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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 211 - (show annotations) (download)
Thu Aug 9 09:52:43 2007 UTC (7 years, 2 months ago) by ph10
File MIME type: text/plain
File size: 294599 byte(s)
Update UTF-8 validity check and documentation.

1 -----------------------------------------------------------------------------
2 This file contains a concatenation of the PCRE man pages, converted to plain
3 text format for ease of searching with a text editor, or for use on systems
4 that do not have a man page processor. The small individual files that give
5 synopses of each function in the library have not been included. There are
6 separate text files for the pcregrep and pcretest commands.
7 -----------------------------------------------------------------------------
8
9
10 PCRE(3) PCRE(3)
11
12
13 NAME
14 PCRE - Perl-compatible regular expressions
15
16
17 INTRODUCTION
18
19 The PCRE library is a set of functions that implement regular expres-
20 sion pattern matching using the same syntax and semantics as Perl, with
21 just a few differences. (Certain features that appeared in Python and
22 PCRE before they appeared in Perl are also available using the Python
23 syntax.)
24
25 The current implementation of PCRE (release 7.x) corresponds approxi-
26 mately with Perl 5.10, including support for UTF-8 encoded strings and
27 Unicode general category properties. However, UTF-8 and Unicode support
28 has to be explicitly enabled; it is not the default. The Unicode tables
29 correspond to Unicode release 5.0.0.
30
31 In addition to the Perl-compatible matching function, PCRE contains an
32 alternative matching function that matches the same compiled patterns
33 in a different way. In certain circumstances, the alternative function
34 has some advantages. For a discussion of the two matching algorithms,
35 see the pcrematching page.
36
37 PCRE is written in C and released as a C library. A number of people
38 have written wrappers and interfaces of various kinds. In particular,
39 Google Inc. have provided a comprehensive C++ wrapper. This is now
40 included as part of the PCRE distribution. The pcrecpp page has details
41 of this interface. Other people's contributions can be found in the
42 Contrib directory at the primary FTP site, which is:
43
44 ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre
45
46 Details of exactly which Perl regular expression features are and are
47 not supported by PCRE are given in separate documents. See the pcrepat-
48 tern and pcrecompat pages. There is a syntax summary in the pcresyntax
49 page.
50
51 Some features of PCRE can be included, excluded, or changed when the
52 library is built. The pcre_config() function makes it possible for a
53 client to discover which features are available. The features them-
54 selves are described in the pcrebuild page. Documentation about build-
55 ing PCRE for various operating systems can be found in the README file
56 in the source distribution.
57
58 The library contains a number of undocumented internal functions and
59 data tables that are used by more than one of the exported external
60 functions, but which are not intended for use by external callers.
61 Their names all begin with "_pcre_", which hopefully will not provoke
62 any name clashes. In some environments, it is possible to control which
63 external symbols are exported when a shared library is built, and in
64 these cases the undocumented symbols are not exported.
65
66
67 USER DOCUMENTATION
68
69 The user documentation for PCRE comprises a number of different sec-
70 tions. In the "man" format, each of these is a separate "man page". In
71 the HTML format, each is a separate page, linked from the index page.
72 In the plain text format, all the sections are concatenated, for ease
73 of searching. The sections are as follows:
74
75 pcre this document
76 pcre-config show PCRE installation configuration information
77 pcreapi details of PCRE's native C API
78 pcrebuild options for building PCRE
79 pcrecallout details of the callout feature
80 pcrecompat discussion of Perl compatibility
81 pcrecpp details of the C++ wrapper
82 pcregrep description of the pcregrep command
83 pcrematching discussion of the two matching algorithms
84 pcrepartial details of the partial matching facility
85 pcrepattern syntax and semantics of supported
86 regular expressions
87 pcresyntax quick syntax reference
88 pcreperform discussion of performance issues
89 pcreposix the POSIX-compatible C API
90 pcreprecompile details of saving and re-using precompiled patterns
91 pcresample discussion of the sample program
92 pcrestack discussion of stack usage
93 pcretest description of the pcretest testing command
94
95 In addition, in the "man" and HTML formats, there is a short page for
96 each C library function, listing its arguments and results.
97
98
99 LIMITATIONS
100
101 There are some size limitations in PCRE but it is hoped that they will
102 never in practice be relevant.
103
104 The maximum length of a compiled pattern is 65539 (sic) bytes if PCRE
105 is compiled with the default internal linkage size of 2. If you want to
106 process regular expressions that are truly enormous, you can compile
107 PCRE with an internal linkage size of 3 or 4 (see the README file in
108 the source distribution and the pcrebuild documentation for details).
109 In these cases the limit is substantially larger. However, the speed
110 of execution is slower.
111
112 All values in repeating quantifiers must be less than 65536.
113
114 There is no limit to the number of parenthesized subpatterns, but there
115 can be no more than 65535 capturing subpatterns.
116
117 The maximum length of name for a named subpattern is 32 characters, and
118 the maximum number of named subpatterns is 10000.
119
120 The maximum length of a subject string is the largest positive number
121 that an integer variable can hold. However, when using the traditional
122 matching function, PCRE uses recursion to handle subpatterns and indef-
123 inite repetition. This means that the available stack space may limit
124 the size of a subject string that can be processed by certain patterns.
125 For a discussion of stack issues, see the pcrestack documentation.
126
127
128 UTF-8 AND UNICODE PROPERTY SUPPORT
129
130 From release 3.3, PCRE has had some support for character strings
131 encoded in the UTF-8 format. For release 4.0 this was greatly extended
132 to cover most common requirements, and in release 5.0 additional sup-
133 port for Unicode general category properties was added.
134
135 In order process UTF-8 strings, you must build PCRE to include UTF-8
136 support in the code, and, in addition, you must call pcre_compile()
137 with the PCRE_UTF8 option flag. When you do this, both the pattern and
138 any subject strings that are matched against it are treated as UTF-8
139 strings instead of just strings of bytes.
140
141 If you compile PCRE with UTF-8 support, but do not use it at run time,
142 the library will be a bit bigger, but the additional run time overhead
143 is limited to testing the PCRE_UTF8 flag occasionally, so should not be
144 very big.
145
146 If PCRE is built with Unicode character property support (which implies
147 UTF-8 support), the escape sequences \p{..}, \P{..}, and \X are sup-
148 ported. The available properties that can be tested are limited to the
149 general category properties such as Lu for an upper case letter or Nd
150 for a decimal number, the Unicode script names such as Arabic or Han,
151 and the derived properties Any and L&. A full list is given in the
152 pcrepattern documentation. Only the short names for properties are sup-
153 ported. For example, \p{L} matches a letter. Its Perl synonym, \p{Let-
154 ter}, is not supported. Furthermore, in Perl, many properties may
155 optionally be prefixed by "Is", for compatibility with Perl 5.6. PCRE
156 does not support this.
157
158 Validity of UTF-8 strings
159
160 When you set the PCRE_UTF8 flag, the strings passed as patterns and
161 subjects are (by default) checked for validity on entry to the relevant
162 functions. From release 7.3 of PCRE, the check is according the rules
163 of RFC 3629, which are themselves derived from the Unicode specifica-
164 tion. Earlier releases of PCRE followed the rules of RFC 2279, which
165 allows the full range of 31-bit values (0 to 0x7FFFFFFF). The current
166 check allows only values in the range U+0 to U+10FFFF, excluding U+D800
167 to U+DFFF.
168
169 The excluded code points are the "Low Surrogate Area" of Unicode, of
170 which the Unicode Standard says this: "The Low Surrogate Area does not
171 contain any character assignments, consequently no character code
172 charts or namelists are provided for this area. Surrogates are reserved
173 for use with UTF-16 and then must be used in pairs." The code points
174 that are encoded by UTF-16 pairs are available as independent code
175 points in the UTF-8 encoding. (In other words, the whole surrogate
176 thing is a fudge for UTF-16 which unfortunately messes up UTF-8.)
177
178 If an invalid UTF-8 string is passed to PCRE, an error return
179 (PCRE_ERROR_BADUTF8) is given. In some situations, you may already know
180 that your strings are valid, and therefore want to skip these checks in
181 order to improve performance. If you set the PCRE_NO_UTF8_CHECK flag at
182 compile time or at run time, PCRE assumes that the pattern or subject
183 it is given (respectively) contains only valid UTF-8 codes. In this
184 case, it does not diagnose an invalid UTF-8 string.
185
186 If you pass an invalid UTF-8 string when PCRE_NO_UTF8_CHECK is set,
187 what happens depends on why the string is invalid. If the string con-
188 forms to the "old" definition of UTF-8 (RFC 2279), it is processed as a
189 string of characters in the range 0 to 0x7FFFFFFF. In other words,
190 apart from the initial validity test, PCRE (when in UTF-8 mode) handles
191 strings according to the more liberal rules of RFC 2279. However, if
192 the string does not even conform to RFC 2279, the result is undefined.
193 Your program may crash.
194
195 If you want to process strings of values in the full range 0 to
196 0x7FFFFFFF, encoded in a UTF-8-like manner as per the old RFC, you can
197 set PCRE_NO_UTF8_CHECK to bypass the more restrictive test. However, in
198 this situation, you will have to apply your own validity check.
199
200 General comments about UTF-8 mode
201
202 1. An unbraced hexadecimal escape sequence (such as \xb3) matches a
203 two-byte UTF-8 character if the value is greater than 127.
204
205 2. Octal numbers up to \777 are recognized, and match two-byte UTF-8
206 characters for values greater than \177.
207
208 3. Repeat quantifiers apply to complete UTF-8 characters, not to indi-
209 vidual bytes, for example: \x{100}{3}.
210
211 4. The dot metacharacter matches one UTF-8 character instead of a sin-
212 gle byte.
213
214 5. The escape sequence \C can be used to match a single byte in UTF-8
215 mode, but its use can lead to some strange effects. This facility is
216 not available in the alternative matching function, pcre_dfa_exec().
217
218 6. The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly
219 test characters of any code value, but the characters that PCRE recog-
220 nizes as digits, spaces, or word characters remain the same set as
221 before, all with values less than 256. This remains true even when PCRE
222 includes Unicode property support, because to do otherwise would slow
223 down PCRE in many common cases. If you really want to test for a wider
224 sense of, say, "digit", you must use Unicode property tests such as
225 \p{Nd}.
226
227 7. Similarly, characters that match the POSIX named character classes
228 are all low-valued characters.
229
230 8. However, the Perl 5.10 horizontal and vertical whitespace matching
231 escapes (\h, \H, \v, and \V) do match all the appropriate Unicode char-
232 acters.
233
234 9. Case-insensitive matching applies only to characters whose values
235 are less than 128, unless PCRE is built with Unicode property support.
236 Even when Unicode property support is available, PCRE still uses its
237 own character tables when checking the case of low-valued characters,
238 so as not to degrade performance. The Unicode property information is
239 used only for characters with higher values. Even when Unicode property
240 support is available, PCRE supports case-insensitive matching only when
241 there is a one-to-one mapping between a letter's cases. There are a
242 small number of many-to-one mappings in Unicode; these are not sup-
243 ported by PCRE.
244
245
246 AUTHOR
247
248 Philip Hazel
249 University Computing Service
250 Cambridge CB2 3QH, England.
251
252 Putting an actual email address here seems to have been a spam magnet,
253 so I've taken it away. If you want to email me, use my two initials,
254 followed by the two digits 10, at the domain cam.ac.uk.
255
256
257 REVISION
258
259 Last updated: 09 August 2007
260 Copyright (c) 1997-2007 University of Cambridge.
261 ------------------------------------------------------------------------------
262
263
264 PCREBUILD(3) PCREBUILD(3)
265
266
267 NAME
268 PCRE - Perl-compatible regular expressions
269
270
271 PCRE BUILD-TIME OPTIONS
272
273 This document describes the optional features of PCRE that can be
274 selected when the library is compiled. They are all selected, or dese-
275 lected, by providing options to the configure script that is run before
276 the make command. The complete list of options for configure (which
277 includes the standard ones such as the selection of the installation
278 directory) can be obtained by running
279
280 ./configure --help
281
282 The following sections include descriptions of options whose names
283 begin with --enable or --disable. These settings specify changes to the
284 defaults for the configure command. Because of the way that configure
285 works, --enable and --disable always come in pairs, so the complemen-
286 tary option always exists as well, but as it specifies the default, it
287 is not described.
288
289
290 C++ SUPPORT
291
292 By default, the configure script will search for a C++ compiler and C++
293 header files. If it finds them, it automatically builds the C++ wrapper
294 library for PCRE. You can disable this by adding
295
296 --disable-cpp
297
298 to the configure command.
299
300
301 UTF-8 SUPPORT
302
303 To build PCRE with support for UTF-8 character strings, add
304
305 --enable-utf8
306
307 to the configure command. Of itself, this does not make PCRE treat
308 strings as UTF-8. As well as compiling PCRE with this option, you also
309 have have to set the PCRE_UTF8 option when you call the pcre_compile()
310 function.
311
312
313 UNICODE CHARACTER PROPERTY SUPPORT
314
315 UTF-8 support allows PCRE to process character values greater than 255
316 in the strings that it handles. On its own, however, it does not pro-
317 vide any facilities for accessing the properties of such characters. If
318 you want to be able to use the pattern escapes \P, \p, and \X, which
319 refer to Unicode character properties, you must add
320
321 --enable-unicode-properties
322
323 to the configure command. This implies UTF-8 support, even if you have
324 not explicitly requested it.
325
326 Including Unicode property support adds around 30K of tables to the
327 PCRE library. Only the general category properties such as Lu and Nd
328 are supported. Details are given in the pcrepattern documentation.
329
330
331 CODE VALUE OF NEWLINE
332
333 By default, PCRE interprets character 10 (linefeed, LF) as indicating
334 the end of a line. This is the normal newline character on Unix-like
335 systems. You can compile PCRE to use character 13 (carriage return, CR)
336 instead, by adding
337
338 --enable-newline-is-cr
339
340 to the configure command. There is also a --enable-newline-is-lf
341 option, which explicitly specifies linefeed as the newline character.
342
343 Alternatively, you can specify that line endings are to be indicated by
344 the two character sequence CRLF. If you want this, add
345
346 --enable-newline-is-crlf
347
348 to the configure command. There is a fourth option, specified by
349
350 --enable-newline-is-anycrlf
351
352 which causes PCRE to recognize any of the three sequences CR, LF, or
353 CRLF as indicating a line ending. Finally, a fifth option, specified by
354
355 --enable-newline-is-any
356
357 causes PCRE to recognize any Unicode newline sequence.
358
359 Whatever line ending convention is selected when PCRE is built can be
360 overridden when the library functions are called. At build time it is
361 conventional to use the standard for your operating system.
362
363
364 BUILDING SHARED AND STATIC LIBRARIES
365
366 The PCRE building process uses libtool to build both shared and static
367 Unix libraries by default. You can suppress one of these by adding one
368 of
369
370 --disable-shared
371 --disable-static
372
373 to the configure command, as required.
374
375
376 POSIX MALLOC USAGE
377
378 When PCRE is called through the POSIX interface (see the pcreposix doc-
379 umentation), additional working storage is required for holding the
380 pointers to capturing substrings, because PCRE requires three integers
381 per substring, whereas the POSIX interface provides only two. If the
382 number of expected substrings is small, the wrapper function uses space
383 on the stack, because this is faster than using malloc() for each call.
384 The default threshold above which the stack is no longer used is 10; it
385 can be changed by adding a setting such as
386
387 --with-posix-malloc-threshold=20
388
389 to the configure command.
390
391
392 HANDLING VERY LARGE PATTERNS
393
394 Within a compiled pattern, offset values are used to point from one
395 part to another (for example, from an opening parenthesis to an alter-
396 nation metacharacter). By default, two-byte values are used for these
397 offsets, leading to a maximum size for a compiled pattern of around
398 64K. This is sufficient to handle all but the most gigantic patterns.
399 Nevertheless, some people do want to process enormous patterns, so it
400 is possible to compile PCRE to use three-byte or four-byte offsets by
401 adding a setting such as
402
403 --with-link-size=3
404
405 to the configure command. The value given must be 2, 3, or 4. Using
406 longer offsets slows down the operation of PCRE because it has to load
407 additional bytes when handling them.
408
409
410 AVOIDING EXCESSIVE STACK USAGE
411
412 When matching with the pcre_exec() function, PCRE implements backtrack-
413 ing by making recursive calls to an internal function called match().
414 In environments where the size of the stack is limited, this can se-
415 verely limit PCRE's operation. (The Unix environment does not usually
416 suffer from this problem, but it may sometimes be necessary to increase
417 the maximum stack size. There is a discussion in the pcrestack docu-
418 mentation.) An alternative approach to recursion that uses memory from
419 the heap to remember data, instead of using recursive function calls,
420 has been implemented to work round the problem of limited stack size.
421 If you want to build a version of PCRE that works this way, add
422
423 --disable-stack-for-recursion
424
425 to the configure command. With this configuration, PCRE will use the
426 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
427 ment functions. By default these point to malloc() and free(), but you
428 can replace the pointers so that your own functions are used.
429
430 Separate functions are provided rather than using pcre_malloc and
431 pcre_free because the usage is very predictable: the block sizes
432 requested are always the same, and the blocks are always freed in
433 reverse order. A calling program might be able to implement optimized
434 functions that perform better than malloc() and free(). PCRE runs
435 noticeably more slowly when built in this way. This option affects only
436 the pcre_exec() function; it is not relevant for the the
437 pcre_dfa_exec() function.
438
439
440 LIMITING PCRE RESOURCE USAGE
441
442 Internally, PCRE has a function called match(), which it calls repeat-
443 edly (sometimes recursively) when matching a pattern with the
444 pcre_exec() function. By controlling the maximum number of times this
445 function may be called during a single matching operation, a limit can
446 be placed on the resources used by a single call to pcre_exec(). The
447 limit can be changed at run time, as described in the pcreapi documen-
448 tation. The default is 10 million, but this can be changed by adding a
449 setting such as
450
451 --with-match-limit=500000
452
453 to the configure command. This setting has no effect on the
454 pcre_dfa_exec() matching function.
455
456 In some environments it is desirable to limit the depth of recursive
457 calls of match() more strictly than the total number of calls, in order
458 to restrict the maximum amount of stack (or heap, if --disable-stack-
459 for-recursion is specified) that is used. A second limit controls this;
460 it defaults to the value that is set for --with-match-limit, which
461 imposes no additional constraints. However, you can set a lower limit
462 by adding, for example,
463
464 --with-match-limit-recursion=10000
465
466 to the configure command. This value can also be overridden at run
467 time.
