be opened can be
5152 referenced as (?(+1), and so on. (The value zero in any of these forms
5153 is not used; it provokes a compile-time error.)
5154
5155 Consider the following pattern, which contains non-significant white
5156 space to make it more readable (assume the PCRE_EXTENDED option) and to
5157 divide it into three parts for ease of discussion:
5158
5159 ( \( )? [^()]+ (?(1) \) )
5160
5161 The first part matches an optional opening parenthesis, and if that
5162 character is present, sets it as the first captured substring. The sec-
5163 ond part matches one or more characters that are not parentheses. The
5164 third part is a conditional subpattern that tests whether or not the
5165 first set of parentheses matched. If they did, that is, if subject
5166 started with an opening parenthesis, the condition is true, and so the
5167 yes-pattern is executed and a closing parenthesis is required. Other-
5168 wise, since no-pattern is not present, the subpattern matches nothing.
5169 In other words, this pattern matches a sequence of non-parentheses,
5170 optionally enclosed in parentheses.
5171
5172 If you were embedding this pattern in a larger one, you could use a
5173 relative reference:
5174
5175 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
5176
5177 This makes the fragment independent of the parentheses in the larger
5178 pattern.
5179
5180 Checking for a used subpattern by name
5181
5182 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
5183 used subpattern by name. For compatibility with earlier versions of
5184 PCRE, which had this facility before Perl, the syntax (?(name)...) is
5185 also recognized. However, there is a possible ambiguity with this syn-
5186 tax, because subpattern names may consist entirely of digits. PCRE
5187 looks first for a named subpattern; if it cannot find one and the name
5188 consists entirely of digits, PCRE looks for a subpattern of that num-
5189 ber, which must be greater than zero. Using subpattern names that con-
5190 sist entirely of digits is not recommended.
5191
5192 Rewriting the above example to use a named subpattern gives this:
5193
5194 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
5195
5196 If the name used in a condition of this kind is a duplicate, the test
5197 is applied to all subpatterns of the same name, and is true if any one
5198 of them has matched.
5199
5200 Checking for pattern recursion
5201
5202 If the condition is the string (R), and there is no subpattern with the
5203 name R, the condition is true if a recursive call to the whole pattern
5204 or any subpattern has been made. If digits or a name preceded by amper-
5205 sand follow the letter R, for example:
5206
5207 (?(R3)...) or (?(R&name)...)
5208
5209 the condition is true if the most recent recursion is into a subpattern
5210 whose number or name is given. This condition does not check the entire
5211 recursion stack. If the name used in a condition of this kind is a
5212 duplicate, the test is applied to all subpatterns of the same name, and
5213 is true if any one of them is the most recent recursion.
5214
5215 At "top level", all these recursion test conditions are false. The
5216 syntax for recursive patterns is described below.
5217
5218 Defining subpatterns for use by reference only
5219
5220 If the condition is the string (DEFINE), and there is no subpattern
5221 with the name DEFINE, the condition is always false. In this case,
5222 there may be only one alternative in the subpattern. It is always
5223 skipped if control reaches this point in the pattern; the idea of
5224 DEFINE is that it can be used to define "subroutines" that can be ref-
5225 erenced from elsewhere. (The use of "subroutines" is described below.)
5226 For example, a pattern to match an IPv4 address such as
5227 "192.168.23.245" could be written like this (ignore whitespace and line
5228 breaks):
5229
5230 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
5231 \b (?&byte) (\.(?&byte)){3} \b
5232
5233 The first part of the pattern is a DEFINE group inside which a another
5234 group named "byte" is defined. This matches an individual component of
5235 an IPv4 address (a number less than 256). When matching takes place,
5236 this part of the pattern is skipped because DEFINE acts like a false
5237 condition. The rest of the pattern uses references to the named group
5238 to match the four dot-separated components of an IPv4 address, insist-
5239 ing on a word boundary at each end.
5240
5241 Assertion conditions
5242
5243 If the condition is not in any of the above formats, it must be an
5244 assertion. This may be a positive or negative lookahead or lookbehind
5245 assertion. Consider this pattern, again containing non-significant
5246 white space, and with the two alternatives on the second line:
5247
5248 (?(?=[^a-z]*[a-z])
5249 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
5250
5251 The condition is a positive lookahead assertion that matches an
5252 optional sequence of non-letters followed by a letter. In other words,
5253 it tests for the presence of at least one letter in the subject. If a
5254 letter is found, the subject is matched against the first alternative;
5255 otherwise it is matched against the second. This pattern matches
5256 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
5257 letters and dd are digits.