468
469
470 CREATING CHARACTER TABLES AT BUILD TIME
471
472 PCRE uses fixed tables for processing characters whose code values are
473 less than 256. By default, PCRE is built with a set of tables that are
474 distributed in the file pcre_chartables.c.dist. These tables are for
475 ASCII codes only. If you add
476
477 --enable-rebuild-chartables
478
479 to the configure command, the distributed tables are no longer used.
480 Instead, a program called dftables is compiled and run. This outputs
481 the source for new set of tables, created in the default locale of your
482 C runtime system. (This method of replacing the tables does not work if
483 you are cross compiling, because dftables is run on the local host. If
484 you need to create alternative tables when cross compiling, you will
485 have to do so "by hand".)
486
487
488 USING EBCDIC CODE
489
490 PCRE assumes by default that it will run in an environment where the
491 character code is ASCII (or Unicode, which is a superset of ASCII).
492 This is the case for most computer operating systems. PCRE can, how-
493 ever, be compiled to run in an EBCDIC environment by adding
494
495 --enable-ebcdic
496
497 to the configure command. This setting implies --enable-rebuild-charta-
498 bles. You should only use it if you know that you are in an EBCDIC
499 environment (for example, an IBM mainframe operating system).
500
501
502 SEE ALSO
503
504 pcreapi(3), pcre_config(3).
505
506
507 AUTHOR
508
509 Philip Hazel
510 University Computing Service
511 Cambridge CB2 3QH, England.
512
513
514 REVISION
515
516 Last updated: 30 July 2007
517 Copyright (c) 1997-2007 University of Cambridge.
518 ------------------------------------------------------------------------------
519
520
521 PCREMATCHING(3) PCREMATCHING(3)
522
523
524 NAME
525 PCRE - Perl-compatible regular expressions
526
527
528 PCRE MATCHING ALGORITHMS
529
530 This document describes the two different algorithms that are available
531 in PCRE for matching a compiled regular expression against a given sub-
532 ject string. The "standard" algorithm is the one provided by the
533 pcre_exec() function. This works in the same was as Perl's matching
534 function, and provides a Perl-compatible matching operation.
535
536 An alternative algorithm is provided by the pcre_dfa_exec() function;
537 this operates in a different way, and is not Perl-compatible. It has
538 advantages and disadvantages compared with the standard algorithm, and
539 these are described below.
540
541 When there is only one possible way in which a given subject string can
542 match a pattern, the two algorithms give the same answer. A difference
543 arises, however, when there are multiple possibilities. For example, if
544 the pattern
545
546 ^<.*>
547
548 is matched against the string
549
550 <something> <something else> <something further>
551
552 there are three possible answers. The standard algorithm finds only one
553 of them, whereas the alternative algorithm finds all three.
554
555
556 REGULAR EXPRESSIONS AS TREES
557
558 The set of strings that are matched by a regular expression can be rep-
559 resented as a tree structure. An unlimited repetition in the pattern
560 makes the tree of infinite size, but it is still a tree. Matching the
561 pattern to a given subject string (from a given starting point) can be
562 thought of as a search of the tree. There are two ways to search a
563 tree: depth-first and breadth-first, and these correspond to the two
564 matching algorithms provided by PCRE.
565
566
567 THE STANDARD MATCHING ALGORITHM
568
569 In the terminology of Jeffrey Friedl's book "Mastering Regular Expres-
570 sions", the standard algorithm is an "NFA algorithm". It conducts a
571 depth-first search of the pattern tree. That is, it proceeds along a
572 single path through the tree, checking that the subject matches what is
573 required. When there is a mismatch, the algorithm tries any alterna-
574 tives at the current point, and if they all fail, it backs up to the
575 previous branch point in the tree, and tries the next alternative
576 branch at that level. This often involves backing up (moving to the
577 left) in the subject string as well. The order in which repetition
578 branches are tried is controlled by the greedy or ungreedy nature of
579 the quantifier.
580
581 If a leaf node is reached, a matching string has been found, and at
582 that point the algorithm stops. Thus, if there is more than one possi-
583 ble match, this algorithm returns the first one that it finds. Whether
584 this is the shortest, the longest, or some intermediate length depends
585 on the way the greedy and ungreedy repetition quantifiers are specified
586 in the pattern.
587
588 Because it ends up with a single path through the tree, it is rela-
589 tively straightforward for this algorithm to keep track of the sub-
590 strings that are matched by portions of the pattern in parentheses.
591 This provides support for capturing parentheses and back references.
592
593
594 THE ALTERNATIVE MATCHING ALGORITHM
595
596 This algorithm conducts a breadth-first search of the tree. Starting
597 from the first matching point in the subject, it scans the subject
598 string from left to right, once, character by character, and as it does
599 this, it remembers all the paths through the tree that represent valid
600 matches. In Friedl's terminology, this is a kind of "DFA algorithm",
601 though it is not implemented as a traditional finite state machine (it
602 keeps multiple states active simultaneously).
603
604 The scan continues until either the end of the subject is reached, or
605 there are no more unterminated paths. At this point, terminated paths
606 represent the different matching possibilities (if there are none, the
607 match has failed). Thus, if there is more than one possible match,
608 this algorithm finds all of them, and in particular, it finds the long-
609 est. In PCRE, there is an option to stop the algorithm after the first
610 match (which is necessarily the shortest) has been found.
611
612 Note that all the matches that are found start at the same point in the
613 subject. If the pattern
614
615 cat(er(pillar)?)
616
617 is matched against the string "the caterpillar catchment", the result
618 will be the three strings "cat", "cater", and "caterpillar" that start
619 at the fourth character of the subject. The algorithm does not automat-
620 ically move on to find matches that start at later positions.
621
622 There are a number of features of PCRE regular expressions that are not
623 supported by the alternative matching algorithm. They are as follows:
624
625 1. Because the algorithm finds all possible matches, the greedy or
626 ungreedy nature of repetition quantifiers is not relevant. Greedy and
627 ungreedy quantifiers are treated in exactly the same way. However, pos-
628 sessive quantifiers can make a difference when what follows could also
629 match what is quantified, for example in a pattern like this:
630
631 ^a++\w!
632
633 This pattern matches "aaab!" but not "aaa!", which would be matched by
634 a non-possessive quantifier. Similarly, if an atomic group is present,
635 it is matched as if it were a standalone pattern at the current point,
636 and the longest match is then "locked in" for the rest of the overall
637 pattern.
638
639 2. When dealing with multiple paths through the tree simultaneously, it
640 is not straightforward to keep track of captured substrings for the
641 different matching possibilities, and PCRE's implementation of this
642 algorithm does not attempt to do this. This means that no captured sub-
643 strings are available.
644
645 3. Because no substrings are captured, back references within the pat-
646 tern are not supported, and cause errors if encountered.
647
648 4. For the same reason, conditional expressions that use a backrefer-
649 ence as the condition or test for a specific group recursion are not
650 supported.
651
652 5. Because many paths through the tree may be active, the \K escape
653 sequence, which resets the start of the match when encountered (but may
654 be on some paths and not on others), is not supported. It causes an
655 error if encountered.
656
657 6. Callouts are supported, but the value of the capture_top field is
658 always 1, and the value of the capture_last field is always -1.
659
660 7. The \C escape sequence, which (in the standard algorithm) matches a
661 single byte, even in UTF-8 mode, is not supported because the alterna-
662 tive algorithm moves through the subject string one character at a
663 time, for all active paths through the tree.
664
665 8. None of the backtracking control verbs such as (*PRUNE) are sup-
666 ported.
667
668
669 ADVANTAGES OF THE ALTERNATIVE ALGORITHM
670
671 Using the alternative matching algorithm provides the following advan-
672 tages:
673
674 1. All possible matches (at a single point in the subject) are automat-
675 ically found, and in particular, the longest match is found. To find
676 more than one match using the standard algorithm, you have to do kludgy
677 things with callouts.
678
679 2. There is much better support for partial matching. The restrictions
680 on the content of the pattern that apply when using the standard algo-
681 rithm for partial matching do not apply to the alternative algorithm.
682 For non-anchored patterns, the starting position of a partial match is
683 available.
684
685 3. Because the alternative algorithm scans the subject string just
686 once, and never needs to backtrack, it is possible to pass very long
687 subject strings to the matching function in several pieces, checking
688 for partial matching each time.
689
690
691 DISADVANTAGES OF THE ALTERNATIVE ALGORITHM
692
693 The alternative algorithm suffers from a number of disadvantages:
694
695 1. It is substantially slower than the standard algorithm. This is
696 partly because it has to search for all possible matches, but is also
697 because it is less susceptible to optimization.
698
699 2. Capturing parentheses and back references are not supported.
700
701 3. Although atomic groups are supported, their use does not provide the
702 performance advantage that it does for the standard algorithm.
703
704
705 AUTHOR
706
707 Philip Hazel
708 University Computing Service
709 Cambridge CB2 3QH, England.
710
711
712 REVISION
713
714 Last updated: 08 August 2007
715 Copyright (c) 1997-2007 University of Cambridge.
716 ------------------------------------------------------------------------------
717
718
719 PCREAPI(3) PCREAPI(3)
720
721
722 NAME
723 PCRE - Perl-compatible regular expressions
724
725
726 PCRE NATIVE API
727
728 #include <pcre.h>
729
730 pcre *pcre_compile(const char *pattern, int options,
731 const char **errptr, int *erroffset,
732 const unsigned char *tableptr);
733
734 pcre *pcre_compile2(const char *pattern, int options,
735 int *errorcodeptr,
736 const char **errptr, int *erroffset,
737 const unsigned char *tableptr);
738
739 pcre_extra *pcre_study(const pcre *code, int options,
740 const char **errptr);
741
742 int pcre_exec(const pcre *code, const pcre_extra *extra,
743 const char *subject, int length, int startoffset,
744 int options, int *ovector, int ovecsize);
745
746 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
747 const char *subject, int length, int startoffset,
748 int options, int *ovector, int ovecsize,
749 int *workspace, int wscount);
750
751 int pcre_copy_named_substring(const pcre *code,
752 const char *subject, int *ovector,
753 int stringcount, const char *stringname,
754 char *buffer, int buffersize);
755
756 int pcre_copy_substring(const char *subject, int *ovector,
757 int stringcount, int stringnumber, char *buffer,
758 int buffersize);
759
760 int pcre_get_named_substring(const pcre *code,
761 const char *subject, int *ovector,
762 int stringcount, const char *stringname,
763 const char **stringptr);
764
765 int pcre_get_stringnumber(const pcre *code,
766 const char *name);
767
768 int pcre_get_stringtable_entries(const pcre *code,
769 const char *name, char **first, char **last);
770
771 int pcre_get_substring(const char *subject, int *ovector,
772 int stringcount, int stringnumber,
773 const char **stringptr);
774
775 int pcre_get_substring_list(const char *subject,
776 int *ovector, int stringcount, const char ***listptr);
777
778 void pcre_free_substring(const char *stringptr);
779
780 void pcre_free_substring_list(const char **stringptr);
781
782 const unsigned char *pcre_maketables(void);
783
784 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
785 int what, void *where);
786
787 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
788
789 int pcre_refcount(pcre *code, int adjust);
790
791 int pcre_config(int what, void *where);
792
793 char *pcre_version(void);
794
795 void *(*pcre_malloc)(size_t);
796
797 void (*pcre_free)(void *);
798
799 void *(*pcre_stack_malloc)(size_t);
800
801 void (*pcre_stack_free)(void *);
802
803 int (*pcre_callout)(pcre_callout_block *);
804
805
806 PCRE API OVERVIEW
807
808 PCRE has its own native API, which is described in this document. There
809 are also some wrapper functions that correspond to the POSIX regular
810 expression API. These are described in the pcreposix documentation.
811 Both of these APIs define a set of C function calls. A C++ wrapper is
812 distributed with PCRE. It is documented in the pcrecpp page.
813
814 The native API C function prototypes are defined in the header file
815 pcre.h, and on Unix systems the library itself is called libpcre. It
816 can normally be accessed by adding -lpcre to the command for linking an
817 application that uses PCRE. The header file defines the macros
818 PCRE_MAJOR and PCRE_MINOR to contain the major and minor release num-
819 bers for the library. Applications can use these to include support
820 for different releases of PCRE.
821
822 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
823 pcre_exec() are used for compiling and matching regular expressions in
824 a Perl-compatible manner. A sample program that demonstrates the sim-
825 plest way of using them is provided in the file called pcredemo.c in
826 the source distribution. The pcresample documentation describes how to
827 run it.
828
829 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
830 ble, is also provided. This uses a different algorithm for the match-
831 ing. The alternative algorithm finds all possible matches (at a given
832 point in the subject), and scans the subject just once. However, this
833 algorithm does not return captured substrings. A description of the two
834 matching algorithms and their advantages and disadvantages is given in
835 the pcrematching documentation.
836
837 In addition to the main compiling and matching functions, there are
838 convenience functions for extracting captured substrings from a subject
839 string that is matched by pcre_exec(). They are:
840
841 pcre_copy_substring()
842 pcre_copy_named_substring()
843 pcre_get_substring()
844 pcre_get_named_substring()
845 pcre_get_substring_list()
846 pcre_get_stringnumber()
847 pcre_get_stringtable_entries()
848
849 pcre_free_substring() and pcre_free_substring_list() are also provided,
850 to free the memory used for extracted strings.
851
852 The function pcre_maketables() is used to build a set of character
853 tables in the current locale for passing to pcre_compile(),
854 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
855 provided for specialist use. Most commonly, no special tables are
856 passed, in which case internal tables that are generated when PCRE is
857 built are used.
858
859 The function pcre_fullinfo() is used to find out information about a
860 compiled pattern; pcre_info() is an obsolete version that returns only
861 some of the available information, but is retained for backwards com-
862 patibility. The function pcre_version() returns a pointer to a string
863 containing the version of PCRE and its date of release.
864
865 The function pcre_refcount() maintains a reference count in a data
866 block containing a compiled pattern. This is provided for the benefit
867 of object-oriented applications.
868
869 The global variables pcre_malloc and pcre_free initially contain the
870 entry points of the standard malloc() and free() functions, respec-
871 tively. PCRE calls the memory management functions via these variables,
872 so a calling program can replace them if it wishes to intercept the
873 calls. This should be done before calling any PCRE functions.
874
875 The global variables pcre_stack_malloc and pcre_stack_free are also
876 indirections to memory management functions. These special functions
877 are used only when PCRE is compiled to use the heap for remembering
878 data, instead of recursive function calls, when running the pcre_exec()
879 function. See the pcrebuild documentation for details of how to do
880 this. It is a non-standard way of building PCRE, for use in environ-
881 ments that have limited stacks. Because of the greater use of memory
882 management, it runs more slowly. Separate functions are provided so
883 that special-purpose external code can be used for this case. When
884 used, these functions are always called in a stack-like manner (last
885 obtained, first freed), and always for memory blocks of the same size.
886 There is a discussion about PCRE's stack usage in the pcrestack docu-
887 mentation.
888
889 The global variable pcre_callout initially contains NULL. It can be set
890 by the caller to a "callout" function, which PCRE will then call at
891 specified points during a matching operation. Details are given in the
892 pcrecallout documentation.
893
894
895 NEWLINES
896
897 PCRE supports five different conventions for indicating line breaks in
898 strings: a single CR (carriage return) character, a single LF (line-
899 feed) character, the two-character sequence CRLF, any of the three pre-
900 ceding, or any Unicode newline sequence. The Unicode newline sequences
901 are the three just mentioned, plus the single characters VT (vertical
902 tab, U+000B), FF (formfeed, U+000C), NEL (next line, U+0085), LS (line
903 separator, U+2028), and PS (paragraph separator, U+2029).
904
905 Each of the first three conventions is used by at least one operating
906 system as its standard newline sequence. When PCRE is built, a default
907 can be specified. The default default is LF, which is the Unix stan-
908 dard. When PCRE is run, the default can be overridden, either when a
909 pattern is compiled, or when it is matched.
910
911 In the PCRE documentation the word "newline" is used to mean "the char-
912 acter or pair of characters that indicate a line break". The choice of
913 newline convention affects the handling of the dot, circumflex, and
914 dollar metacharacters, the handling of #-comments in /x mode, and, when
915 CRLF is a recognized line ending sequence, the match position advance-
916 ment for a non-anchored pattern. The choice of newline convention does
917 not affect the interpretation of the \n or \r escape sequences.
918
919
920 MULTITHREADING
921
922 The PCRE functions can be used in multi-threading applications, with
923 the proviso that the memory management functions pointed to by
924 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
925 callout function pointed to by pcre_callout, are shared by all threads.
926
927 The compiled form of a regular expression is not altered during match-
928 ing, so the same compiled pattern can safely be used by several threads
929 at once.
930
931
932 SAVING PRECOMPILED PATTERNS FOR LATER USE
933
934 The compiled form of a regular expression can be saved and re-used at a
935 later time, possibly by a different program, and even on a host other
936 than the one on which it was compiled. Details are given in the
937 pcreprecompile documentation. However, compiling a regular expression
938 with one version of PCRE for use with a different version is not guar-
939 anteed to work and may cause crashes.
940
941
942 CHECKING BUILD-TIME OPTIONS
943
944 int pcre_config(int what, void *where);
945
946 The function pcre_config() makes it possible for a PCRE client to dis-
947 cover which optional features have been compiled into the PCRE library.
948 The pcrebuild documentation has more details about these optional fea-
949 tures.
950
951 The first argument for pcre_config() is an integer, specifying which
952 information is required; the second argument is a pointer to a variable
953 into which the information is placed. The following information is
954 available:
955
956 PCRE_CONFIG_UTF8
957
958 The output is an integer that is set to one if UTF-8 support is avail-
959 able; otherwise it is set to zero.
960
961 PCRE_CONFIG_UNICODE_PROPERTIES
962
963 The output is an integer that is set to one if support for Unicode
964 character properties is available; otherwise it is set to zero.
965
966 PCRE_CONFIG_NEWLINE
967
968 The output is an integer whose value specifies the default character
969 sequence that is recognized as meaning "newline". The four values that
970 are supported are: 10 for LF, 13 for CR, 3338 for CRLF, -2 for ANYCRLF,
971 and -1 for ANY. The default should normally be the standard sequence
972 for your operating system.
973
974 PCRE_CONFIG_LINK_SIZE
975
976 The output is an integer that contains the number of bytes used for
977 internal linkage in compiled regular expressions. The value is 2, 3, or
978 4. Larger values allow larger regular expressions to be compiled, at
979 the expense of slower matching. The default value of 2 is sufficient
980 for all but the most massive patterns, since it allows the compiled
981 pattern to be up to 64K in size.
982
983 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
984
985 The output is an integer that contains the threshold above which the
986 POSIX interface uses malloc() for output vectors. Further details are
987 given in the pcreposix documentation.
988
989 PCRE_CONFIG_MATCH_LIMIT
990
991 The output is an integer that gives the default limit for the number of
992 internal matching function calls in a pcre_exec() execution. Further
993 details are given with pcre_exec() below.