5258
5259
5260 COMMENTS
5261
5262 There are two ways of including comments in patterns that are processed
5263 by PCRE. In both cases, the start of the comment must not be in a char-
5264 acter class, nor in the middle of any other sequence of related charac-
5265 ters such as (?: or a subpattern name or number. The characters that
5266 make up a comment play no part in the pattern matching.
5267
5268 The sequence (?# marks the start of a comment that continues up to the
5269 next closing parenthesis. Nested parentheses are not permitted. If the
5270 PCRE_EXTENDED option is set, an unescaped # character also introduces a
5271 comment, which in this case continues to immediately after the next
5272 newline character or character sequence in the pattern. Which charac-
5273 ters are interpreted as newlines is controlled by the options passed to
5274 pcre_compile() or by a special sequence at the start of the pattern, as
5275 described in the section entitled "Newline conventions" above. Note
5276 that the end of this type of comment is a literal newline sequence in
5277 the pattern; escape sequences that happen to represent a newline do not
5278 count. For example, consider this pattern when PCRE_EXTENDED is set,
5279 and the default newline convention is in force:
5280
5281 abc #comment \n still comment
5282
5283 On encountering the # character, pcre_compile() skips along, looking
5284 for a newline in the pattern. The sequence \n is still literal at this
5285 stage, so it does not terminate the comment. Only an actual character
5286 with the code value 0x0a (the default newline) does so.
5287
5288
5289 RECURSIVE PATTERNS
5290
5291 Consider the problem of matching a string in parentheses, allowing for
5292 unlimited nested parentheses. Without the use of recursion, the best
5293 that can be done is to use a pattern that matches up to some fixed
5294 depth of nesting. It is not possible to handle an arbitrary nesting
5295 depth.
5296
5297 For some time, Perl has provided a facility that allows regular expres-
5298 sions to recurse (amongst other things). It does this by interpolating
5299 Perl code in the expression at run time, and the code can refer to the
5300 expression itself. A Perl pattern using code interpolation to solve the
5301 parentheses problem can be created like this:
5302
5303 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
5304
5305 The (?p{...}) item interpolates Perl code at run time, and in this case
5306 refers recursively to the pattern in which it appears.
5307
5308 Obviously, PCRE cannot support the interpolation of Perl code. Instead,
5309 it supports special syntax for recursion of the entire pattern, and
5310 also for individual subpattern recursion. After its introduction in
5311 PCRE and Python, this kind of recursion was subsequently introduced
5312 into Perl at release 5.10.
5313
5314 A special item that consists of (? followed by a number greater than
5315 zero and a closing parenthesis is a recursive call of the subpattern of
5316 the given number, provided that it occurs inside that subpattern. (If
5317 not, it is a "subroutine" call, which is described in the next sec-
5318 tion.) The special item (?R) or (?0) is a recursive call of the entire
5319 regular expression.
5320
5321 This PCRE pattern solves the nested parentheses problem (assume the
5322 PCRE_EXTENDED option is set so that white space is ignored):
5323
5324 \( ( [^()]++ | (?R) )* \)
5325
5326 First it matches an opening parenthesis. Then it matches any number of
5327 substrings which can either be a sequence of non-parentheses, or a
5328 recursive match of the pattern itself (that is, a correctly parenthe-
5329 sized substring). Finally there is a closing parenthesis. Note the use
5330 of a possessive quantifier to avoid backtracking into sequences of non-
5331 parentheses.
5332
5333 If this were part of a larger pattern, you would not want to recurse
5334 the entire pattern, so instead you could use this:
5335
5336 ( \( ( [^()]++ | (?1) )* \) )
5337
5338 We have put the pattern into parentheses, and caused the recursion to
5339 refer to them instead of the whole pattern.
5340
5341 In a larger pattern, keeping track of parenthesis numbers can be
5342 tricky. This is made easier by the use of relative references. Instead
5343 of (?1) in the pattern above you can write (?-2) to refer to the second
5344 most recently opened parentheses preceding the recursion. In other
5345 words, a negative number counts capturing parentheses leftwards from
5346 the point at which it is encountered.
5347
5348 It is also possible to refer to subsequently opened parentheses, by
5349 writing references such as (?+2). However, these cannot be recursive
5350 because the reference is not inside the parentheses that are refer-
5351 enced. They are always "subroutine" calls, as described in the next
5352 section.
5353
5354 An alternative approach is to use named parentheses instead. The Perl
5355 syntax for this is (?&name); PCRE's earlier syntax (?P>name) is also
5356 supported. We could rewrite the above example as follows:
5357
5358 (?<pn> \( ( [^()]++ | (?&pn) )* \) )
5359
5360 If there is more than one subpattern with the same name, the earliest
5361 one is used.
5362
5363 This particular example pattern that we have been looking at contains
5364 nested unlimited repeats, and so the use of a possessive quantifier for
5365 matching strings of non-parentheses is important when applying the pat-
5366 tern to strings that do not match. For example, when this pattern is
5367 applied to
5368
5369 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
5370
5371 it yields "no match" quickly. However, if a possessive quantifier is
5372 not used, the match runs for a very long time indeed because there are
5373 so many different ways the + and * repeats can carve up the subject,
5374 and all have to be tested before failure can be reported.
5375
5376 At the end of a match, the values of capturing parentheses are those
5377 from the outermost level. If you want to obtain intermediate values, a
5378 callout function can be used (see below and the pcrecallout documenta-
5379 tion). If the pattern above is matched against
5380
5381 (ab(cd)ef)
5382
5383 the value for the inner capturing parentheses (numbered 2) is "ef",
5384 which is the last value taken on at the top level. If a capturing sub-
5385 pattern is not matched at the top level, its final value is unset, even
5386 if it is (temporarily) set at a deeper level.
5387
5388 If there are more than 15 capturing parentheses in a pattern, PCRE has
5389 to obtain extra memory to store data during a recursion, which it does
5390 by using pcre_malloc, freeing it via pcre_free afterwards. If no memory
5391 can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
5392
5393 Do not confuse the (?R) item with the condition (R), which tests for
5394 recursion. Consider this pattern, which matches text in angle brack-
5395 ets, allowing for arbitrary nesting. Only digits are allowed in nested
5396 brackets (that is, when recursing), whereas any characters are permit-
5397 ted at the outer level.
5398
5399 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
5400
5401 In this pattern, (?(R) is the start of a conditional subpattern, with
5402 two different alternatives for the recursive and non-recursive cases.
5403 The (?R) item is the actual recursive call.
5404
5405 Recursion difference from Perl
5406
5407 In PCRE (like Python, but unlike Perl), a recursive subpattern call is
5408 always treated as an atomic group. That is, once it has matched some of
5409 the subject string, it is never re-entered, even if it contains untried
5410 alternatives and there is a subsequent matching failure. This can be
5411 illustrated by the following pattern, which purports to match a palin-
5412 dromic string that contains an odd number of characters (for example,
5413 "a", "aba", "abcba", "abcdcba"):
5414
5415 ^(.|(.)(?1)\2)$
5416
5417 The idea is that it either matches a single character, or two identical
5418 characters surrounding a sub-palindrome. In Perl, this pattern works;
5419 in PCRE it does not if the pattern is longer than three characters.
5420 Consider the subject string "abcba":
5421
5422 At the top level, the first character is matched, but as it is not at
5423 the end of the string, the first alternative fails; the second alterna-
5424 tive is taken and the recursion kicks in. The recursive call to subpat-
5425 tern 1 successfully matches the next character ("b"). (Note that the
5426 beginning and end of line tests are not part of the recursion).
5427
5428 Back at the top level, the next character ("c") is compared with what
5429 subpattern 2 matched, which was "a". This fails. Because the recursion
5430 is treated as an atomic group, there are now no backtracking points,
5431 and so the entire match fails. (Perl is able, at this point, to re-
5432 enter the recursion and try the second alternative.) However, if the
5433 pattern is written with the alternatives in the other order, things are
5434 different:
5435
5436 ^((.)(?1)\2|.)$
5437
5438 This time, the recursing alternative is tried first, and continues to
5439 recurse until it runs out of characters, at which point the recursion
5440 fails. But this time we do have another alternative to try at the
5441 higher level. That is the big difference: in the previous case the
5442 remaining alternative is at a deeper recursion level, which PCRE cannot
5443 use.
5444
5445 To change the pattern so that it matches all palindromic strings, not
5446 just those with an odd number of characters, it is tempting to change
5447 the pattern to this:
5448
5449 ^((.)(?1)\2|.?)$
5450
5451 Again, this works in Perl, but not in PCRE, and for the same reason.
5452 When a deeper recursion has matched a single character, it cannot be
5453 entered again in order to match an empty string. The solution is to
5454 separate the two cases, and write out the odd and even cases as alter-
5455 natives at the higher level:
5456
5457 ^(?:((.)(?1)\2|)|((.)(?3)\4|.))