994
995 PCRE_CONFIG_MATCH_LIMIT_RECURSION
996
997 The output is an integer that gives the default limit for the depth of
998 recursion when calling the internal matching function in a pcre_exec()
999 execution. Further details are given with pcre_exec() below.
1000
1001 PCRE_CONFIG_STACKRECURSE
1002
1003 The output is an integer that is set to one if internal recursion when
1004 running pcre_exec() is implemented by recursive function calls that use
1005 the stack to remember their state. This is the usual way that PCRE is
1006 compiled. The output is zero if PCRE was compiled to use blocks of data
1007 on the heap instead of recursive function calls. In this case,
1008 pcre_stack_malloc and pcre_stack_free are called to manage memory
1009 blocks on the heap, thus avoiding the use of the stack.
1010
1011
1012 COMPILING A PATTERN
1013
1014 pcre *pcre_compile(const char *pattern, int options,
1015 const char **errptr, int *erroffset,
1016 const unsigned char *tableptr);
1017
1018 pcre *pcre_compile2(const char *pattern, int options,
1019 int *errorcodeptr,
1020 const char **errptr, int *erroffset,
1021 const unsigned char *tableptr);
1022
1023 Either of the functions pcre_compile() or pcre_compile2() can be called
1024 to compile a pattern into an internal form. The only difference between
1025 the two interfaces is that pcre_compile2() has an additional argument,
1026 errorcodeptr, via which a numerical error code can be returned.
1027
1028 The pattern is a C string terminated by a binary zero, and is passed in
1029 the pattern argument. A pointer to a single block of memory that is
1030 obtained via pcre_malloc is returned. This contains the compiled code
1031 and related data. The pcre type is defined for the returned block; this
1032 is a typedef for a structure whose contents are not externally defined.
1033 It is up to the caller to free the memory (via pcre_free) when it is no
1034 longer required.
1035
1036 Although the compiled code of a PCRE regex is relocatable, that is, it
1037 does not depend on memory location, the complete pcre data block is not
1038 fully relocatable, because it may contain a copy of the tableptr argu-
1039 ment, which is an address (see below).
1040
1041 The options argument contains various bit settings that affect the com-
1042 pilation. It should be zero if no options are required. The available
1043 options are described below. Some of them, in particular, those that
1044 are compatible with Perl, can also be set and unset from within the
1045 pattern (see the detailed description in the pcrepattern documenta-
1046 tion). For these options, the contents of the options argument speci-
1047 fies their initial settings at the start of compilation and execution.
1048 The PCRE_ANCHORED and PCRE_NEWLINE_xxx options can be set at the time
1049 of matching as well as at compile time.
1050
1051 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
1052 if compilation of a pattern fails, pcre_compile() returns NULL, and
1053 sets the variable pointed to by errptr to point to a textual error mes-
1054 sage. This is a static string that is part of the library. You must not
1055 try to free it. The offset from the start of the pattern to the charac-
1056 ter where the error was discovered is placed in the variable pointed to
1057 by erroffset, which must not be NULL. If it is, an immediate error is
1058 given.
1059
1060 If pcre_compile2() is used instead of pcre_compile(), and the error-
1061 codeptr argument is not NULL, a non-zero error code number is returned
1062 via this argument in the event of an error. This is in addition to the
1063 textual error message. Error codes and messages are listed below.
1064
1065 If the final argument, tableptr, is NULL, PCRE uses a default set of
1066 character tables that are built when PCRE is compiled, using the
1067 default C locale. Otherwise, tableptr must be an address that is the
1068 result of a call to pcre_maketables(). This value is stored with the
1069 compiled pattern, and used again by pcre_exec(), unless another table
1070 pointer is passed to it. For more discussion, see the section on locale
1071 support below.
1072
1073 This code fragment shows a typical straightforward call to pcre_com-
1074 pile():
1075
1076 pcre *re;
1077 const char *error;
1078 int erroffset;
1079 re = pcre_compile(
1080 "^A.*Z", /* the pattern */
1081 0, /* default options */
1082 &error, /* for error message */
1083 &erroffset, /* for error offset */
1084 NULL); /* use default character tables */
1085
1086 The following names for option bits are defined in the pcre.h header
1087 file:
1088
1089 PCRE_ANCHORED
1090
1091 If this bit is set, the pattern is forced to be "anchored", that is, it
1092 is constrained to match only at the first matching point in the string
1093 that is being searched (the "subject string"). This effect can also be
1094 achieved by appropriate constructs in the pattern itself, which is the
1095 only way to do it in Perl.
1096
1097 PCRE_AUTO_CALLOUT
1098
1099 If this bit is set, pcre_compile() automatically inserts callout items,
1100 all with number 255, before each pattern item. For discussion of the
1101 callout facility, see the pcrecallout documentation.
1102
1103 PCRE_CASELESS
1104
1105 If this bit is set, letters in the pattern match both upper and lower
1106 case letters. It is equivalent to Perl's /i option, and it can be
1107 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
1108 always understands the concept of case for characters whose values are
1109 less than 128, so caseless matching is always possible. For characters
1110 with higher values, the concept of case is supported if PCRE is com-
1111 piled with Unicode property support, but not otherwise. If you want to
1112 use caseless matching for characters 128 and above, you must ensure
1113 that PCRE is compiled with Unicode property support as well as with
1114 UTF-8 support.
1115
1116 PCRE_DOLLAR_ENDONLY
1117
1118 If this bit is set, a dollar metacharacter in the pattern matches only
1119 at the end of the subject string. Without this option, a dollar also
1120 matches immediately before a newline at the end of the string (but not
1121 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
1122 if PCRE_MULTILINE is set. There is no equivalent to this option in
1123 Perl, and no way to set it within a pattern.
1124
1125 PCRE_DOTALL
1126
1127 If this bit is set, a dot metacharater in the pattern matches all char-
1128 acters, including those that indicate newline. Without it, a dot does
1129 not match when the current position is at a newline. This option is
1130 equivalent to Perl's /s option, and it can be changed within a pattern
1131 by a (?s) option setting. A negative class such as [^a] always matches
1132 newline characters, independent of the setting of this option.
1133
1134 PCRE_DUPNAMES
1135
1136 If this bit is set, names used to identify capturing subpatterns need
1137 not be unique. This can be helpful for certain types of pattern when it
1138 is known that only one instance of the named subpattern can ever be
1139 matched. There are more details of named subpatterns below; see also
1140 the pcrepattern documentation.
1141
1142 PCRE_EXTENDED
1143
1144 If this bit is set, whitespace data characters in the pattern are
1145 totally ignored except when escaped or inside a character class. White-
1146 space does not include the VT character (code 11). In addition, charac-
1147 ters between an unescaped # outside a character class and the next new-
1148 line, inclusive, are also ignored. This is equivalent to Perl's /x
1149 option, and it can be changed within a pattern by a (?x) option set-
1150 ting.
1151
1152 This option makes it possible to include comments inside complicated
1153 patterns. Note, however, that this applies only to data characters.
1154 Whitespace characters may never appear within special character
1155 sequences in a pattern, for example within the sequence (?( which
1156 introduces a conditional subpattern.
1157
1158 PCRE_EXTRA
1159
1160 This option was invented in order to turn on additional functionality
1161 of PCRE that is incompatible with Perl, but it is currently of very
1162 little use. When set, any backslash in a pattern that is followed by a
1163 letter that has no special meaning causes an error, thus reserving
1164 these combinations for future expansion. By default, as in Perl, a
1165 backslash followed by a letter with no special meaning is treated as a
1166 literal. (Perl can, however, be persuaded to give a warning for this.)
1167 There are at present no other features controlled by this option. It
1168 can also be set by a (?X) option setting within a pattern.
1169
1170 PCRE_FIRSTLINE
1171
1172 If this option is set, an unanchored pattern is required to match
1173 before or at the first newline in the subject string, though the
1174 matched text may continue over the newline.
1175
1176 PCRE_MULTILINE
1177
1178 By default, PCRE treats the subject string as consisting of a single
1179 line of characters (even if it actually contains newlines). The "start
1180 of line" metacharacter (^) matches only at the start of the string,
1181 while the "end of line" metacharacter ($) matches only at the end of
1182 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1183 is set). This is the same as Perl.
1184
1185 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1186 constructs match immediately following or immediately before internal
1187 newlines in the subject string, respectively, as well as at the very
1188 start and end. This is equivalent to Perl's /m option, and it can be
1189 changed within a pattern by a (?m) option setting. If there are no new-
1190 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1191 setting PCRE_MULTILINE has no effect.
1192
1193 PCRE_NEWLINE_CR
1194 PCRE_NEWLINE_LF
1195 PCRE_NEWLINE_CRLF
1196 PCRE_NEWLINE_ANYCRLF
1197 PCRE_NEWLINE_ANY
1198
1199 These options override the default newline definition that was chosen
1200 when PCRE was built. Setting the first or the second specifies that a
1201 newline is indicated by a single character (CR or LF, respectively).
1202 Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
1203 two-character CRLF sequence. Setting PCRE_NEWLINE_ANYCRLF specifies
1204 that any of the three preceding sequences should be recognized. Setting
1205 PCRE_NEWLINE_ANY specifies that any Unicode newline sequence should be
1206 recognized. The Unicode newline sequences are the three just mentioned,
1207 plus the single characters VT (vertical tab, U+000B), FF (formfeed,
1208 U+000C), NEL (next line, U+0085), LS (line separator, U+2028), and PS
1209 (paragraph separator, U+2029). The last two are recognized only in
1210 UTF-8 mode.
1211
1212 The newline setting in the options word uses three bits that are
1213 treated as a number, giving eight possibilities. Currently only six are
1214 used (default plus the five values above). This means that if you set
1215 more than one newline option, the combination may or may not be sensi-
1216 ble. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equivalent to
1217 PCRE_NEWLINE_CRLF, but other combinations may yield unused numbers and
1218 cause an error.
1219
1220 The only time that a line break is specially recognized when compiling
1221 a pattern is if PCRE_EXTENDED is set, and an unescaped # outside a
1222 character class is encountered. This indicates a comment that lasts
1223 until after the next line break sequence. In other circumstances, line
1224 break sequences are treated as literal data, except that in
1225 PCRE_EXTENDED mode, both CR and LF are treated as whitespace characters
1226 and are therefore ignored.
1227
1228 The newline option that is set at compile time becomes the default that
1229 is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1230
1231 PCRE_NO_AUTO_CAPTURE
1232
1233 If this option is set, it disables the use of numbered capturing paren-
1234 theses in the pattern. Any opening parenthesis that is not followed by
1235 ? behaves as if it were followed by ?: but named parentheses can still
1236 be used for capturing (and they acquire numbers in the usual way).
1237 There is no equivalent of this option in Perl.
1238
1239 PCRE_UNGREEDY
1240
1241 This option inverts the "greediness" of the quantifiers so that they
1242 are not greedy by default, but become greedy if followed by "?". It is
1243 not compatible with Perl. It can also be set by a (?U) option setting
1244 within the pattern.
1245
1246 PCRE_UTF8
1247
1248 This option causes PCRE to regard both the pattern and the subject as
1249 strings of UTF-8 characters instead of single-byte character strings.
1250 However, it is available only when PCRE is built to include UTF-8 sup-
1251 port. If not, the use of this option provokes an error. Details of how
1252 this option changes the behaviour of PCRE are given in the section on
1253 UTF-8 support in the main pcre page.
1254
1255 PCRE_NO_UTF8_CHECK
1256
1257 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1258 automatically checked. There is a discussion about the validity of
1259 UTF-8 strings in the main pcre page. If an invalid UTF-8 sequence of
1260 bytes is found, pcre_compile() returns an error. If you already know
1261 that your pattern is valid, and you want to skip this check for perfor-
1262 mance reasons, you can set the PCRE_NO_UTF8_CHECK option. When it is
1263 set, the effect of passing an invalid UTF-8 string as a pattern is
1264 undefined. It may cause your program to crash. Note that this option
1265 can also be passed to pcre_exec() and pcre_dfa_exec(), to suppress the
1266 UTF-8 validity checking of subject strings.
1267
1268
1269 COMPILATION ERROR CODES
1270
1271 The following table lists the error codes than may be returned by
1272 pcre_compile2(), along with the error messages that may be returned by
1273 both compiling functions. As PCRE has developed, some error codes have
1274 fallen out of use. To avoid confusion, they have not been re-used.
1275
1276 0 no error
1277 1 \ at end of pattern
1278 2 \c at end of pattern
1279 3 unrecognized character follows \
1280 4 numbers out of order in {} quantifier
1281 5 number too big in {} quantifier
1282 6 missing terminating ] for character class
1283 7 invalid escape sequence in character class
1284 8 range out of order in character class
1285 9 nothing to repeat
1286 10 [this code is not in use]
1287 11 internal error: unexpected repeat
1288 12 unrecognized character after (?
1289 13 POSIX named classes are supported only within a class
1290 14 missing )
1291 15 reference to non-existent subpattern
1292 16 erroffset passed as NULL
1293 17 unknown option bit(s) set
1294 18 missing ) after comment
1295 19 [this code is not in use]
1296 20 regular expression too large
1297 21 failed to get memory
1298 22 unmatched parentheses
1299 23 internal error: code overflow
1300 24 unrecognized character after (?<
1301 25 lookbehind assertion is not fixed length
1302 26 malformed number or name after (?(
1303 27 conditional group contains more than two branches
1304 28 assertion expected after (?(
1305 29 (?R or (?[+-]digits must be followed by )
1306 30 unknown POSIX class name
1307 31 POSIX collating elements are not supported
1308 32 this version of PCRE is not compiled with PCRE_UTF8 support
1309 33 [this code is not in use]
1310 34 character value in \x{...} sequence is too large
1311 35 invalid condition (?(0)
1312 36 \C not allowed in lookbehind assertion
1313 37 PCRE does not support \L, \l, \N, \U, or \u
1314 38 number after (?C is > 255
1315 39 closing ) for (?C expected
1316 40 recursive call could loop indefinitely
1317 41 unrecognized character after (?P
1318 42 syntax error in subpattern name (missing terminator)
1319 43 two named subpatterns have the same name
1320 44 invalid UTF-8 string
1321 45 support for \P, \p, and \X has not been compiled
1322 46 malformed \P or \p sequence
1323 47 unknown property name after \P or \p
1324 48 subpattern name is too long (maximum 32 characters)
1325 49 too many named subpatterns (maximum 10,000)
1326 50 [this code is not in use]
1327 51 octal value is greater than \377 (not in UTF-8 mode)
1328 52 internal error: overran compiling workspace
1329 53 internal error: previously-checked referenced subpattern not
1330 found
1331 54 DEFINE group contains more than one branch
1332 55 repeating a DEFINE group is not allowed
1333 56 inconsistent NEWLINE options"
1334 57 \g is not followed by a braced name or an optionally braced
1335 non-zero number
1336 58 (?+ or (?- or (?(+ or (?(- must be followed by a non-zero number
1337
1338
1339 STUDYING A PATTERN
1340
1341 pcre_extra *pcre_study(const pcre *code, int options
1342 const char **errptr);
1343
1344 If a compiled pattern is going to be used several times, it is worth
1345 spending more time analyzing it in order to speed up the time taken for
1346 matching. The function pcre_study() takes a pointer to a compiled pat-
1347 tern as its first argument. If studying the pattern produces additional
1348 information that will help speed up matching, pcre_study() returns a
1349 pointer to a pcre_extra block, in which the study_data field points to
1350 the results of the study.
1351
1352 The returned value from pcre_study() can be passed directly to
1353 pcre_exec(). However, a pcre_extra block also contains other fields
1354 that can be set by the caller before the block is passed; these are
1355 described below in the section on matching a pattern.
1356
1357 If studying the pattern does not produce any additional information
1358 pcre_study() returns NULL. In that circumstance, if the calling program
1359 wants to pass any of the other fields to pcre_exec(), it must set up
1360 its own pcre_extra block.
1361
1362 The second argument of pcre_study() contains option bits. At present,
1363 no options are defined, and this argument should always be zero.
1364
1365 The third argument for pcre_study() is a pointer for an error message.
1366 If studying succeeds (even if no data is returned), the variable it
1367 points to is set to NULL. Otherwise it is set to point to a textual
1368 error message. This is a static string that is part of the library. You
1369 must not try to free it. You should test the error pointer for NULL
1370 after calling pcre_study(), to be sure that it has run successfully.
1371
1372 This is a typical call to pcre_study():
1373
1374 pcre_extra *pe;
1375 pe = pcre_study(
1376 re, /* result of pcre_compile() */
1377 0, /* no options exist */
1378 &error); /* set to NULL or points to a message */
1379
1380 At present, studying a pattern is useful only for non-anchored patterns
1381 that do not have a single fixed starting character. A bitmap of possi-
1382 ble starting bytes is created.
1383
1384
1385 LOCALE SUPPORT
1386
1387 PCRE handles caseless matching, and determines whether characters are
1388 letters, digits, or whatever, by reference to a set of tables, indexed
1389 by character value. When running in UTF-8 mode, this applies only to
1390 characters with codes less than 128. Higher-valued codes never match
1391 escapes such as \w or \d, but can be tested with \p if PCRE is built
1392 with Unicode character property support. The use of locales with Uni-
1393 code is discouraged. If you are handling characters with codes greater
1394 than 128, you should either use UTF-8 and Unicode, or use locales, but
1395 not try to mix the two.
1396
1397 PCRE contains an internal set of tables that are used when the final
1398 argument of pcre_compile() is NULL. These are sufficient for many
1399 applications. Normally, the internal tables recognize only ASCII char-
1400 acters. However, when PCRE is built, it is possible to cause the inter-
1401 nal tables to be rebuilt in the default "C" locale of the local system,
1402 which may cause them to be different.
1403
1404 The internal tables can always be overridden by tables supplied by the
1405 application that calls PCRE. These may be created in a different locale
1406 from the default. As more and more applications change to using Uni-
1407 code, the need for this locale support is expected to die away.
1408
1409 External tables are built by calling the pcre_maketables() function,
1410 which has no arguments, in the relevant locale. The result can then be
1411 passed to pcre_compile() or pcre_exec() as often as necessary. For
1412 example, to build and use tables that are appropriate for the French
1413 locale (where accented characters with values greater than 128 are
1414 treated as letters), the following code could be used:
1415
1416 setlocale(LC_CTYPE, "fr_FR");
1417 tables = pcre_maketables();
1418 re = pcre_compile(..., tables);
1419
1420 The locale name "fr_FR" is used on Linux and other Unix-like systems;
1421 if you are using Windows, the name for the French locale is "french".
1422
1423 When pcre_maketables() runs, the tables are built in memory that is
1424 obtained via pcre_malloc. It is the caller's responsibility to ensure
1425 that the memory containing the tables remains available for as long as
1426 it is needed.
1427
1428 The pointer that is passed to pcre_compile() is saved with the compiled
1429 pattern, and the same tables are used via this pointer by pcre_study()
1430 and normally also by pcre_exec(). Thus, by default, for any single pat-
1431 tern, compilation, studying and matching all happen in the same locale,
1432 but different patterns can be compiled in different locales.
1433
1434 It is possible to pass a table pointer or NULL (indicating the use of
1435 the internal tables) to pcre_exec(). Although not intended for this
1436 purpose, this facility could be used to match a pattern in a different
1437 locale from the one in which it was compiled. Passing table pointers at
1438 run time is discussed below in the section on matching a pattern.
1439
1440
1441 INFORMATION ABOUT A PATTERN
1442
1443 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1444 int what, void *where);
1445
1446 The pcre_fullinfo() function returns information about a compiled pat-
1447 tern. It replaces the obsolete pcre_info() function, which is neverthe-
1448 less retained for backwards compability (and is documented below).
1449
1450 The first argument for pcre_fullinfo() is a pointer to the compiled
1451 pattern. The second argument is the result of pcre_study(), or NULL if
1452 the pattern was not studied. The third argument specifies which piece
1453 of information is required, and the fourth argument is a pointer to a
1454 variable to receive the data. The yield of the function is zero for
1455 success, or one of the following negative numbers:
1456
1457 PCRE_ERROR_NULL the argument code was NULL
1458 the argument where was NULL
1459 PCRE_ERROR_BADMAGIC the "magic number" was not found
1460 PCRE_ERROR_BADOPTION the value of what was invalid
1461
1462 The "magic number" is placed at the start of each compiled pattern as
1463 an simple check against passing an arbitrary memory pointer. Here is a
1464 typical call of pcre_fullinfo(), to obtain the length of the compiled
1465 pattern:
1466
1467 int rc;
1468 size_t length;
1469 rc = pcre_fullinfo(
1470 re, /* result of pcre_compile() */
1471 pe, /* result of pcre_study(), or NULL */
1472 PCRE_INFO_SIZE, /* what is required */
1473 &length); /* where to put the data */
1474
1475 The possible values for the third argument are defined in pcre.h, and
1476 are as follows:
1477
1478 PCRE_INFO_BACKREFMAX
1479
1480 Return the number of the highest back reference in the pattern. The
1481 fourth argument should point to an int variable. Zero is returned if
1482 there are no back references.
1483
1484 PCRE_INFO_CAPTURECOUNT
1485
1486 Return the number of capturing subpatterns in the pattern. The fourth
1487 argument should point to an int variable.
1488
1489 PCRE_INFO_DEFAULT_TABLES
1490
1491 Return a pointer to the internal default character tables within PCRE.
1492 The fourth argument should point to an unsigned char * variable. This
1493 information call is provided for internal use by the pcre_study() func-
1494 tion. External callers can cause PCRE to use its internal tables by
1495 passing a NULL table pointer.
1496
1497 PCRE_INFO_FIRSTBYTE
1498
1499 Return information about the first byte of any matched string, for a
1500 non-anchored pattern. The fourth argument should point to an int vari-
1501 able. (This option used to be called PCRE_INFO_FIRSTCHAR; the old name
1502 is still recognized for backwards compatibility.)
1503
1504 If there is a fixed first byte, for example, from a pattern such as
1505 (cat|cow|coyote), its value is returned. Otherwise, if either
1506
1507 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
1508 branch starts with "^", or
1509
1510 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
1511 set (if it were set, the pattern would be anchored),
1512
1513 -1 is returned, indicating that the pattern matches only at the start
1514 of a subject string or after any newline within the string. Otherwise
1515 -2 is returned. For anchored patterns, -2 is returned.
1516
1517 PCRE_INFO_FIRSTTABLE
1518
1519 If the pattern was studied, and this resulted in the construction of a
1520 256-bit table indicating a fixed set of bytes for the first byte in any
1521 matching string, a pointer to the table is returned. Otherwise NULL is
1522 returned. The fourth argument should point to an unsigned char * vari-
1523 able.
1524
1525 PCRE_INFO_JCHANGED
1526
1527 Return 1 if the (?J) option setting is used in the pattern, otherwise
1528 0. The fourth argument should point to an int variable. The (?J) inter-
1529 nal option setting changes the local PCRE_DUPNAMES option.
1530
1531 PCRE_INFO_LASTLITERAL
1532
1533 Return the value of the rightmost literal byte that must exist in any
1534 matched string, other than at its start, if such a byte has been
1535 recorded. The fourth argument should point to an int variable. If there
1536 is no such byte, -1 is returned. For anchored patterns, a last literal
1537 byte is recorded only if it follows something of variable length. For
1538 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
1539 /^a\dz\d/ the returned value is -1.
1540
1541 PCRE_INFO_NAMECOUNT
1542 PCRE_INFO_NAMEENTRYSIZE
1543 PCRE_INFO_NAMETABLE
1544
1545 PCRE supports the use of named as well as numbered capturing parenthe-
1546 ses. The names are just an additional way of identifying the parenthe-
1547 ses, which still acquire numbers. Several convenience functions such as
1548 pcre_get_named_substring() are provided for extracting captured sub-
1549 strings by name. It is also possible to extract the data directly, by
1550 first converting the name to a number in order to access the correct
1551 pointers in the output vector (described with pcre_exec() below). To do
1552 the conversion, you need to use the name-to-number map, which is
1553 described by these three values.
1554
1555 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
1556 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
1557 of each entry; both of these return an int value. The entry size
1558 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
1559 a pointer to the first entry of the table (a pointer to char). The
1560 first two bytes of each entry are the number of the capturing parenthe-
1561 sis, most significant byte first. The rest of the entry is the corre-
1562 sponding name, zero terminated. The names are in alphabetical order.
1563 When PCRE_DUPNAMES is set, duplicate names are in order of their paren-
1564 theses numbers. For example, consider the following pattern (assume
1565 PCRE_EXTENDED is set, so white space - including newlines - is
1566 ignored):
1567
1568 (?<date> (?<year>(\d\d)?\d\d) -
1569 (?<month>\d\d) - (?<day>\d\d) )
1570
1571 There are four named subpatterns, so the table has four entries, and
1572 each entry in the table is eight bytes long. The table is as follows,
1573 with non-printing bytes shows in hexadecimal, and undefined bytes shown
1574 as ??:
1575
1576 00 01 d a t e 00 ??
1577 00 05 d a y 00 ?? ??
1578 00 04 m o n t h 00
1579 00 02 y e a r 00 ??
1580
1581 When writing code to extract data from named subpatterns using the
1582 name-to-number map, remember that the length of the entries is likely
1583 to be different for each compiled pattern.
1584
1585 PCRE_INFO_OKPARTIAL
1586
1587 Return 1 if the pattern can be used for partial matching, otherwise 0.
1588 The fourth argument should point to an int variable. The pcrepartial
1589 documentation lists the restrictions that apply to patterns when par-
1590 tial matching is used.
1591
1592 PCRE_INFO_OPTIONS
1593
1594 Return a copy of the options with which the pattern was compiled. The
1595 fourth argument should point to an unsigned long int variable. These
1596 option bits are those specified in the call to pcre_compile(), modified
1597 by any top-level option settings at the start of the pattern itself. In
1598 other words, they are the options that will be in force when matching
1599 starts. For example, if the pattern /(?im)abc(?-i)d/ is compiled with
1600 the PCRE_EXTENDED option, the result is PCRE_CASELESS, PCRE_MULTILINE,
1601 and PCRE_EXTENDED.
1602
1603 A pattern is automatically anchored by PCRE if all of its top-level
1604 alternatives begin with one of the following:
1605
1606 ^ unless PCRE_MULTILINE is set
1607 \A always
1608 \G always
1609 .* if PCRE_DOTALL is set and there are no back
1610 references to the subpattern in which .* appears
1611
1612 For such patterns, the PCRE_ANCHORED bit is set in the options returned
1613 by pcre_fullinfo().
1614
1615 PCRE_INFO_SIZE
1616
1617 Return the size of the compiled pattern, that is, the value that was
1618 passed as the argument to pcre_malloc() when PCRE was getting memory in
1619 which to place the compiled data. The fourth argument should point to a
1620 size_t variable.
1621
1622 PCRE_INFO_STUDYSIZE
1623
1624 Return the size of the data block pointed to by the study_data field in
1625 a pcre_extra block. That is, it is the value that was passed to
1626 pcre_malloc() when PCRE was getting memory into which to place the data
1627 created by pcre_study(). The fourth argument should point to a size_t
1628 variable.
1629
1630
1631 OBSOLETE INFO FUNCTION
1632
1633 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
1634
1635 The pcre_info() function is now obsolete because its interface is too
1636 restrictive to return all the available data about a compiled pattern.
1637 New programs should use pcre_fullinfo() instead. The yield of
1638 pcre_info() is the number of capturing subpatterns, or one of the fol-
1639 lowing negative numbers:
1640
1641 PCRE_ERROR_NULL the argument code was NULL
1642 PCRE_ERROR_BADMAGIC the "magic number" was not found
1643
1644 If the optptr argument is not NULL, a copy of the options with which
1645 the pattern was compiled is placed in the integer it points to (see
1646 PCRE_INFO_OPTIONS above).
1647
1648 If the pattern is not anchored and the firstcharptr argument is not
1649 NULL, it is used to pass back information about the first character of
1650 any matched string (see PCRE_INFO_FIRSTBYTE above).
1651
1652
1653 REFERENCE COUNTS
1654
1655 int pcre_refcount(pcre *code, int adjust);
1656
1657 The pcre_refcount() function is used to maintain a reference count in
1658 the data block that contains a compiled pattern. It is provided for the
1659 benefit of applications that operate in an object-oriented manner,
1660 where different parts of the application may be using the same compiled
1661 pattern, but you want to free the block when they are all done.
1662
1663 When a pattern is compiled, the reference count field is initialized to
1664 zero. It is changed only by calling this function, whose action is to
1665 add the adjust value (which may be positive or negative) to it. The
1666 yield of the function is the new value. However, the value of the count
1667 is constrained to lie between 0 and 65535, inclusive. If the new value
1668 is outside these limits, it is forced to the appropriate limit value.
1669
1670 Except when it is zero, the reference count is not correctly preserved
1671 if a pattern is compiled on one host and then transferred to a host
1672 whose byte-order is different. (This seems a highly unlikely scenario.)
1673
1674
1675 MATCHING A PATTERN: THE TRADITIONAL FUNCTION
1676
1677 int pcre_exec(const pcre *code, const pcre_extra *extra,
1678 const char *subject, int length, int startoffset,
1679 int options, int *ovector, int ovecsize);
1680
1681 The function pcre_exec() is called to match a subject string against a
1682 compiled pattern, which is passed in the code argument. If the pattern
1683 has been studied, the result of the study should be passed in the extra
1684 argument. This function is the main matching facility of the library,
1685 and it operates in a Perl-like manner. For specialist use there is also
1686 an alternative matching function, which is described below in the sec-
1687 tion about the pcre_dfa_exec() function.
1688
1689 In most applications, the pattern will have been compiled (and option-
1690 ally studied) in the same process that calls pcre_exec(). However, it
1691 is possible to save compiled patterns and study data, and then use them
1692 later in different processes, possibly even on different hosts. For a
1693 discussion about this, see the pcreprecompile documentation.
1694
1695 Here is an example of a simple call to pcre_exec():
1696
1697 int rc;
1698 int ovector[30];
1699 rc = pcre_exec(
1700 re, /* result of pcre_compile() */
1701 NULL, /* we didn't study the pattern */
1702 "some string", /* the subject string */
1703 11, /* the length of the subject string */
1704 0, /* start at offset 0 in the subject */
1705 0, /* default options */
1706 ovector, /* vector of integers for substring information */
1707 30); /* number of elements (NOT size in bytes) */
1708
1709 Extra data for pcre_exec()
1710
1711 If the extra argument is not NULL, it must point to a pcre_extra data
1712 block. The pcre_study() function returns such a block (when it doesn't
1713 return NULL), but you can also create one for yourself, and pass addi-
1714 tional information in it. The pcre_extra block contains the following
1715 fields (not necessarily in this order):
1716
1717 unsigned long int flags;
1718 void *study_data;
1719 unsigned long int match_limit;
1720 unsigned long int match_limit_recursion;
1721 void *callout_data;
1722 const unsigned char *tables;
1723
1724 The flags field is a bitmap that specifies which of the other fields
1725 are set. The flag bits are:
1726
1727 PCRE_EXTRA_STUDY_DATA
1728 PCRE_EXTRA_MATCH_LIMIT
1729 PCRE_EXTRA_MATCH_LIMIT_RECURSION
1730 PCRE_EXTRA_CALLOUT_DATA
1731 PCRE_EXTRA_TABLES
1732
1733 Other flag bits should be set to zero. The study_data field is set in
1734 the pcre_extra block that is returned by pcre_study(), together with
1735 the appropriate flag bit. You should not set this yourself, but you may
1736 add to the block by setting the other fields and their corresponding
1737 flag bits.
1738
1739 The match_limit field provides a means of preventing PCRE from using up
1740 a vast amount of resources when running patterns that are not going to
1741 match, but which have a very large number of possibilities in their
1742 search trees. The classic example is the use of nested unlimited
1743 repeats.
1744
1745 Internally, PCRE uses a function called match() which it calls repeat-
1746 edly (sometimes recursively). The limit set by match_limit is imposed
1747 on the number of times this function is called during a match, which
1748 has the effect of limiting the amount of backtracking that can take
1749 place. For patterns that are not anchored, the count restarts from zero
1750 for each position in the subject string.
1751
1752 The default value for the limit can be set when PCRE is built; the
1753 default default is 10 million, which handles all but the most extreme
1754 cases. You can override the default by suppling pcre_exec() with a
1755 pcre_extra block in which match_limit is set, and
1756 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
1757 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
1758
1759 The match_limit_recursion field is similar to match_limit, but instead
1760 of limiting the total number of times that match() is called, it limits
1761 the depth of recursion. The recursion depth is a smaller number than
1762 the total number of calls, because not all calls to match() are recur-
1763 sive. This limit is of use only if it is set smaller than match_limit.
1764
1765 Limiting the recursion depth limits the amount of stack that can be
1766 used, or, when PCRE has been compiled to use memory on the heap instead
1767 of the stack, the amount of heap memory that can be used.
1768
1769 The default value for match_limit_recursion can be set when PCRE is
1770 built; the default default is the same value as the default for
1771 match_limit. You can override the default by suppling pcre_exec() with
1772 a pcre_extra block in which match_limit_recursion is set, and
1773 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
1774 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
1775
1776 The pcre_callout field is used in conjunction with the "callout" fea-
1777 ture, which is described in the pcrecallout documentation.
1778
1779 The tables field is used to pass a character tables pointer to
1780 pcre_exec(); this overrides the value that is stored with the compiled
1781 pattern. A non-NULL value is stored with the compiled pattern only if
1782 custom tables were supplied to pcre_compile() via its tableptr argu-
1783 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
1784 PCRE's internal tables to be used. This facility is helpful when re-
1785 using patterns that have been saved after compiling with an external
1786 set of tables, because the external tables might be at a different
1787 address when pcre_exec() is called. See the pcreprecompile documenta-
1788 tion for a discussion of saving compiled patterns for later use.
1789
1790 Option bits for pcre_exec()
1791
1792 The unused bits of the options argument for pcre_exec() must be zero.
1793 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
1794 PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK and
1795 PCRE_PARTIAL.
1796
1797 PCRE_ANCHORED
1798
1799 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
1800 matching position. If a pattern was compiled with PCRE_ANCHORED, or
1801 turned out to be anchored by virtue of its contents, it cannot be made
1802 unachored at matching time.
1803
1804 PCRE_NEWLINE_CR
1805 PCRE_NEWLINE_LF
1806 PCRE_NEWLINE_CRLF
1807 PCRE_NEWLINE_ANYCRLF
1808 PCRE_NEWLINE_ANY
1809
1810 These options override the newline definition that was chosen or
1811 defaulted when the pattern was compiled. For details, see the descrip-
1812 tion of pcre_compile() above. During matching, the newline choice
1813 affects the behaviour of the dot, circumflex, and dollar metacharac-
1814 ters. It may also alter the way the match position is advanced after a
1815 match failure for an unanchored pattern. When PCRE_NEWLINE_CRLF,
1816 PCRE_NEWLINE_ANYCRLF, or PCRE_NEWLINE_ANY is set, and a match attempt
1817 fails when the current position is at a CRLF sequence, the match posi-
1818 tion is advanced by two characters instead of one, in other words, to
1819 after the CRLF.
1820
1821 PCRE_NOTBOL
1822
1823 This option specifies that first character of the subject string is not
1824 the beginning of a line, so the circumflex metacharacter should not
1825 match before it. Setting this without PCRE_MULTILINE (at compile time)
1826 causes circumflex never to match. This option affects only the behav-
1827 iour of the circumflex metacharacter. It does not affect \A.
1828
1829 PCRE_NOTEOL
1830
1831 This option specifies that the end of the subject string is not the end
1832 of a line, so the dollar metacharacter should not match it nor (except
1833 in multiline mode) a newline immediately before it. Setting this with-
1834 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
1835 option affects only the behaviour of the dollar metacharacter. It does
1836 not affect \Z or \z.
1837
1838 PCRE_NOTEMPTY
1839
1840 An empty string is not considered to be a valid match if this option is
1841 set. If there are alternatives in the pattern, they are tried. If all
1842 the alternatives match the empty string, the entire match fails. For
1843 example, if the pattern
1844
1845 a?b?
1846
1847 is applied to a string not beginning with "a" or "b", it matches the
1848 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
1849 match is not valid, so PCRE searches further into the string for occur-
1850 rences of "a" or "b".
1851
1852 Perl has no direct equivalent of PCRE_NOTEMPTY, but it does make a spe-
1853 cial case of a pattern match of the empty string within its split()
1854 function, and when using the /g modifier. It is possible to emulate
1855 Perl's behaviour after matching a null string by first trying the match
1856 again at the same offset with PCRE_NOTEMPTY and PCRE_ANCHORED, and then
1857 if that fails by advancing the starting offset (see below) and trying
1858 an ordinary match again. There is some code that demonstrates how to do
1859 this in the pcredemo.c sample program.