5458
5459 If you want to match typical palindromic phrases, the pattern has to
5460 ignore all non-word characters, which can be done like this:
5461
5462 ^\W*+(?:((.)\W*+(?1)\W*+\2|)|((.)\W*+(?3)\W*+\4|\W*+.\W*+))\W*+$
5463
5464 If run with the PCRE_CASELESS option, this pattern matches phrases such
5465 as "A man, a plan, a canal: Panama!" and it works well in both PCRE and
5466 Perl. Note the use of the possessive quantifier *+ to avoid backtrack-
5467 ing into sequences of non-word characters. Without this, PCRE takes a
5468 great deal longer (ten times or more) to match typical phrases, and
5469 Perl takes so long that you think it has gone into a loop.
5470
5471 WARNING: The palindrome-matching patterns above work only if the sub-
5472 ject string does not start with a palindrome that is shorter than the
5473 entire string. For example, although "abcba" is correctly matched, if
5474 the subject is "ababa", PCRE finds the palindrome "aba" at the start,
5475 then fails at top level because the end of the string does not follow.
5476 Once again, it cannot jump back into the recursion to try other alter-
5477 natives, so the entire match fails.
5478
5479
5480 SUBPATTERNS AS SUBROUTINES
5481
5482 If the syntax for a recursive subpattern reference (either by number or
5483 by name) is used outside the parentheses to which it refers, it oper-
5484 ates like a subroutine in a programming language. The "called" subpat-
5485 tern may be defined before or after the reference. A numbered reference
5486 can be absolute or relative, as in these examples:
5487
5488 (...(absolute)...)...(?2)...
5489 (...(relative)...)...(?-1)...
5490 (...(?+1)...(relative)...
5491
5492 An earlier example pointed out that the pattern
5493
5494 (sens|respons)e and \1ibility
5495
5496 matches "sense and sensibility" and "response and responsibility", but
5497 not "sense and responsibility". If instead the pattern
5498
5499 (sens|respons)e and (?1)ibility
5500
5501 is used, it does match "sense and responsibility" as well as the other
5502 two strings. Another example is given in the discussion of DEFINE
5503 above.
5504
5505 Like recursive subpatterns, a subroutine call is always treated as an
5506 atomic group. That is, once it has matched some of the subject string,
5507 it is never re-entered, even if it contains untried alternatives and
5508 there is a subsequent matching failure. Any capturing parentheses that
5509 are set during the subroutine call revert to their previous values
5510 afterwards.
5511
5512 When a subpattern is used as a subroutine, processing options such as
5513 case-independence are fixed when the subpattern is defined. They cannot
5514 be changed for different calls. For example, consider this pattern:
5515
5516 (abc)(?i:(?-1))
5517
5518 It matches "abcabc". It does not match "abcABC" because the change of
5519 processing option does not affect the called subpattern.
5520
5521
5522 ONIGURUMA SUBROUTINE SYNTAX
5523
5524 For compatibility with Oniguruma, the non-Perl syntax \g followed by a
5525 name or a number enclosed either in angle brackets or single quotes, is
5526 an alternative syntax for referencing a subpattern as a subroutine,
5527 possibly recursively. Here are two of the examples used above, rewrit-
5528 ten using this syntax:
5529
5530 (?<pn> \( ( (?>[^()]+) | \g<pn> )* \) )
5531 (sens|respons)e and \g'1'ibility
5532
5533 PCRE supports an extension to Oniguruma: if a number is preceded by a
5534 plus or a minus sign it is taken as a relative reference. For example:
5535
5536 (abc)(?i:\g<-1>)
5537
5538 Note that \g{...} (Perl syntax) and \g<...> (Oniguruma syntax) are not
5539 synonymous. The former is a back reference; the latter is a subroutine
5540 call.
5541
5542
5543 CALLOUTS
5544
5545 Perl has a feature whereby using the sequence (?{...}) causes arbitrary
5546 Perl code to be obeyed in the middle of matching a regular expression.
5547 This makes it possible, amongst other things, to extract different sub-
5548 strings that match the same pair of parentheses when there is a repeti-
5549 tion.
5550
5551 PCRE provides a similar feature, but of course it cannot obey arbitrary
5552 Perl code. The feature is called "callout". The caller of PCRE provides
5553 an external function by putting its entry point in the global variable
5554 pcre_callout. By default, this variable contains NULL, which disables
5555 all calling out.
5556
5557 Within a regular expression, (?C) indicates the points at which the
5558 external function is to be called. If you want to identify different
5559 callout points, you can put a number less than 256 after the letter C.
5560 The default value is zero. For example, this pattern has two callout
5561 points:
5562
5563 (?C1)abc(?C2)def
5564
5565 If the PCRE_AUTO_CALLOUT flag is passed to pcre_compile(), callouts are
5566 automatically installed before each item in the pattern. They are all
5567 numbered 255.