1860
1861 PCRE_NO_UTF8_CHECK
1862
1863 When PCRE_UTF8 is set at compile time, the validity of the subject as a
1864 UTF-8 string is automatically checked when pcre_exec() is subsequently
1865 called. The value of startoffset is also checked to ensure that it
1866 points to the start of a UTF-8 character. There is a discussion about
1867 the validity of UTF-8 strings in the section on UTF-8 support in the
1868 main pcre page. If an invalid UTF-8 sequence of bytes is found,
1869 pcre_exec() returns the error PCRE_ERROR_BADUTF8. If startoffset con-
1870 tains an invalid value, PCRE_ERROR_BADUTF8_OFFSET is returned.
1871
1872 If you already know that your subject is valid, and you want to skip
1873 these checks for performance reasons, you can set the
1874 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
1875 do this for the second and subsequent calls to pcre_exec() if you are
1876 making repeated calls to find all the matches in a single subject
1877 string. However, you should be sure that the value of startoffset
1878 points to the start of a UTF-8 character. When PCRE_NO_UTF8_CHECK is
1879 set, the effect of passing an invalid UTF-8 string as a subject, or a
1880 value of startoffset that does not point to the start of a UTF-8 char-
1881 acter, is undefined. Your program may crash.
1882
1883 PCRE_PARTIAL
1884
1885 This option turns on the partial matching feature. If the subject
1886 string fails to match the pattern, but at some point during the match-
1887 ing process the end of the subject was reached (that is, the subject
1888 partially matches the pattern and the failure to match occurred only
1889 because there were not enough subject characters), pcre_exec() returns
1890 PCRE_ERROR_PARTIAL instead of PCRE_ERROR_NOMATCH. When PCRE_PARTIAL is
1891 used, there are restrictions on what may appear in the pattern. These
1892 are discussed in the pcrepartial documentation.
1893
1894 The string to be matched by pcre_exec()
1895
1896 The subject string is passed to pcre_exec() as a pointer in subject, a
1897 length in length, and a starting byte offset in startoffset. In UTF-8
1898 mode, the byte offset must point to the start of a UTF-8 character.
1899 Unlike the pattern string, the subject may contain binary zero bytes.
1900 When the starting offset is zero, the search for a match starts at the
1901 beginning of the subject, and this is by far the most common case.
1902
1903 A non-zero starting offset is useful when searching for another match
1904 in the same subject by calling pcre_exec() again after a previous suc-
1905 cess. Setting startoffset differs from just passing over a shortened
1906 string and setting PCRE_NOTBOL in the case of a pattern that begins
1907 with any kind of lookbehind. For example, consider the pattern
1908
1909 \Biss\B
1910
1911 which finds occurrences of "iss" in the middle of words. (\B matches
1912 only if the current position in the subject is not a word boundary.)
1913 When applied to the string "Mississipi" the first call to pcre_exec()
1914 finds the first occurrence. If pcre_exec() is called again with just
1915 the remainder of the subject, namely "issipi", it does not match,
1916 because \B is always false at the start of the subject, which is deemed
1917 to be a word boundary. However, if pcre_exec() is passed the entire
1918 string again, but with startoffset set to 4, it finds the second occur-
1919 rence of "iss" because it is able to look behind the starting point to
1920 discover that it is preceded by a letter.
1921
1922 If a non-zero starting offset is passed when the pattern is anchored,
1923 one attempt to match at the given offset is made. This can only succeed
1924 if the pattern does not require the match to be at the start of the
1925 subject.
1926
1927 How pcre_exec() returns captured substrings
1928
1929 In general, a pattern matches a certain portion of the subject, and in
1930 addition, further substrings from the subject may be picked out by
1931 parts of the pattern. Following the usage in Jeffrey Friedl's book,
1932 this is called "capturing" in what follows, and the phrase "capturing
1933 subpattern" is used for a fragment of a pattern that picks out a sub-
1934 string. PCRE supports several other kinds of parenthesized subpattern
1935 that do not cause substrings to be captured.
1936
1937 Captured substrings are returned to the caller via a vector of integer
1938 offsets whose address is passed in ovector. The number of elements in
1939 the vector is passed in ovecsize, which must be a non-negative number.
1940 Note: this argument is NOT the size of ovector in bytes.
1941
1942 The first two-thirds of the vector is used to pass back captured sub-
1943 strings, each substring using a pair of integers. The remaining third
1944 of the vector is used as workspace by pcre_exec() while matching cap-
1945 turing subpatterns, and is not available for passing back information.
1946 The length passed in ovecsize should always be a multiple of three. If
1947 it is not, it is rounded down.
1948
1949 When a match is successful, information about captured substrings is
1950 returned in pairs of integers, starting at the beginning of ovector,
1951 and continuing up to two-thirds of its length at the most. The first
1952 element of a pair is set to the offset of the first character in a sub-
1953 string, and the second is set to the offset of the first character
1954 after the end of a substring. The first pair, ovector[0] and ovec-
1955 tor[1], identify the portion of the subject string matched by the
1956 entire pattern. The next pair is used for the first capturing subpat-
1957 tern, and so on. The value returned by pcre_exec() is one more than the
1958 highest numbered pair that has been set. For example, if two substrings
1959 have been captured, the returned value is 3. If there are no capturing
1960 subpatterns, the return value from a successful match is 1, indicating
1961 that just the first pair of offsets has been set.
1962
1963 If a capturing subpattern is matched repeatedly, it is the last portion
1964 of the string that it matched that is returned.
1965
1966 If the vector is too small to hold all the captured substring offsets,
1967 it is used as far as possible (up to two-thirds of its length), and the
1968 function returns a value of zero. In particular, if the substring off-
1969 sets are not of interest, pcre_exec() may be called with ovector passed
1970 as NULL and ovecsize as zero. However, if the pattern contains back
1971 references and the ovector is not big enough to remember the related
1972 substrings, PCRE has to get additional memory for use during matching.
1973 Thus it is usually advisable to supply an ovector.
1974
1975 The pcre_info() function can be used to find out how many capturing
1976 subpatterns there are in a compiled pattern. The smallest size for
1977 ovector that will allow for n captured substrings, in addition to the
1978 offsets of the substring matched by the whole pattern, is (n+1)*3.
1979
1980 It is possible for capturing subpattern number n+1 to match some part
1981 of the subject when subpattern n has not been used at all. For example,
1982 if the string "abc" is matched against the pattern (a|(z))(bc) the
1983 return from the function is 4, and subpatterns 1 and 3 are matched, but
1984 2 is not. When this happens, both values in the offset pairs corre-
1985 sponding to unused subpatterns are set to -1.
1986
1987 Offset values that correspond to unused subpatterns at the end of the
1988 expression are also set to -1. For example, if the string "abc" is
1989 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
1990 matched. The return from the function is 2, because the highest used
1991 capturing subpattern number is 1. However, you can refer to the offsets
1992 for the second and third capturing subpatterns if you wish (assuming
1993 the vector is large enough, of course).
1994
1995 Some convenience functions are provided for extracting the captured
1996 substrings as separate strings. These are described below.
1997
1998 Error return values from pcre_exec()
1999
2000 If pcre_exec() fails, it returns a negative number. The following are
2001 defined in the header file:
2002
2003 PCRE_ERROR_NOMATCH (-1)
2004
2005 The subject string did not match the pattern.
2006
2007 PCRE_ERROR_NULL (-2)
2008
2009 Either code or subject was passed as NULL, or ovector was NULL and
2010 ovecsize was not zero.
2011
2012 PCRE_ERROR_BADOPTION (-3)
2013
2014 An unrecognized bit was set in the options argument.
2015
2016 PCRE_ERROR_BADMAGIC (-4)
2017
2018 PCRE stores a 4-byte "magic number" at the start of the compiled code,
2019 to catch the case when it is passed a junk pointer and to detect when a
2020 pattern that was compiled in an environment of one endianness is run in
2021 an environment with the other endianness. This is the error that PCRE
2022 gives when the magic number is not present.
2023
2024 PCRE_ERROR_UNKNOWN_OPCODE (-5)
2025
2026 While running the pattern match, an unknown item was encountered in the
2027 compiled pattern. This error could be caused by a bug in PCRE or by
2028 overwriting of the compiled pattern.
2029
2030 PCRE_ERROR_NOMEMORY (-6)
2031
2032 If a pattern contains back references, but the ovector that is passed
2033 to pcre_exec() is not big enough to remember the referenced substrings,
2034 PCRE gets a block of memory at the start of matching to use for this
2035 purpose. If the call via pcre_malloc() fails, this error is given. The
2036 memory is automatically freed at the end of matching.
2037
2038 PCRE_ERROR_NOSUBSTRING (-7)
2039
2040 This error is used by the pcre_copy_substring(), pcre_get_substring(),
2041 and pcre_get_substring_list() functions (see below). It is never
2042 returned by pcre_exec().
2043
2044 PCRE_ERROR_MATCHLIMIT (-8)
2045
2046 The backtracking limit, as specified by the match_limit field in a
2047 pcre_extra structure (or defaulted) was reached. See the description
2048 above.
2049
2050 PCRE_ERROR_CALLOUT (-9)
2051
2052 This error is never generated by pcre_exec() itself. It is provided for
2053 use by callout functions that want to yield a distinctive error code.
2054 See the pcrecallout documentation for details.
2055
2056 PCRE_ERROR_BADUTF8 (-10)
2057
2058 A string that contains an invalid UTF-8 byte sequence was passed as a
2059 subject.
2060
2061 PCRE_ERROR_BADUTF8_OFFSET (-11)
2062
2063 The UTF-8 byte sequence that was passed as a subject was valid, but the
2064 value of startoffset did not point to the beginning of a UTF-8 charac-
2065 ter.
2066
2067 PCRE_ERROR_PARTIAL (-12)
2068
2069 The subject string did not match, but it did match partially. See the
2070 pcrepartial documentation for details of partial matching.
2071
2072 PCRE_ERROR_BADPARTIAL (-13)
2073
2074 The PCRE_PARTIAL option was used with a compiled pattern containing
2075 items that are not supported for partial matching. See the pcrepartial
2076 documentation for details of partial matching.
2077
2078 PCRE_ERROR_INTERNAL (-14)
2079
2080 An unexpected internal error has occurred. This error could be caused
2081 by a bug in PCRE or by overwriting of the compiled pattern.
2082
2083 PCRE_ERROR_BADCOUNT (-15)
2084
2085 This error is given if the value of the ovecsize argument is negative.
2086
2087 PCRE_ERROR_RECURSIONLIMIT (-21)
2088
2089 The internal recursion limit, as specified by the match_limit_recursion
2090 field in a pcre_extra structure (or defaulted) was reached. See the
2091 description above.
2092
2093 PCRE_ERROR_BADNEWLINE (-23)
2094
2095 An invalid combination of PCRE_NEWLINE_xxx options was given.
2096
2097 Error numbers -16 to -20 and -22 are not used by pcre_exec().
2098
2099
2100 EXTRACTING CAPTURED SUBSTRINGS BY NUMBER
2101
2102 int pcre_copy_substring(const char *subject, int *ovector,
2103 int stringcount, int stringnumber, char *buffer,
2104 int buffersize);
2105
2106 int pcre_get_substring(const char *subject, int *ovector,
2107 int stringcount, int stringnumber,
2108 const char **stringptr);
2109
2110 int pcre_get_substring_list(const char *subject,
2111 int *ovector, int stringcount, const char ***listptr);
2112
2113 Captured substrings can be accessed directly by using the offsets
2114 returned by pcre_exec() in ovector. For convenience, the functions
2115 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
2116 string_list() are provided for extracting captured substrings as new,
2117 separate, zero-terminated strings. These functions identify substrings
2118 by number. The next section describes functions for extracting named
2119 substrings.
2120
2121 A substring that contains a binary zero is correctly extracted and has
2122 a further zero added on the end, but the result is not, of course, a C
2123 string. However, you can process such a string by referring to the
2124 length that is returned by pcre_copy_substring() and pcre_get_sub-
2125 string(). Unfortunately, the interface to pcre_get_substring_list() is
2126 not adequate for handling strings containing binary zeros, because the
2127 end of the final string is not independently indicated.
2128
2129 The first three arguments are the same for all three of these func-
2130 tions: subject is the subject string that has just been successfully
2131 matched, ovector is a pointer to the vector of integer offsets that was
2132 passed to pcre_exec(), and stringcount is the number of substrings that
2133 were captured by the match, including the substring that matched the
2134 entire regular expression. This is the value returned by pcre_exec() if
2135 it is greater than zero. If pcre_exec() returned zero, indicating that
2136 it ran out of space in ovector, the value passed as stringcount should
2137 be the number of elements in the vector divided by three.
2138
2139 The functions pcre_copy_substring() and pcre_get_substring() extract a
2140 single substring, whose number is given as stringnumber. A value of
2141 zero extracts the substring that matched the entire pattern, whereas
2142 higher values extract the captured substrings. For pcre_copy_sub-
2143 string(), the string is placed in buffer, whose length is given by
2144 buffersize, while for pcre_get_substring() a new block of memory is
2145 obtained via pcre_malloc, and its address is returned via stringptr.
2146 The yield of the function is the length of the string, not including
2147 the terminating zero, or one of these error codes:
2148
2149 PCRE_ERROR_NOMEMORY (-6)
2150
2151 The buffer was too small for pcre_copy_substring(), or the attempt to
2152 get memory failed for pcre_get_substring().
2153
2154 PCRE_ERROR_NOSUBSTRING (-7)
2155
2156 There is no substring whose number is stringnumber.
2157
2158 The pcre_get_substring_list() function extracts all available sub-
2159 strings and builds a list of pointers to them. All this is done in a
2160 single block of memory that is obtained via pcre_malloc. The address of
2161 the memory block is returned via listptr, which is also the start of
2162 the list of string pointers. The end of the list is marked by a NULL
2163 pointer. The yield of the function is zero if all went well, or the
2164 error code
2165
2166 PCRE_ERROR_NOMEMORY (-6)
2167
2168 if the attempt to get the memory block failed.
2169
2170 When any of these functions encounter a substring that is unset, which
2171 can happen when capturing subpattern number n+1 matches some part of
2172 the subject, but subpattern n has not been used at all, they return an
2173 empty string. This can be distinguished from a genuine zero-length sub-
2174 string by inspecting the appropriate offset in ovector, which is nega-
2175 tive for unset substrings.
2176
2177 The two convenience functions pcre_free_substring() and pcre_free_sub-
2178 string_list() can be used to free the memory returned by a previous
2179 call of pcre_get_substring() or pcre_get_substring_list(), respec-
2180 tively. They do nothing more than call the function pointed to by
2181 pcre_free, which of course could be called directly from a C program.
2182 However, PCRE is used in some situations where it is linked via a spe-
2183 cial interface to another programming language that cannot use
2184 pcre_free directly; it is for these cases that the functions are pro-
2185 vided.
2186
2187
2188 EXTRACTING CAPTURED SUBSTRINGS BY NAME
2189
2190 int pcre_get_stringnumber(const pcre *code,
2191 const char *name);
2192
2193 int pcre_copy_named_substring(const pcre *code,
2194 const char *subject, int *ovector,
2195 int stringcount, const char *stringname,
2196 char *buffer, int buffersize);
2197
2198 int pcre_get_named_substring(const pcre *code,
2199 const char *subject, int *ovector,
2200 int stringcount, const char *stringname,
2201 const char **stringptr);
2202
2203 To extract a substring by name, you first have to find associated num-
2204 ber. For example, for this pattern
2205
2206 (a+)b(?<xxx>\d+)...
2207
2208 the number of the subpattern called "xxx" is 2. If the name is known to
2209 be unique (PCRE_DUPNAMES was not set), you can find the number from the
2210 name by calling pcre_get_stringnumber(). The first argument is the com-
2211 piled pattern, and the second is the name. The yield of the function is
2212 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
2213 subpattern of that name.
2214
2215 Given the number, you can extract the substring directly, or use one of
2216 the functions described in the previous section. For convenience, there
2217 are also two functions that do the whole job.
2218
2219 Most of the arguments of pcre_copy_named_substring() and
2220 pcre_get_named_substring() are the same as those for the similarly
2221 named functions that extract by number. As these are described in the
2222 previous section, they are not re-described here. There are just two
2223 differences:
2224
2225 First, instead of a substring number, a substring name is given. Sec-
2226 ond, there is an extra argument, given at the start, which is a pointer
2227 to the compiled pattern. This is needed in order to gain access to the
2228 name-to-number translation table.
2229
2230 These functions call pcre_get_stringnumber(), and if it succeeds, they
2231 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
2232 ate. NOTE: If PCRE_DUPNAMES is set and there are duplicate names, the
2233 behaviour may not be what you want (see the next section).
2234
2235
2236 DUPLICATE SUBPATTERN NAMES
2237
2238 int pcre_get_stringtable_entries(const pcre *code,
2239 const char *name, char **first, char **last);
2240
2241 When a pattern is compiled with the PCRE_DUPNAMES option, names for
2242 subpatterns are not required to be unique. Normally, patterns with
2243 duplicate names are such that in any one match, only one of the named
2244 subpatterns participates. An example is shown in the pcrepattern docu-
2245 mentation.
2246
2247 When duplicates are present, pcre_copy_named_substring() and
2248 pcre_get_named_substring() return the first substring corresponding to
2249 the given name that is set. If none are set, PCRE_ERROR_NOSUBSTRING
2250 (-7) is returned; no data is returned. The pcre_get_stringnumber()
2251 function returns one of the numbers that are associated with the name,
2252 but it is not defined which it is.
2253
2254 If you want to get full details of all captured substrings for a given
2255 name, you must use the pcre_get_stringtable_entries() function. The
2256 first argument is the compiled pattern, and the second is the name. The
2257 third and fourth are pointers to variables which are updated by the
2258 function. After it has run, they point to the first and last entries in
2259 the name-to-number table for the given name. The function itself
2260 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
2261 there are none. The format of the table is described above in the sec-
2262 tion entitled Information about a pattern. Given all the relevant
2263 entries for the name, you can extract each of their numbers, and hence
2264 the captured data, if any.
2265
2266
2267 FINDING ALL POSSIBLE MATCHES
2268
2269 The traditional matching function uses a similar algorithm to Perl,
2270 which stops when it finds the first match, starting at a given point in
2271 the subject. If you want to find all possible matches, or the longest
2272 possible match, consider using the alternative matching function (see
2273 below) instead. If you cannot use the alternative function, but still
2274 need to find all possible matches, you can kludge it up by making use
2275 of the callout facility, which is described in the pcrecallout documen-
2276 tation.
2277
2278 What you have to do is to insert a callout right at the end of the pat-
2279 tern. When your callout function is called, extract and save the cur-
2280 rent matched substring. Then return 1, which forces pcre_exec() to
2281 backtrack and try other alternatives. Ultimately, when it runs out of
2282 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
2283
2284
2285 MATCHING A PATTERN: THE ALTERNATIVE FUNCTION
2286
2287 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
2288 const char *subject, int length, int startoffset,
2289 int options, int *ovector, int ovecsize,
2290 int *workspace, int wscount);
2291
2292 The function pcre_dfa_exec() is called to match a subject string
2293 against a compiled pattern, using a matching algorithm that scans the
2294 subject string just once, and does not backtrack. This has different
2295 characteristics to the normal algorithm, and is not compatible with
2296 Perl. Some of the features of PCRE patterns are not supported. Never-
2297 theless, there are times when this kind of matching can be useful. For
2298 a discussion of the two matching algorithms, see the pcrematching docu-
2299 mentation.