5568
5569 During matching, when PCRE reaches a callout point (and pcre_callout is
5570 set), the external function is called. It is provided with the number
5571 of the callout, the position in the pattern, and, optionally, one item
5572 of data originally supplied by the caller of pcre_exec(). The callout
5573 function may cause matching to proceed, to backtrack, or to fail alto-
5574 gether. A complete description of the interface to the callout function
5575 is given in the pcrecallout documentation.
5576
5577
5578 BACKTRACKING CONTROL
5579
5580 Perl 5.10 introduced a number of "Special Backtracking Control Verbs",
5581 which are described in the Perl documentation as "experimental and sub-
5582 ject to change or removal in a future version of Perl". It goes on to
5583 say: "Their usage in production code should be noted to avoid problems
5584 during upgrades." The same remarks apply to the PCRE features described
5585 in this section.
5586
5587 Since these verbs are specifically related to backtracking, most of
5588 them can be used only when the pattern is to be matched using
5589 pcre_exec(), which uses a backtracking algorithm. With the exception of
5590 (*FAIL), which behaves like a failing negative assertion, they cause an
5591 error if encountered by pcre_dfa_exec().
5592
5593 If any of these verbs are used in an assertion or subroutine subpattern
5594 (including recursive subpatterns), their effect is confined to that
5595 subpattern; it does not extend to the surrounding pattern, with one
5596 exception: a *MARK that is encountered in a positive assertion is
5597 passed back (compare capturing parentheses in assertions). Note that
5598 such subpatterns are processed as anchored at the point where they are
5599 tested.
5600
5601 The new verbs make use of what was previously invalid syntax: an open-
5602 ing parenthesis followed by an asterisk. They are generally of the form
5603 (*VERB) or (*VERB:NAME). Some may take either form, with differing be-
5604 haviour, depending on whether or not an argument is present. An name is
5605 a sequence of letters, digits, and underscores. If the name is empty,
5606 that is, if the closing parenthesis immediately follows the colon, the
5607 effect is as if the colon were not there. Any number of these verbs may
5608 occur in a pattern.
5609
5610 PCRE contains some optimizations that are used to speed up matching by
5611 running some checks at the start of each match attempt. For example, it
5612 may know the minimum length of matching subject, or that a particular
5613 character must be present. When one of these optimizations suppresses
5614 the running of a match, any included backtracking verbs will not, of
5615 course, be processed. You can suppress the start-of-match optimizations
5616 by setting the PCRE_NO_START_OPTIMIZE option when calling pcre_com-
5617 pile() or pcre_exec(), or by starting the pattern with (*NO_START_OPT).
5618
5619 Verbs that act immediately
5620
5621 The following verbs act as soon as they are encountered. They may not
5622 be followed by a name.
5623
5624 (*ACCEPT)
5625
5626 This verb causes the match to end successfully, skipping the remainder
5627 of the pattern. When inside a recursion, only the innermost pattern is
5628 ended immediately. If (*ACCEPT) is inside capturing parentheses, the
5629 data so far is captured. (This feature was added to PCRE at release
5630 8.00.) For example:
5631
5632 A((?:A|B(*ACCEPT)|C)D)
5633
5634 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is cap-
5635 tured by the outer parentheses.
5636
5637 (*FAIL) or (*F)
5638
5639 This verb causes the match to fail, forcing backtracking to occur. It
5640 is equivalent to (?!) but easier to read. The Perl documentation notes
5641 that it is probably useful only when combined with (?{}) or (??{}).
5642 Those are, of course, Perl features that are not present in PCRE. The
5643 nearest equivalent is the callout feature, as for example in this pat-
5644 tern:
5645
5646 a+(?C)(*FAIL)
5647
5648 A match with the string "aaaa" always fails, but the callout is taken
5649 before each backtrack happens (in this example, 10 times).
5650
5651 Recording which path was taken
5652
5653 There is one verb whose main purpose is to track how a match was
5654 arrived at, though it also has a secondary use in conjunction with
5655 advancing the match starting point (see (*SKIP) below).
5656
5657 (*MARK:NAME) or (*:NAME)
5658
5659 A name is always required with this verb. There may be as many
5660 instances of (*MARK) as you like in a pattern, and their names do not
5661 have to be unique.
5662
5663 When a match succeeds, the name of the last-encountered (*MARK) is
5664 passed back to the caller via the pcre_extra data structure, as
5665 described in the section on pcre_extra in the pcreapi documentation. No
5666 data is returned for a partial match. Here is an example of pcretest
5667 output, where the /K modifier requests the retrieval and outputting of
5668 (*MARK) data:
5669
5670 /X(*MARK:A)Y|X(*MARK:B)Z/K
5671 XY
5672 0: XY
5673 MK: A
5674 XZ
5675 0: XZ
5676 MK: B
5677
5678 The (*MARK) name is tagged with "MK:" in this output, and in this exam-
5679 ple it indicates which of the two alternatives matched. This is a more
5680 efficient way of obtaining this information than putting each alterna-
5681 tive in its own capturing parentheses.