2300
2301 The arguments for the pcre_dfa_exec() function are the same as for
2302 pcre_exec(), plus two extras. The ovector argument is used in a differ-
2303 ent way, and this is described below. The other common arguments are
2304 used in the same way as for pcre_exec(), so their description is not
2305 repeated here.
2306
2307 The two additional arguments provide workspace for the function. The
2308 workspace vector should contain at least 20 elements. It is used for
2309 keeping track of multiple paths through the pattern tree. More
2310 workspace will be needed for patterns and subjects where there are a
2311 lot of potential matches.
2312
2313 Here is an example of a simple call to pcre_dfa_exec():
2314
2315 int rc;
2316 int ovector[10];
2317 int wspace[20];
2318 rc = pcre_dfa_exec(
2319 re, /* result of pcre_compile() */
2320 NULL, /* we didn't study the pattern */
2321 "some string", /* the subject string */
2322 11, /* the length of the subject string */
2323 0, /* start at offset 0 in the subject */
2324 0, /* default options */
2325 ovector, /* vector of integers for substring information */
2326 10, /* number of elements (NOT size in bytes) */
2327 wspace, /* working space vector */
2328 20); /* number of elements (NOT size in bytes) */
2329
2330 Option bits for pcre_dfa_exec()
2331
2332 The unused bits of the options argument for pcre_dfa_exec() must be
2333 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
2334 LINE_xxx, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK,
2335 PCRE_PARTIAL, PCRE_DFA_SHORTEST, and PCRE_DFA_RESTART. All but the last
2336 three of these are the same as for pcre_exec(), so their description is
2337 not repeated here.
2338
2339 PCRE_PARTIAL
2340
2341 This has the same general effect as it does for pcre_exec(), but the
2342 details are slightly different. When PCRE_PARTIAL is set for
2343 pcre_dfa_exec(), the return code PCRE_ERROR_NOMATCH is converted into
2344 PCRE_ERROR_PARTIAL if the end of the subject is reached, there have
2345 been no complete matches, but there is still at least one matching pos-
2346 sibility. The portion of the string that provided the partial match is
2347 set as the first matching string.
2348
2349 PCRE_DFA_SHORTEST
2350
2351 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
2352 stop as soon as it has found one match. Because of the way the alterna-
2353 tive algorithm works, this is necessarily the shortest possible match
2354 at the first possible matching point in the subject string.
2355
2356 PCRE_DFA_RESTART
2357
2358 When pcre_dfa_exec() is called with the PCRE_PARTIAL option, and
2359 returns a partial match, it is possible to call it again, with addi-
2360 tional subject characters, and have it continue with the same match.
2361 The PCRE_DFA_RESTART option requests this action; when it is set, the
2362 workspace and wscount options must reference the same vector as before
2363 because data about the match so far is left in them after a partial
2364 match. There is more discussion of this facility in the pcrepartial
2365 documentation.
2366
2367 Successful returns from pcre_dfa_exec()
2368
2369 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
2370 string in the subject. Note, however, that all the matches from one run
2371 of the function start at the same point in the subject. The shorter
2372 matches are all initial substrings of the longer matches. For example,
2373 if the pattern
2374
2375 <.*>
2376
2377 is matched against the string
2378
2379 This is <something> <something else> <something further> no more
2380
2381 the three matched strings are
2382
2383 <something>
2384 <something> <something else>
2385 <something> <something else> <something further>
2386
2387 On success, the yield of the function is a number greater than zero,
2388 which is the number of matched substrings. The substrings themselves
2389 are returned in ovector. Each string uses two elements; the first is
2390 the offset to the start, and the second is the offset to the end. In
2391 fact, all the strings have the same start offset. (Space could have
2392 been saved by giving this only once, but it was decided to retain some
2393 compatibility with the way pcre_exec() returns data, even though the
2394 meaning of the strings is different.)
2395
2396 The strings are returned in reverse order of length; that is, the long-
2397 est matching string is given first. If there were too many matches to
2398 fit into ovector, the yield of the function is zero, and the vector is
2399 filled with the longest matches.
2400
2401 Error returns from pcre_dfa_exec()
2402
2403 The pcre_dfa_exec() function returns a negative number when it fails.
2404 Many of the errors are the same as for pcre_exec(), and these are
2405 described above. There are in addition the following errors that are
2406 specific to pcre_dfa_exec():
2407
2408 PCRE_ERROR_DFA_UITEM (-16)
2409
2410 This return is given if pcre_dfa_exec() encounters an item in the pat-
2411 tern that it does not support, for instance, the use of \C or a back
2412 reference.
2413
2414 PCRE_ERROR_DFA_UCOND (-17)
2415
2416 This return is given if pcre_dfa_exec() encounters a condition item
2417 that uses a back reference for the condition, or a test for recursion
2418 in a specific group. These are not supported.
2419
2420 PCRE_ERROR_DFA_UMLIMIT (-18)
2421
2422 This return is given if pcre_dfa_exec() is called with an extra block
2423 that contains a setting of the match_limit field. This is not supported
2424 (it is meaningless).
2425
2426 PCRE_ERROR_DFA_WSSIZE (-19)
2427
2428 This return is given if pcre_dfa_exec() runs out of space in the
2429 workspace vector.
2430
2431 PCRE_ERROR_DFA_RECURSE (-20)
2432
2433 When a recursive subpattern is processed, the matching function calls
2434 itself recursively, using private vectors for ovector and workspace.
2435 This error is given if the output vector is not large enough. This
2436 should be extremely rare, as a vector of size 1000 is used.
2437
2438
2439 SEE ALSO
2440
2441 pcrebuild(3), pcrecallout(3), pcrecpp(3)(3), pcrematching(3), pcrepar-
2442 tial(3), pcreposix(3), pcreprecompile(3), pcresample(3), pcrestack(3).
2443
2444
2445 AUTHOR
2446
2447 Philip Hazel
2448 University Computing Service
2449 Cambridge CB2 3QH, England.
2450
2451
2452 REVISION
2453
2454 Last updated: 09 August 2007
2455 Copyright (c) 1997-2007 University of Cambridge.
2456 ------------------------------------------------------------------------------
2457
2458
2459 PCRECALLOUT(3) PCRECALLOUT(3)
2460
2461
2462 NAME
2463 PCRE - Perl-compatible regular expressions
2464
2465
2466 PCRE CALLOUTS
2467
2468 int (*pcre_callout)(pcre_callout_block *);
2469
2470 PCRE provides a feature called "callout", which is a means of temporar-
2471 ily passing control to the caller of PCRE in the middle of pattern
2472 matching. The caller of PCRE provides an external function by putting
2473 its entry point in the global variable pcre_callout. By default, this
2474 variable contains NULL, which disables all calling out.
2475
2476 Within a regular expression, (?C) indicates the points at which the
2477 external function is to be called. Different callout points can be
2478 identified by putting a number less than 256 after the letter C. The
2479 default value is zero. For example, this pattern has two callout
2480 points:
2481
2482 (?C1)abc(?C2)def
2483
2484 If the PCRE_AUTO_CALLOUT option bit is set when pcre_compile() is
2485 called, PCRE automatically inserts callouts, all with number 255,
2486 before each item in the pattern. For example, if PCRE_AUTO_CALLOUT is
2487 used with the pattern
2488
2489 A(\d{2}|--)
2490
2491 it is processed as if it were
2492
2493 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
2494
2495 Notice that there is a callout before and after each parenthesis and
2496 alternation bar. Automatic callouts can be used for tracking the
2497 progress of pattern matching. The pcretest command has an option that
2498 sets automatic callouts; when it is used, the output indicates how the
2499 pattern is matched. This is useful information when you are trying to
2500 optimize the performance of a particular pattern.
2501
2502
2503 MISSING CALLOUTS
2504
2505 You should be aware that, because of optimizations in the way PCRE
2506 matches patterns, callouts sometimes do not happen. For example, if the
2507 pattern is
2508
2509 ab(?C4)cd
2510
2511 PCRE knows that any matching string must contain the letter "d". If the
2512 subject string is "abyz", the lack of "d" means that matching doesn't
2513 ever start, and the callout is never reached. However, with "abyd",
2514 though the result is still no match, the callout is obeyed.
2515
2516
2517 THE CALLOUT INTERFACE
2518
2519 During matching, when PCRE reaches a callout point, the external func-
2520 tion defined by pcre_callout is called (if it is set). This applies to
2521 both the pcre_exec() and the pcre_dfa_exec() matching functions. The
2522 only argument to the callout function is a pointer to a pcre_callout
2523 block. This structure contains the following fields:
2524
2525 int version;
2526 int callout_number;
2527 int *offset_vector;
2528 const char *subject;
2529 int subject_length;
2530 int start_match;
2531 int current_position;
2532 int capture_top;
2533 int capture_last;
2534 void *callout_data;
2535 int pattern_position;
2536 int next_item_length;
2537
2538 The version field is an integer containing the version number of the
2539 block format. The initial version was 0; the current version is 1. The
2540 version number will change again in future if additional fields are
2541 added, but the intention is never to remove any of the existing fields.
2542
2543 The callout_number field contains the number of the callout, as com-
2544 piled into the pattern (that is, the number after ?C for manual call-
2545 outs, and 255 for automatically generated callouts).
2546
2547 The offset_vector field is a pointer to the vector of offsets that was
2548 passed by the caller to pcre_exec() or pcre_dfa_exec(). When
2549 pcre_exec() is used, the contents can be inspected in order to extract
2550 substrings that have been matched so far, in the same way as for
2551 extracting substrings after a match has completed. For pcre_dfa_exec()
2552 this field is not useful.
2553
2554 The subject and subject_length fields contain copies of the values that
2555 were passed to pcre_exec().
2556
2557 The start_match field normally contains the offset within the subject
2558 at which the current match attempt started. However, if the escape
2559 sequence \K has been encountered, this value is changed to reflect the
2560 modified starting point. If the pattern is not anchored, the callout
2561 function may be called several times from the same point in the pattern
2562 for different starting points in the subject.
2563
2564 The current_position field contains the offset within the subject of
2565 the current match pointer.
2566
2567 When the pcre_exec() function is used, the capture_top field contains
2568 one more than the number of the highest numbered captured substring so
2569 far. If no substrings have been captured, the value of capture_top is
2570 one. This is always the case when pcre_dfa_exec() is used, because it
2571 does not support captured substrings.
2572
2573 The capture_last field contains the number of the most recently cap-
2574 tured substring. If no substrings have been captured, its value is -1.
2575 This is always the case when pcre_dfa_exec() is used.
2576
2577 The callout_data field contains a value that is passed to pcre_exec()
2578 or pcre_dfa_exec() specifically so that it can be passed back in call-
2579 outs. It is passed in the pcre_callout field of the pcre_extra data
2580 structure. If no such data was passed, the value of callout_data in a
2581 pcre_callout block is NULL. There is a description of the pcre_extra
2582 structure in the pcreapi documentation.
2583
2584 The pattern_position field is present from version 1 of the pcre_call-
2585 out structure. It contains the offset to the next item to be matched in
2586 the pattern string.
2587
2588 The next_item_length field is present from version 1 of the pcre_call-
2589 out structure. It contains the length of the next item to be matched in
2590 the pattern string. When the callout immediately precedes an alterna-
2591 tion bar, a closing parenthesis, or the end of the pattern, the length
2592 is zero. When the callout precedes an opening parenthesis, the length
2593 is that of the entire subpattern.
2594
2595 The pattern_position and next_item_length fields are intended to help
2596 in distinguishing between different automatic callouts, which all have
2597 the same callout number. However, they are set for all callouts.
2598
2599
2600 RETURN VALUES
2601
2602 The external callout function returns an integer to PCRE. If the value
2603 is zero, matching proceeds as normal. If the value is greater than
2604 zero, matching fails at the current point, but the testing of other
2605 matching possibilities goes ahead, just as if a lookahead assertion had
2606 failed. If the value is less than zero, the match is abandoned, and
2607 pcre_exec() (or pcre_dfa_exec()) returns the negative value.
2608
2609 Negative values should normally be chosen from the set of
2610 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
2611 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
2612 reserved for use by callout functions; it will never be used by PCRE
2613 itself.
2614
2615
2616 AUTHOR
2617
2618 Philip Hazel
2619 University Computing Service
2620 Cambridge CB2 3QH, England.
2621
2622
2623 REVISION
2624
2625 Last updated: 29 May 2007
2626 Copyright (c) 1997-2007 University of Cambridge.
2627 ------------------------------------------------------------------------------
2628
2629
2630 PCRECOMPAT(3) PCRECOMPAT(3)
2631
2632
2633 NAME
2634 PCRE - Perl-compatible regular expressions
2635
2636
2637 DIFFERENCES BETWEEN PCRE AND PERL
2638
2639 This document describes the differences in the ways that PCRE and Perl
2640 handle regular expressions. The differences described here are mainly
2641 with respect to Perl 5.8, though PCRE versions 7.0 and later contain
2642 some features that are expected to be in the forthcoming Perl 5.10.
2643
2644 1. PCRE has only a subset of Perl's UTF-8 and Unicode support. Details
2645 of what it does have are given in the section on UTF-8 support in the
2646 main pcre page.
2647
2648 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl
2649 permits them, but they do not mean what you might think. For example,
2650 (?!a){3} does not assert that the next three characters are not "a". It
2651 just asserts that the next character is not "a" three times.
2652
2653 3. Capturing subpatterns that occur inside negative lookahead asser-
2654 tions are counted, but their entries in the offsets vector are never
2655 set. Perl sets its numerical variables from any such patterns that are
2656 matched before the assertion fails to match something (thereby succeed-
2657 ing), but only if the negative lookahead assertion contains just one
2658 branch.
2659
2660 4. Though binary zero characters are supported in the subject string,
2661 they are not allowed in a pattern string because it is passed as a nor-
2662 mal C string, terminated by zero. The escape sequence \0 can be used in
2663 the pattern to represent a binary zero.
2664
2665 5. The following Perl escape sequences are not supported: \l, \u, \L,
2666 \U, and \N. In fact these are implemented by Perl's general string-han-
2667 dling and are not part of its pattern matching engine. If any of these
2668 are encountered by PCRE, an error is generated.
2669
2670 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
2671 is built with Unicode character property support. The properties that
2672 can be tested with \p and \P are limited to the general category prop-
2673 erties such as Lu and Nd, script names such as Greek or Han, and the
2674 derived properties Any and L&.
2675
2676 7. PCRE does support the \Q...\E escape for quoting substrings. Charac-
2677 ters in between are treated as literals. This is slightly different
2678 from Perl in that $ and @ are also handled as literals inside the
2679 quotes. In Perl, they cause variable interpolation (but of course PCRE
2680 does not have variables). Note the following examples:
2681
2682 Pattern PCRE matches Perl matches
2683
2684 \Qabc$xyz\E abc$xyz abc followed by the
2685 contents of $xyz
2686 \Qabc\$xyz\E abc\$xyz abc\$xyz
2687 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2688
2689 The \Q...\E sequence is recognized both inside and outside character
2690 classes.
2691
2692 8. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
2693 constructions. However, there is support for recursive patterns. This
2694 is not available in Perl 5.8, but will be in Perl 5.10. Also, the PCRE
2695 "callout" feature allows an external function to be called during pat-
2696 tern matching. See the pcrecallout documentation for details.
2697
2698 9. Subpatterns that are called recursively or as "subroutines" are
2699 always treated as atomic groups in PCRE. This is like Python, but
2700 unlike Perl.
2701
2702 10. There are some differences that are concerned with the settings of
2703 captured strings when part of a pattern is repeated. For example,
2704 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
2705 unset, but in PCRE it is set to "b".
2706
2707 11. PCRE does support Perl 5.10's backtracking verbs (*ACCEPT),
2708 (*FAIL), (*F), (*COMMIT), (*PRUNE), (*SKIP), and (*THEN), but only in
2709 the forms without an argument. PCRE does not support (*MARK). If
2710 (*ACCEPT) is within capturing parentheses, PCRE does not set that cap-
2711 ture group; this is different to Perl.
2712
2713 12. PCRE provides some extensions to the Perl regular expression facil-
2714 ities. Perl 5.10 will include new features that are not in earlier
2715 versions, some of which (such as named parentheses) have been in PCRE
2716 for some time. This list is with respect to Perl 5.10:
2717
2718 (a) Although lookbehind assertions must match fixed length strings,
2719 each alternative branch of a lookbehind assertion can match a different
2720 length of string. Perl requires them all to have the same length.
2721
2722 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
2723 meta-character matches only at the very end of the string.
2724
2725 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
2726 cial meaning is faulted. Otherwise, like Perl, the backslash is quietly
2727 ignored. (Perl can be made to issue a warning.)
2728
2729 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
2730 fiers is inverted, that is, by default they are not greedy, but if fol-
2731 lowed by a question mark they are.
2732
2733 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
2734 tried only at the first matching position in the subject string.
2735
2736 (f) The PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, and PCRE_NO_AUTO_CAP-
2737 TURE options for pcre_exec() have no Perl equivalents.
2738
2739 (g) The callout facility is PCRE-specific.
2740
2741 (h) The partial matching facility is PCRE-specific.
2742
2743 (i) Patterns compiled by PCRE can be saved and re-used at a later time,
2744 even on different hosts that have the other endianness.
2745
2746 (j) The alternative matching function (pcre_dfa_exec()) matches in a
2747 different way and is not Perl-compatible.
2748
2749
2750 AUTHOR
2751
2752 Philip Hazel
2753 University Computing Service
2754 Cambridge CB2 3QH, England.
2755
2756
2757 REVISION
2758
2759 Last updated: 08 August 2007
2760 Copyright (c) 1997-2007 University of Cambridge.
2761 ------------------------------------------------------------------------------
2762
2763
2764 PCREPATTERN(3) PCREPATTERN(3)
2765
2766
2767 NAME
2768 PCRE - Perl-compatible regular expressions
2769
2770
2771 PCRE REGULAR EXPRESSION DETAILS
2772
2773 The syntax and semantics of the regular expressions that are supported
2774 by PCRE are described in detail below. There is a quick-reference syn-
2775 tax summary in the pcresyntax page. Perl's regular expressions are
2776 described in its own documentation, and regular expressions in general
2777 are covered in a number of books, some of which have copious examples.