5682
5683 If (*MARK) is encountered in a positive assertion, its name is recorded
5684 and passed back if it is the last-encountered. This does not happen for
5685 negative assetions.
5686
5687 A name may also be returned after a failed match if the final path
5688 through the pattern involves (*MARK). However, unless (*MARK) used in
5689 conjunction with (*COMMIT), this is unlikely to happen for an unan-
5690 chored pattern because, as the starting point for matching is advanced,
5691 the final check is often with an empty string, causing a failure before
5692 (*MARK) is reached. For example:
5693
5694 /X(*MARK:A)Y|X(*MARK:B)Z/K
5695 XP
5696 No match
5697
5698 There are three potential starting points for this match (starting with
5699 X, starting with P, and with an empty string). If the pattern is
5700 anchored, the result is different:
5701
5702 /^X(*MARK:A)Y|^X(*MARK:B)Z/K
5703 XP
5704 No match, mark = B
5705
5706 PCRE's start-of-match optimizations can also interfere with this. For
5707 example, if, as a result of a call to pcre_study(), it knows the mini-
5708 mum subject length for a match, a shorter subject will not be scanned
5709 at all.
5710
5711 Note that similar anomalies (though different in detail) exist in Perl,
5712 no doubt for the same reasons. The use of (*MARK) data after a failed
5713 match of an unanchored pattern is not recommended, unless (*COMMIT) is
5714 involved.
5715
5716 Verbs that act after backtracking
5717
5718 The following verbs do nothing when they are encountered. Matching con-
5719 tinues with what follows, but if there is no subsequent match, causing
5720 a backtrack to the verb, a failure is forced. That is, backtracking
5721 cannot pass to the left of the verb. However, when one of these verbs
5722 appears inside an atomic group, its effect is confined to that group,
5723 because once the group has been matched, there is never any backtrack-
5724 ing into it. In this situation, backtracking can "jump back" to the
5725 left of the entire atomic group. (Remember also, as stated above, that
5726 this localization also applies in subroutine calls and assertions.)
5727
5728 These verbs differ in exactly what kind of failure occurs when back-
5729 tracking reaches them.
5730
5731 (*COMMIT)
5732
5733 This verb, which may not be followed by a name, causes the whole match
5734 to fail outright if the rest of the pattern does not match. Even if the
5735 pattern is unanchored, no further attempts to find a match by advancing
5736 the starting point take place. Once (*COMMIT) has been passed,
5737 pcre_exec() is committed to finding a match at the current starting
5738 point, or not at all. For example:
5739
5740 a+(*COMMIT)b
5741
5742 This matches "xxaab" but not "aacaab". It can be thought of as a kind
5743 of dynamic anchor, or "I've started, so I must finish." The name of the
5744 most recently passed (*MARK) in the path is passed back when (*COMMIT)
5745 forces a match failure.
5746
5747 Note that (*COMMIT) at the start of a pattern is not the same as an
5748 anchor, unless PCRE's start-of-match optimizations are turned off, as
5749 shown in this pcretest example:
5750
5751 /(*COMMIT)abc/
5752 xyzabc
5753 0: abc
5754 xyzabc\Y
5755 No match
5756
5757 PCRE knows that any match must start with "a", so the optimization
5758 skips along the subject to "a" before running the first match attempt,
5759 which succeeds. When the optimization is disabled by the \Y escape in
5760 the second subject, the match starts at "x" and so the (*COMMIT) causes
5761 it to fail without trying any other starting points.
5762
5763 (*PRUNE) or (*PRUNE:NAME)