2778 Jeffrey Friedl's "Mastering Regular Expressions", published by
2779 O'Reilly, covers regular expressions in great detail. This description
2780 of PCRE's regular expressions is intended as reference material.
2781
2782 The original operation of PCRE was on strings of one-byte characters.
2783 However, there is now also support for UTF-8 character strings. To use
2784 this, you must build PCRE to include UTF-8 support, and then call
2785 pcre_compile() with the PCRE_UTF8 option. How this affects pattern
2786 matching is mentioned in several places below. There is also a summary
2787 of UTF-8 features in the section on UTF-8 support in the main pcre
2788 page.
2789
2790 The remainder of this document discusses the patterns that are sup-
2791 ported by PCRE when its main matching function, pcre_exec(), is used.
2792 From release 6.0, PCRE offers a second matching function,
2793 pcre_dfa_exec(), which matches using a different algorithm that is not
2794 Perl-compatible. Some of the features discussed below are not available
2795 when pcre_dfa_exec() is used. The advantages and disadvantages of the
2796 alternative function, and how it differs from the normal function, are
2797 discussed in the pcrematching page.
2798
2799
2800 CHARACTERS AND METACHARACTERS
2801
2802 A regular expression is a pattern that is matched against a subject
2803 string from left to right. Most characters stand for themselves in a
2804 pattern, and match the corresponding characters in the subject. As a
2805 trivial example, the pattern
2806
2807 The quick brown fox
2808
2809 matches a portion of a subject string that is identical to itself. When
2810 caseless matching is specified (the PCRE_CASELESS option), letters are
2811 matched independently of case. In UTF-8 mode, PCRE always understands
2812 the concept of case for characters whose values are less than 128, so
2813 caseless matching is always possible. For characters with higher val-
2814 ues, the concept of case is supported if PCRE is compiled with Unicode
2815 property support, but not otherwise. If you want to use caseless
2816 matching for characters 128 and above, you must ensure that PCRE is
2817 compiled with Unicode property support as well as with UTF-8 support.
2818
2819 The power of regular expressions comes from the ability to include
2820 alternatives and repetitions in the pattern. These are encoded in the
2821 pattern by the use of metacharacters, which do not stand for themselves
2822 but instead are interpreted in some special way.
2823
2824 There are two different sets of metacharacters: those that are recog-
2825 nized anywhere in the pattern except within square brackets, and those
2826 that are recognized within square brackets. Outside square brackets,
2827 the metacharacters are as follows:
2828
2829 \ general escape character with several uses
2830 ^ assert start of string (or line, in multiline mode)
2831 $ assert end of string (or line, in multiline mode)
2832 . match any character except newline (by default)
2833 [ start character class definition
2834 | start of alternative branch
2835 ( start subpattern
2836 ) end subpattern
2837 ? extends the meaning of (
2838 also 0 or 1 quantifier
2839 also quantifier minimizer
2840 * 0 or more quantifier
2841 + 1 or more quantifier
2842 also "possessive quantifier"
2843 { start min/max quantifier
2844
2845 Part of a pattern that is in square brackets is called a "character
2846 class". In a character class the only metacharacters are:
2847
2848 \ general escape character
2849 ^ negate the class, but only if the first character
2850 - indicates character range
2851 [ POSIX character class (only if followed by POSIX
2852 syntax)
2853 ] terminates the character class
2854
2855 The following sections describe the use of each of the metacharacters.
2856
2857
2858 BACKSLASH
2859
2860 The backslash character has several uses. Firstly, if it is followed by
2861 a non-alphanumeric character, it takes away any special meaning that
2862 character may have. This use of backslash as an escape character
2863 applies both inside and outside character classes.
2864
2865 For example, if you want to match a * character, you write \* in the
2866 pattern. This escaping action applies whether or not the following
2867 character would otherwise be interpreted as a metacharacter, so it is
2868 always safe to precede a non-alphanumeric with backslash to specify
2869 that it stands for itself. In particular, if you want to match a back-
2870 slash, you write \\.
2871
2872 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
2873 the pattern (other than in a character class) and characters between a
2874 # outside a character class and the next newline are ignored. An escap-
2875 ing backslash can be used to include a whitespace or # character as
2876 part of the pattern.
2877
2878 If you want to remove the special meaning from a sequence of charac-
2879 ters, you can do so by putting them between \Q and \E. This is differ-
2880 ent from Perl in that $ and @ are handled as literals in \Q...\E
2881 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
2882 tion. Note the following examples:
2883
2884 Pattern PCRE matches Perl matches
2885
2886 \Qabc$xyz\E abc$xyz abc followed by the
2887 contents of $xyz
2888 \Qabc\$xyz\E abc\$xyz abc\$xyz
2889 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2890
2891 The \Q...\E sequence is recognized both inside and outside character
2892 classes.
2893
2894 Non-printing characters
2895
2896 A second use of backslash provides a way of encoding non-printing char-
2897 acters in patterns in a visible manner. There is no restriction on the
2898 appearance of non-printing characters, apart from the binary zero that
2899 terminates a pattern, but when a pattern is being prepared by text
2900 editing, it is usually easier to use one of the following escape
2901 sequences than the binary character it represents:
2902
2903 \a alarm, that is, the BEL character (hex 07)
2904 \cx "control-x", where x is any character
2905 \e escape (hex 1B)
2906 \f formfeed (hex 0C)
2907 \n newline (hex 0A)
2908 \r carriage return (hex 0D)
2909 \t tab (hex 09)
2910 \ddd character with octal code ddd, or backreference
2911 \xhh character with hex code hh
2912 \x{hhh..} character with hex code hhh..
2913
2914 The precise effect of \cx is as follows: if x is a lower case letter,
2915 it is converted to upper case. Then bit 6 of the character (hex 40) is
2916 inverted. Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c;
2917 becomes hex 7B.
2918
2919 After \x, from zero to two hexadecimal digits are read (letters can be
2920 in upper or lower case). Any number of hexadecimal digits may appear
2921 between \x{ and }, but the value of the character code must be less
2922 than 256 in non-UTF-8 mode, and less than 2**31 in UTF-8 mode. That is,
2923 the maximum value in hexadecimal is 7FFFFFFF. Note that this is bigger
2924 than the largest Unicode code point, which is 10FFFF.
2925
2926 If characters other than hexadecimal digits appear between \x{ and },
2927 or if there is no terminating }, this form of escape is not recognized.
2928 Instead, the initial \x will be interpreted as a basic hexadecimal
2929 escape, with no following digits, giving a character whose value is
2930 zero.
2931
2932 Characters whose value is less than 256 can be defined by either of the
2933 two syntaxes for \x. There is no difference in the way they are han-
2934 dled. For example, \xdc is exactly the same as \x{dc}.
2935
2936 After \0 up to two further octal digits are read. If there are fewer
2937 than two digits, just those that are present are used. Thus the
2938 sequence \0\x\07 specifies two binary zeros followed by a BEL character
2939 (code value 7). Make sure you supply two digits after the initial zero
2940 if the pattern character that follows is itself an octal digit.
2941
2942 The handling of a backslash followed by a digit other than 0 is compli-
2943 cated. Outside a character class, PCRE reads it and any following dig-
2944 its as a decimal number. If the number is less than 10, or if there
2945 have been at least that many previous capturing left parentheses in the
2946 expression, the entire sequence is taken as a back reference. A
2947 description of how this works is given later, following the discussion
2948 of parenthesized subpatterns.
2949
2950 Inside a character class, or if the decimal number is greater than 9
2951 and there have not been that many capturing subpatterns, PCRE re-reads
2952 up to three octal digits following the backslash, and uses them to gen-
2953 erate a data character. Any subsequent digits stand for themselves. In
2954 non-UTF-8 mode, the value of a character specified in octal must be
2955 less than \400. In UTF-8 mode, values up to \777 are permitted. For
2956 example:
2957
2958 \040 is another way of writing a space
2959 \40 is the same, provided there are fewer than 40
2960 previous capturing subpatterns
2961 \7 is always a back reference
2962 \11 might be a back reference, or another way of
2963 writing a tab
2964 \011 is always a tab
2965 \0113 is a tab followed by the character "3"
2966 \113 might be a back reference, otherwise the
2967 character with octal code 113
2968 \377 might be a back reference, otherwise
2969 the byte consisting entirely of 1 bits
2970 \81 is either a back reference, or a binary zero
2971 followed by the two characters "8" and "1"
2972
2973 Note that octal values of 100 or greater must not be introduced by a
2974 leading zero, because no more than three octal digits are ever read.
2975
2976 All the sequences that define a single character value can be used both
2977 inside and outside character classes. In addition, inside a character
2978 class, the sequence \b is interpreted as the backspace character (hex
2979 08), and the sequences \R and \X are interpreted as the characters "R"
2980 and "X", respectively. Outside a character class, these sequences have
2981 different meanings (see below).
2982
2983 Absolute and relative back references
2984
2985 The sequence \g followed by an unsigned or a negative number, option-
2986 ally enclosed in braces, is an absolute or relative back reference. A
2987 named back reference can be coded as \g{name}. Back references are dis-
2988 cussed later, following the discussion of parenthesized subpatterns.
2989
2990 Generic character types
2991
2992 Another use of backslash is for specifying generic character types. The
2993 following are always recognized:
2994
2995 \d any decimal digit
2996 \D any character that is not a decimal digit
2997 \h any horizontal whitespace character
2998 \H any character that is not a horizontal whitespace character
2999 \s any whitespace character
3000 \S any character that is not a whitespace character
3001 \v any vertical whitespace character
3002 \V any character that is not a vertical whitespace character
3003 \w any "word" character
3004 \W any "non-word" character
3005
3006 Each pair of escape sequences partitions the complete set of characters
3007 into two disjoint sets. Any given character matches one, and only one,
3008 of each pair.
3009
3010 These character type sequences can appear both inside and outside char-
3011 acter classes. They each match one character of the appropriate type.
3012 If the current matching point is at the end of the subject string, all
3013 of them fail, since there is no character to match.
3014
3015 For compatibility with Perl, \s does not match the VT character (code
3016 11). This makes it different from the the POSIX "space" class. The \s
3017 characters are HT (9), LF (10), FF (12), CR (13), and space (32). If
3018 "use locale;" is included in a Perl script, \s may match the VT charac-
3019 ter. In PCRE, it never does.
3020
3021 In UTF-8 mode, characters with values greater than 128 never match \d,
3022 \s, or \w, and always match \D, \S, and \W. This is true even when Uni-
3023 code character property support is available. These sequences retain
3024 their original meanings from before UTF-8 support was available, mainly
3025 for efficiency reasons.
3026
3027 The sequences \h, \H, \v, and \V are Perl 5.10 features. In contrast to
3028 the other sequences, these do match certain high-valued codepoints in
3029 UTF-8 mode. The horizontal space characters are:
3030
3031 U+0009 Horizontal tab
3032 U+0020 Space
3033 U+00A0 Non-break space
3034 U+1680 Ogham space mark
3035 U+180E Mongolian vowel separator
3036 U+2000 En quad
3037 U+2001 Em quad
3038 U+2002 En space
3039 U+2003 Em space
3040 U+2004 Three-per-em space
3041 U+2005 Four-per-em space
3042 U+2006 Six-per-em space
3043 U+2007 Figure space
3044 U+2008 Punctuation space
3045 U+2009 Thin space
3046 U+200A Hair space
3047 U+202F Narrow no-break space
3048 U+205F Medium mathematical space
3049 U+3000 Ideographic space
3050
3051 The vertical space characters are:
3052
3053 U+000A Linefeed
3054 U+000B Vertical tab
3055 U+000C Formfeed
3056 U+000D Carriage return
3057 U+0085 Next line
3058 U+2028 Line separator
3059 U+2029 Paragraph separator
3060
3061 A "word" character is an underscore or any character less than 256 that
3062 is a letter or digit. The definition of letters and digits is con-
3063 trolled by PCRE's low-valued character tables, and may vary if locale-
3064 specific matching is taking place (see "Locale support" in the pcreapi
3065 page). For example, in a French locale such as "fr_FR" in Unix-like
3066 systems, or "french" in Windows, some character codes greater than 128
3067 are used for accented letters, and these are matched by \w. The use of
3068 locales with Unicode is discouraged.
3069
3070 Newline sequences
3071
3072 Outside a character class, the escape sequence \R matches any Unicode
3073 newline sequence. This is a Perl 5.10 feature. In non-UTF-8 mode \R is
3074 equivalent to the following:
3075
3076 (?>\r\n|\n|\x0b|\f|\r|\x85)
3077
3078 This is an example of an "atomic group", details of which are given
3079 below. This particular group matches either the two-character sequence
3080 CR followed by LF, or one of the single characters LF (linefeed,
3081 U+000A), VT (vertical tab, U+000B), FF (formfeed, U+000C), CR (carriage
3082 return, U+000D), or NEL (next line, U+0085). The two-character sequence
3083 is treated as a single unit that cannot be split.
3084
3085 In UTF-8 mode, two additional characters whose codepoints are greater
3086 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
3087 rator, U+2029). Unicode character property support is not needed for
3088 these characters to be recognized.
3089
3090 Inside a character class, \R matches the letter "R".
3091
3092 Unicode character properties
3093
3094 When PCRE is built with Unicode character property support, three addi-
3095 tional escape sequences that match characters with specific properties
3096 are available. When not in UTF-8 mode, these sequences are of course
3097 limited to testing characters whose codepoints are less than 256, but
3098 they do work in this mode. The extra escape sequences are:
3099
3100 \p{xx} a character with the xx property
3101 \P{xx} a character without the xx property
3102 \X an extended Unicode sequence
3103
3104 The property names represented by xx above are limited to the Unicode
3105 script names, the general category properties, and "Any", which matches
3106 any character (including newline). Other properties such as "InMusical-
3107 Symbols" are not currently supported by PCRE. Note that \P{Any} does
3108 not match any characters, so always causes a match failure.
3109
3110 Sets of Unicode characters are defined as belonging to certain scripts.
3111 A character from one of these sets can be matched using a script name.
3112 For example:
3113
3114 \p{Greek}
3115 \P{Han}
3116
3117 Those that are not part of an identified script are lumped together as
3118 "Common". The current list of scripts is:
3119
3120 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
3121 Buhid, Canadian_Aboriginal, Cherokee, Common, Coptic, Cuneiform,
3122 Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian, Glagolitic,
3123 Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
3124 gana, Inherited, Kannada, Katakana, Kharoshthi, Khmer, Lao, Latin,
3125 Limbu, Linear_B, Malayalam, Mongolian, Myanmar, New_Tai_Lue, Nko,
3126 Ogham, Old_Italic, Old_Persian, Oriya, Osmanya, Phags_Pa, Phoenician,
3127 Runic, Shavian, Sinhala, Syloti_Nagri, Syriac, Tagalog, Tagbanwa,
3128 Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Yi.
3129
3130 Each character has exactly one general category property, specified by
3131 a two-letter abbreviation. For compatibility with Perl, negation can be
3132 specified by including a circumflex between the opening brace and the
3133 property name. For example, \p{^Lu} is the same as \P{Lu}.
3134
3135 If only one letter is specified with \p or \P, it includes all the gen-
3136 eral category properties that start with that letter. In this case, in
3137 the absence of negation, the curly brackets in the escape sequence are
3138 optional; these two examples have the same effect:
3139
3140 \p{L}
3141 \pL
3142
3143 The following general category property codes are supported:
3144
3145 C Other
3146 Cc Control
3147 Cf Format
3148 Cn Unassigned
3149 Co Private use
3150 Cs Surrogate
3151
3152 L Letter
3153 Ll Lower case letter
3154 Lm Modifier letter
3155 Lo Other letter
3156 Lt Title case letter
3157 Lu Upper case letter
3158
3159 M Mark
3160 Mc Spacing mark
3161 Me Enclosing mark
3162 Mn Non-spacing mark
3163
3164 N Number
3165 Nd Decimal number
3166 Nl Letter number
3167 No Other number
3168
3169 P Punctuation
3170 Pc Connector punctuation
3171 Pd Dash punctuation
3172 Pe Close punctuation
3173 Pf Final punctuation
3174 Pi Initial punctuation
3175 Po Other punctuation
3176 Ps Open punctuation
3177
3178 S Symbol
3179 Sc Currency symbol
3180 Sk Modifier symbol
3181 Sm Mathematical symbol
3182 So Other symbol
3183
3184 Z Separator
3185 Zl Line separator
3186 Zp Paragraph separator
3187 Zs Space separator
3188
3189 The special property L& is also supported: it matches a character that
3190 has the Lu, Ll, or Lt property, in other words, a letter that is not
3191 classified as a modifier or "other".
3192
3193 The Cs (Surrogate) property applies only to characters in the range
3194 U+D800 to U+DFFF. Such characters are not valid in UTF-8 strings (see
3195 RFC 3629) and so cannot be tested by PCRE, unless UTF-8 validity check-
3196 ing has been turned off (see the discussion of PCRE_NO_UTF8_CHECK in
3197 the pcreapi page).
3198
3199 The long synonyms for these properties that Perl supports (such as
3200 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
3201 any of these properties with "Is".
3202
3203 No character that is in the Unicode table has the Cn (unassigned) prop-
3204 erty. Instead, this property is assumed for any code point that is not
3205 in the Unicode table.
3206
3207 Specifying caseless matching does not affect these escape sequences.
3208 For example, \p{Lu} always matches only upper case letters.
3209
3210 The \X escape matches any number of Unicode characters that form an
3211 extended Unicode sequence. \X is equivalent to
3212
3213 (?>\PM\pM*)
3214
3215 That is, it matches a character without the "mark" property, followed
3216 by zero or more characters with the "mark" property, and treats the
3217 sequence as an atomic group (see below). Characters with the "mark"
3218 property are typically accents that affect the preceding character.
3219 None of them have codepoints less than 256, so in non-UTF-8 mode \X
3220 matches any one character.
3221
3222 Matching characters by Unicode property is not fast, because PCRE has
3223 to search a structure that contains data for over fifteen thousand
3224 characters. That is why the traditional escape sequences such as \d and
3225 \w do not use Unicode properties in PCRE.
3226
3227 Resetting the match start
3228
3229 The escape sequence \K, which is a Perl 5.10 feature, causes any previ-
3230 ously matched characters not to be included in the final matched
3231 sequence. For example, the pattern:
3232
3233 foo\Kbar
3234
3235 matches "foobar", but reports that it has matched "bar". This feature
3236 is similar to a lookbehind assertion (described below). However, in
3237 this case, the part of the subject before the real match does not have
3238 to be of fixed length, as lookbehind assertions do. The use of \K does
3239 not interfere with the setting of captured substrings. For example,
3240 when the pattern
3241
3242 (foo)\Kbar
3243
3244 matches "foobar", the first substring is still set to "foo".