5764
5765 This verb causes the match to fail at the current starting position in
5766 the subject if the rest of the pattern does not match. If the pattern
5767 is unanchored, the normal "bumpalong" advance to the next starting
5768 character then happens. Backtracking can occur as usual to the left of
5769 (*PRUNE), before it is reached, or when matching to the right of
5770 (*PRUNE), but if there is no match to the right, backtracking cannot
5771 cross (*PRUNE). In simple cases, the use of (*PRUNE) is just an alter-
5772 native to an atomic group or possessive quantifier, but there are some
5773 uses of (*PRUNE) that cannot be expressed in any other way. The behav-
5774 iour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE) when the
5775 match fails completely; the name is passed back if this is the final
5776 attempt. (*PRUNE:NAME) does not pass back a name if the match suc-
5777 ceeds. In an anchored pattern (*PRUNE) has the same effect as (*COM-
5778 MIT).
5779
5780 (*SKIP)
5781
5782 This verb, when given without a name, is like (*PRUNE), except that if
5783 the pattern is unanchored, the "bumpalong" advance is not to the next
5784 character, but to the position in the subject where (*SKIP) was encoun-
5785 tered. (*SKIP) signifies that whatever text was matched leading up to
5786 it cannot be part of a successful match. Consider:
5787
5788 a+(*SKIP)b
5789
5790 If the subject is "aaaac...", after the first match attempt fails
5791 (starting at the first character in the string), the starting point
5792 skips on to start the next attempt at "c". Note that a possessive quan-
5793 tifer does not have the same effect as this example; although it would
5794 suppress backtracking during the first match attempt, the second
5795 attempt would start at the second character instead of skipping on to
5796 "c".
5797
5798 (*SKIP:NAME)
5799
5800 When (*SKIP) has an associated name, its behaviour is modified. If the
5801 following pattern fails to match, the previous path through the pattern
5802 is searched for the most recent (*MARK) that has the same name. If one
5803 is found, the "bumpalong" advance is to the subject position that cor-
5804 responds to that (*MARK) instead of to where (*SKIP) was encountered.
5805 If no (*MARK) with a matching name is found, normal "bumpalong" of one
5806 character happens (the (*SKIP) is ignored).
5807
5808 (*THEN) or (*THEN:NAME)
5809
5810 This verb causes a skip to the next alternation in the innermost
5811 enclosing group if the rest of the pattern does not match. That is, it
5812 cancels pending backtracking, but only within the current alternation.
5813 Its name comes from the observation that it can be used for a pattern-
5814 based if-then-else block:
5815
5816 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
5817
5818 If the COND1 pattern matches, FOO is tried (and possibly further items
5819 after the end of the group if FOO succeeds); on failure the matcher
5820 skips to the second alternative and tries COND2, without backtracking
5821 into COND1. The behaviour of (*THEN:NAME) is exactly the same as
5822 (*MARK:NAME)(*THEN) if the overall match fails. If (*THEN) is not
5823 directly inside an alternation, it acts like (*PRUNE).
5824
5825 The above verbs provide four different "strengths" of control when sub-
5826 sequent matching fails. (*THEN) is the weakest, carrying on the match
5827 at the next alternation. (*PRUNE) comes next, failing the match at the
5828 current starting position, but allowing an advance to the next charac-
5829 ter (for an unanchored pattern). (*SKIP) is similar, except that the
5830 advance may be more than one character. (*COMMIT) is the strongest,
5831 causing the entire match to fail.
5832
5833 If more than one is present in a pattern, the "stongest" one wins. For
5834 example, consider this pattern, where A, B, etc. are complex pattern
5835 fragments:
5836
5837 (A(*COMMIT)B(*THEN)C|D)
5838
5839 Once A has matched, PCRE is committed to this match, at the current
5840 starting position. If subsequently B matches, but C does not, the nor-
5841 mal (*THEN) action of trying the next alternation (that is, D) does not
5842 happen because (*COMMIT) overrides.
5843
5844
5845 SEE ALSO
5846
5847 pcreapi(3), pcrecallout(3), pcrematching(3), pcresyntax(3), pcre(3).
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: 24 July 2011
5860 Copyright (c) 1997-2011 University of Cambridge.
5861 ------------------------------------------------------------------------------
5862
5863
5864 PCRESYNTAX(3) PCRESYNTAX(3)
5865
5866
5867 NAME
5868 PCRE - Perl-compatible regular expressions
5869
5870
5871 PCRE REGULAR EXPRESSION SYNTAX SUMMARY
5872
5873 The full syntax and semantics of the regular expressions that are sup-
5874 ported by PCRE are described in the pcrepattern documentation. This
5875 document contains just a quick-reference summary of the syntax.
5876
5877
5878 QUOTING
5879
5880 \x where x is non-alphanumeric is a literal x
5881 \Q...\E treat enclosed characters as literal
5882
5883
5884 CHARACTERS
5885
5886 \a alarm, that is, the BEL character (hex 07)
5887 \cx "control-x", where x is any ASCII character
5888 \e escape (hex 1B)
5889 \f formfeed (hex 0C)
5890 \n newline (hex 0A)
5891 \r carriage return (hex 0D)
5892 \t tab (hex 09)
5893 \ddd character with octal code ddd, or backreference
5894 \xhh character with hex code hh
5895 \x{hhh..} character with hex code hhh..
5896
5897
5898 CHARACTER TYPES
5899
5900 . any character except newline;
5901 in dotall mode, any character whatsoever
5902 \C one byte, even in UTF-8 mode (best avoided)
5903 \d a decimal digit
5904 \D a character that is not a decimal digit
5905 \h a horizontal whitespace character
5906 \H a character that is not a horizontal whitespace character
5907 \N a character that is not a newline
5908 \p{xx} a character with the xx property
5909 \P{xx} a character without the xx property
5910 \R a newline sequence
5911 \s a whitespace character
5912 \S a character that is not a whitespace character
5913 \v a vertical whitespace character
5914 \V a character that is not a vertical whitespace character
5915 \w a "word" character
5916 \W a "non-word" character
5917 \X an extended Unicode sequence
5918
5919 In PCRE, by default, \d, \D, \s, \S, \w, and \W recognize only ASCII
5920 characters, even in UTF-8 mode. However, this can be changed by setting
5921 the PCRE_UCP option.
5922
5923
5924 GENERAL CATEGORY PROPERTIES FOR \p and \P
5925
5926 C Other
5927 Cc Control
5928 Cf Format
5929 Cn Unassigned
5930 Co Private use
5931 Cs Surrogate
5932
5933 L Letter
5934 Ll Lower case letter
5935 Lm Modifier letter
5936 Lo Other letter
5937 Lt Title case letter
5938 Lu Upper case letter
5939 L& Ll, Lu, or Lt
5940
5941 M Mark
5942 Mc Spacing mark
5943 Me Enclosing mark
5944 Mn Non-spacing mark
5945
5946 N Number
5947 Nd Decimal number
5948 Nl Letter number
5949 No Other number
5950
5951 P Punctuation
5952 Pc Connector punctuation
5953 Pd Dash punctuation
5954 Pe Close punctuation
5955 Pf Final punctuation
5956 Pi Initial punctuation
5957 Po Other punctuation
5958 Ps Open punctuation
5959
5960 S Symbol
5961 Sc Currency symbol
5962 Sk Modifier symbol
5963 Sm Mathematical symbol
5964 So Other symbol
5965
5966 Z Separator
5967 Zl Line separator
5968 Zp Paragraph separator
5969 Zs Space separator
5970
5971
5972 PCRE SPECIAL CATEGORY PROPERTIES FOR \p and \P
5973
5974 Xan Alphanumeric: union of properties L and N
5975 Xps POSIX space: property Z or tab, NL, VT, FF, CR
5976 Xsp Perl space: property Z or tab, NL, FF, CR
5977 Xwd Perl word: property Xan or underscore
5978
5979
5980 SCRIPT NAMES FOR \p AND \P
5981
5982 Arabic, Armenian, Avestan, Balinese, Bamum, Bengali, Bopomofo, Braille,
5983 Buginese, Buhid, Canadian_Aboriginal, Carian, Cham, Cherokee, Common,
5984 Coptic, Cuneiform, Cypriot, Cyrillic, Deseret, Devanagari, Egyp-
5985 tian_Hieroglyphs, Ethiopic, Georgian, Glagolitic, Gothic, Greek,
5986 Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hiragana, Impe-
5987 rial_Aramaic, Inherited, Inscriptional_Pahlavi, Inscriptional_Parthian,
5988 Javanese, Kaithi, Kannada, Katakana, Kayah_Li, Kharoshthi, Khmer, Lao,
5989 Latin, Lepcha, Limbu, Linear_B, Lisu, Lycian, Lydian, Malayalam,
5990 Meetei_Mayek, Mongolian, Myanmar, New_Tai_Lue, Nko, Ogham, Old_Italic,
5991 Old_Persian, Old_South_Arabian, Old_Turkic, Ol_Chiki, Oriya, Osmanya,
5992 Phags_Pa, Phoenician, Rejang, Runic, Samaritan, Saurashtra, Shavian,
5993 Sinhala, Sundanese, Syloti_Nagri, Syriac, Tagalog, Tagbanwa, Tai_Le,
5994 Tai_Tham, Tai_Viet, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh,
5995 Ugaritic, Vai, Yi.
5996
5997
5998 CHARACTER CLASSES
5999
6000 [...] positive character class
6001 [^...] negative character class
6002 [x-y] range (can be used for hex characters)
6003 [[:xxx:]] positive POSIX named set
6004 [[:^xxx:]] negative POSIX named set
6005
6006 alnum alphanumeric
6007 alpha alphabetic
6008 ascii 0-127
6009 blank space or tab
6010 cntrl control character
6011 digit decimal digit
6012 graph printing, excluding space
6013 lower lower case letter
6014 print printing, including space
6015 punct printing, excluding alphanumeric
6016 space whitespace
6017 upper upper case letter
6018