3245
3246 Simple assertions
3247
3248 The final use of backslash is for certain simple assertions. An asser-
3249 tion specifies a condition that has to be met at a particular point in
3250 a match, without consuming any characters from the subject string. The
3251 use of subpatterns for more complicated assertions is described below.
3252 The backslashed assertions are:
3253
3254 \b matches at a word boundary
3255 \B matches when not at a word boundary
3256 \A matches at the start of the subject
3257 \Z matches at the end of the subject
3258 also matches before a newline at the end of the subject
3259 \z matches only at the end of the subject
3260 \G matches at the first matching position in the subject
3261
3262 These assertions may not appear in character classes (but note that \b
3263 has a different meaning, namely the backspace character, inside a char-
3264 acter class).
3265
3266 A word boundary is a position in the subject string where the current
3267 character and the previous character do not both match \w or \W (i.e.
3268 one matches \w and the other matches \W), or the start or end of the
3269 string if the first or last character matches \w, respectively.
3270
3271 The \A, \Z, and \z assertions differ from the traditional circumflex
3272 and dollar (described in the next section) in that they only ever match
3273 at the very start and end of the subject string, whatever options are
3274 set. Thus, they are independent of multiline mode. These three asser-
3275 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
3276 affect only the behaviour of the circumflex and dollar metacharacters.
3277 However, if the startoffset argument of pcre_exec() is non-zero, indi-
3278 cating that matching is to start at a point other than the beginning of
3279 the subject, \A can never match. The difference between \Z and \z is
3280 that \Z matches before a newline at the end of the string as well as at
3281 the very end, whereas \z matches only at the end.
3282
3283 The \G assertion is true only when the current matching position is at
3284 the start point of the match, as specified by the startoffset argument
3285 of pcre_exec(). It differs from \A when the value of startoffset is
3286 non-zero. By calling pcre_exec() multiple times with appropriate argu-
3287 ments, you can mimic Perl's /g option, and it is in this kind of imple-
3288 mentation where \G can be useful.
3289
3290 Note, however, that PCRE's interpretation of \G, as the start of the
3291 current match, is subtly different from Perl's, which defines it as the
3292 end of the previous match. In Perl, these can be different when the
3293 previously matched string was empty. Because PCRE does just one match
3294 at a time, it cannot reproduce this behaviour.
3295
3296 If all the alternatives of a pattern begin with \G, the expression is
3297 anchored to the starting match position, and the "anchored" flag is set
3298 in the compiled regular expression.
3299
3300
3301 CIRCUMFLEX AND DOLLAR
3302
3303 Outside a character class, in the default matching mode, the circumflex
3304 character is an assertion that is true only if the current matching
3305 point is at the start of the subject string. If the startoffset argu-
3306 ment of pcre_exec() is non-zero, circumflex can never match if the
3307 PCRE_MULTILINE option is unset. Inside a character class, circumflex
3308 has an entirely different meaning (see below).
3309
3310 Circumflex need not be the first character of the pattern if a number
3311 of alternatives are involved, but it should be the first thing in each
3312 alternative in which it appears if the pattern is ever to match that
3313 branch. If all possible alternatives start with a circumflex, that is,
3314 if the pattern is constrained to match only at the start of the sub-
3315 ject, it is said to be an "anchored" pattern. (There are also other
3316 constructs that can cause a pattern to be anchored.)
3317
3318 A dollar character is an assertion that is true only if the current
3319 matching point is at the end of the subject string, or immediately
3320 before a newline at the end of the string (by default). Dollar need not
3321 be the last character of the pattern if a number of alternatives are
3322 involved, but it should be the last item in any branch in which it
3323 appears. Dollar has no special meaning in a character class.
3324
3325 The meaning of dollar can be changed so that it matches only at the
3326 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
3327 compile time. This does not affect the \Z assertion.
3328
3329 The meanings of the circumflex and dollar characters are changed if the
3330 PCRE_MULTILINE option is set. When this is the case, a circumflex
3331 matches immediately after internal newlines as well as at the start of
3332 the subject string. It does not match after a newline that ends the
3333 string. A dollar matches before any newlines in the string, as well as
3334 at the very end, when PCRE_MULTILINE is set. When newline is specified
3335 as the two-character sequence CRLF, isolated CR and LF characters do
3336 not indicate newlines.
3337
3338 For example, the pattern /^abc$/ matches the subject string "def\nabc"
3339 (where \n represents a newline) in multiline mode, but not otherwise.
3340 Consequently, patterns that are anchored in single line mode because
3341 all branches start with ^ are not anchored in multiline mode, and a
3342 match for circumflex is possible when the startoffset argument of
3343 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
3344 PCRE_MULTILINE is set.
3345
3346 Note that the sequences \A, \Z, and \z can be used to match the start
3347 and end of the subject in both modes, and if all branches of a pattern
3348 start with \A it is always anchored, whether or not PCRE_MULTILINE is
3349 set.
3350
3351
3352 FULL STOP (PERIOD, DOT)
3353
3354 Outside a character class, a dot in the pattern matches any one charac-
3355 ter in the subject string except (by default) a character that signi-
3356 fies the end of a line. In UTF-8 mode, the matched character may be
3357 more than one byte long.
3358
3359 When a line ending is defined as a single character, dot never matches
3360 that character; when the two-character sequence CRLF is used, dot does
3361 not match CR if it is immediately followed by LF, but otherwise it
3362 matches all characters (including isolated CRs and LFs). When any Uni-
3363 code line endings are being recognized, dot does not match CR or LF or
3364 any of the other line ending characters.
3365
3366 The behaviour of dot with regard to newlines can be changed. If the
3367 PCRE_DOTALL option is set, a dot matches any one character, without
3368 exception. If the two-character sequence CRLF is present in the subject
3369 string, it takes two dots to match it.
3370
3371 The handling of dot is entirely independent of the handling of circum-
3372 flex and dollar, the only relationship being that they both involve
3373 newlines. Dot has no special meaning in a character class.
3374
3375
3376 MATCHING A SINGLE BYTE
3377
3378 Outside a character class, the escape sequence \C matches any one byte,
3379 both in and out of UTF-8 mode. Unlike a dot, it always matches any
3380 line-ending characters. The feature is provided in Perl in order to
3381 match individual bytes in UTF-8 mode. Because it breaks up UTF-8 char-
3382 acters into individual bytes, what remains in the string may be a mal-
3383 formed UTF-8 string. For this reason, the \C escape sequence is best
3384 avoided.
3385
3386 PCRE does not allow \C to appear in lookbehind assertions (described
3387 below), because in UTF-8 mode this would make it impossible to calcu-
3388 late the length of the lookbehind.
3389
3390
3391 SQUARE BRACKETS AND CHARACTER CLASSES
3392
3393 An opening square bracket introduces a character class, terminated by a
3394 closing square bracket. A closing square bracket on its own is not spe-
3395 cial. If a closing square bracket is required as a member of the class,
3396 it should be the first data character in the class (after an initial
3397 circumflex, if present) or escaped with a backslash.
3398
3399 A character class matches a single character in the subject. In UTF-8
3400 mode, the character may occupy more than one byte. A matched character
3401 must be in the set of characters defined by the class, unless the first
3402 character in the class definition is a circumflex, in which case the
3403 subject character must not be in the set defined by the class. If a
3404 circumflex is actually required as a member of the class, ensure it is
3405 not the first character, or escape it with a backslash.
3406
3407 For example, the character class [aeiou] matches any lower case vowel,
3408 while [^aeiou] matches any character that is not a lower case vowel.
3409 Note that a circumflex is just a convenient notation for specifying the
3410 characters that are in the class by enumerating those that are not. A
3411 class that starts with a circumflex is not an assertion: it still con-
3412 sumes a character from the subject string, and therefore it fails if
3413 the current pointer is at the end of the string.
3414
3415 In UTF-8 mode, characters with values greater than 255 can be included
3416 in a class as a literal string of bytes, or by using the \x{ escaping
3417 mechanism.
3418
3419 When caseless matching is set, any letters in a class represent both
3420 their upper case and lower case versions, so for example, a caseless
3421 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
3422 match "A", whereas a caseful version would. In UTF-8 mode, PCRE always
3423 understands the concept of case for characters whose values are less
3424 than 128, so caseless matching is always possible. For characters with
3425 higher values, the concept of case is supported if PCRE is compiled
3426 with Unicode property support, but not otherwise. If you want to use
3427 caseless matching for characters 128 and above, you must ensure that
3428 PCRE is compiled with Unicode property support as well as with UTF-8
3429 support.
3430
3431 Characters that might indicate line breaks are never treated in any
3432 special way when matching character classes, whatever line-ending
3433 sequence is in use, and whatever setting of the PCRE_DOTALL and
3434 PCRE_MULTILINE options is used. A class such as [^a] always matches one
3435 of these characters.
3436
3437 The minus (hyphen) character can be used to specify a range of charac-
3438 ters in a character class. For example, [d-m] matches any letter
3439 between d and m, inclusive. If a minus character is required in a
3440 class, it must be escaped with a backslash or appear in a position
3441 where it cannot be interpreted as indicating a range, typically as the
3442 first or last character in the class.
3443
3444 It is not possible to have the literal character "]" as the end charac-
3445 ter of a range. A pattern such as [W-]46] is interpreted as a class of
3446 two characters ("W" and "-") followed by a literal string "46]", so it
3447 would match "W46]" or "-46]". However, if the "]" is escaped with a
3448 backslash it is interpreted as the end of range, so [W-\]46] is inter-
3449 preted as a class containing a range followed by two other characters.
3450 The octal or hexadecimal representation of "]" can also be used to end
3451 a range.
3452
3453 Ranges operate in the collating sequence of character values. They can
3454 also be used for characters specified numerically, for example
3455 [\000-\037]. In UTF-8 mode, ranges can include characters whose values
3456 are greater than 255, for example [\x{100}-\x{2ff}].
3457
3458 If a range that includes letters is used when caseless matching is set,
3459 it matches the letters in either case. For example, [W-c] is equivalent
3460 to [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if
3461 character tables for a French locale are in use, [\xc8-\xcb] matches
3462 accented E characters in both cases. In UTF-8 mode, PCRE supports the
3463 concept of case for characters with values greater than 128 only when
3464 it is compiled with Unicode property support.
3465
3466 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
3467 in a character class, and add the characters that they match to the
3468 class. For example, [\dABCDEF] matches any hexadecimal digit. A circum-
3469 flex can conveniently be used with the upper case character types to
3470 specify a more restricted set of characters than the matching lower
3471 case type. For example, the class [^\W_] matches any letter or digit,
3472 but not underscore.
3473
3474 The only metacharacters that are recognized in character classes are
3475 backslash, hyphen (only where it can be interpreted as specifying a
3476 range), circumflex (only at the start), opening square bracket (only
3477 when it can be interpreted as introducing a POSIX class name - see the
3478 next section), and the terminating closing square bracket. However,
3479 escaping other non-alphanumeric characters does no harm.
3480
3481
3482 POSIX CHARACTER CLASSES
3483
3484 Perl supports the POSIX notation for character classes. This uses names
3485 enclosed by [: and :] within the enclosing square brackets. PCRE also
3486 supports this notation. For example,
3487
3488 [01[:alpha:]%]
3489
3490 matches "0", "1", any alphabetic character, or "%". The supported class
3491 names are
3492
3493 alnum letters and digits
3494 alpha letters
3495 ascii character codes 0 - 127
3496 blank space or tab only
3497 cntrl control characters
3498 digit decimal digits (same as \d)
3499 graph printing characters, excluding space
3500 lower lower case letters
3501 print printing characters, including space
3502 punct printing characters, excluding letters and digits
3503 space white space (not quite the same as \s)
3504 upper upper case letters
3505 word "word" characters (same as \w)
3506 xdigit hexadecimal digits
3507
3508 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
3509 and space (32). Notice that this list includes the VT character (code
3510 11). This makes "space" different to \s, which does not include VT (for
3511 Perl compatibility).
3512
3513 The name "word" is a Perl extension, and "blank" is a GNU extension
3514 from Perl 5.8. Another Perl extension is negation, which is indicated
3515 by a ^ character after the colon. For example,
3516
3517 [12[:^digit:]]
3518
3519 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
3520 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
3521 these are not supported, and an error is given if they are encountered.
3522
3523 In UTF-8 mode, characters with values greater than 128 do not match any
3524 of the POSIX character classes.
3525
3526
3527 VERTICAL BAR
3528
3529 Vertical bar characters are used to separate alternative patterns. For
3530 example, the pattern
3531
3532 gilbert|sullivan
3533
3534 matches either "gilbert" or "sullivan". Any number of alternatives may
3535 appear, and an empty alternative is permitted (matching the empty
3536 string). The matching process tries each alternative in turn, from left
3537 to right, and the first one that succeeds is used. If the alternatives
3538 are within a subpattern (defined below), "succeeds" means matching the
3539 rest of the main pattern as well as the alternative in the subpattern.
3540
3541
3542 INTERNAL OPTION SETTING
3543
3544 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
3545 PCRE_EXTENDED options can be changed from within the pattern by a
3546 sequence of Perl option letters enclosed between "(?" and ")". The
3547 option letters are
3548
3549 i for PCRE_CASELESS
3550 m for PCRE_MULTILINE
3551 s for PCRE_DOTALL
3552 x for PCRE_EXTENDED
3553
3554 For example, (?im) sets caseless, multiline matching. It is also possi-
3555 ble to unset these options by preceding the letter with a hyphen, and a
3556 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
3557 LESS and PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED,
3558 is also permitted. If a letter appears both before and after the
3559 hyphen, the option is unset.
3560
3561 When an option change occurs at top level (that is, not inside subpat-
3562 tern parentheses), the change applies to the remainder of the pattern
3563 that follows. If the change is placed right at the start of a pattern,
3564 PCRE extracts it into the global options (and it will therefore show up
3565 in data extracted by the pcre_fullinfo() function).
3566
3567 An option change within a subpattern (see below for a description of
3568 subpatterns) affects only that part of the current pattern that follows
3569 it, so
3570
3571 (a(?i)b)c
3572
3573 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
3574 used). By this means, options can be made to have different settings
3575 in different parts of the pattern. Any changes made in one alternative
3576 do carry on into subsequent branches within the same subpattern. For
3577 example,
3578
3579 (a(?i)b|c)
3580
3581 matches "ab", "aB", "c", and "C", even though when matching "C" the
3582 first branch is abandoned before the option setting. This is because
3583 the effects of option settings happen at compile time. There would be
3584 some very weird behaviour otherwise.
3585
3586 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA
3587 can be changed in the same way as the Perl-compatible options by using
3588 the characters J, U and X respectively.
3589
3590
3591 SUBPATTERNS
3592
3593 Subpatterns are delimited by parentheses (round brackets), which can be
3594 nested. Turning part of a pattern into a subpattern does two things:
3595
3596 1. It localizes a set of alternatives. For example, the pattern
3597
3598 cat(aract|erpillar|)
3599
3600 matches one of the words "cat", "cataract", or "caterpillar". Without
3601 the parentheses, it would match "cataract", "erpillar" or an empty
3602 string.
3603
3604 2. It sets up the subpattern as a capturing subpattern. This means
3605 that, when the whole pattern matches, that portion of the subject
3606 string that matched the subpattern is passed back to the caller via the
3607 ovector argument of pcre_exec(). Opening parentheses are counted from
3608 left to right (starting from 1) to obtain numbers for the capturing
3609 subpatterns.
3610
3611 For example, if the string "the red king" is matched against the pat-
3612 tern
3613
3614 the ((red|white) (king|queen))
3615
3616 the captured substrings are "red king", "red", and "king", and are num-
3617 bered 1, 2, and 3, respectively.
3618
3619 The fact that plain parentheses fulfil two functions is not always
3620 helpful. There are often times when a grouping subpattern is required
3621 without a capturing requirement. If an opening parenthesis is followed
3622 by a question mark and a colon, the subpattern does not do any captur-
3623 ing, and is not counted when computing the number of any subsequent
3624 capturing subpatterns. For example, if the string "the white queen" is
3625 matched against the pattern
3626
3627 the ((?:red|white) (king|queen))
3628
3629 the captured substrings are "white queen" and "queen", and are numbered
3630 1 and 2. The maximum number of capturing subpatterns is 65535.
3631
3632 As a convenient shorthand, if any option settings are required at the
3633 start of a non-capturing subpattern, the option letters may appear
3634 between the "?" and the ":". Thus the two patterns
3635
3636 (?i:saturday|sunday)
3637 (?:(?i)saturday|sunday)
3638
3639 match exactly the same set of strings. Because alternative branches are
3640 tried from left to right, and options are not reset until the end of
3641 the subpattern is reached, an option setting in one branch does affect
3642 subsequent branches, so the above patterns match "SUNDAY" as well as
3643 "Saturday".
3644
3645
3646 DUPLICATE SUBPATTERN NUMBERS
3647
3648 Perl 5.10 introduced a feature whereby each alternative in a subpattern
3649 uses the same numbers for its capturing parentheses. Such a subpattern
3650 starts with (?| and is itself a non-capturing subpattern. For example,
3651 consider this pattern:
3652
3653 (?|(Sat)ur|(Sun))day
3654
3655 Because the two alternatives are inside a (?| group, both sets of cap-
3656 turing parentheses are numbered one. Thus, when the pattern matches,
3657 you can look at captured substring number one, whichever alternative
3658 matched. This construct is useful when you want to capture part, but
3659 not all, of one of a number of alternatives. Inside a (?| group, paren-
3660 theses are numbered as usual, but the number is reset at the start of
3661 each branch. The numbers of any capturing buffers that follow the sub-
3662 pattern start after the highest number used in any branch. The follow-
3663 ing example is taken from the Perl documentation. The numbers under-
3664 neath show in which buffer the captured content will be stored.
3665
3666 # before ---------------branch-reset----------- after
3667 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
3668 # 1 2 2 3 2 3 4
3669
3670 A backreference or a recursive call to a numbered subpattern always
3671 refers to the first one in the pattern with the given number.
3672
3673 An alternative approach to using this "branch reset" feature is to use
3674 duplicate named subpatterns, as described in the next section.
3675
3676
3677 NAMED SUBPATTERNS
3678
3679 Identifying capturing parentheses by number is simple, but it can be
3680 very hard to keep track of the numbers in complicated regular expres-
3681 sions. Furthermore, if an expression is modified, the numbers may
3682 change. To help with this difficulty, PCRE supports the naming of sub-
3683 patterns. This feature was not added to Perl until release 5.10. Python
3684 had the feature earlier, and PCRE introduced it at release 4.0, using
3685 the Python syntax. PCRE now supports both the Perl and the Python syn-
3686 tax.
3687
3688 In PCRE, a subpattern can be named in one of three ways: (?<name>...)
3689 or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
3690 to capturing parentheses from other parts of the pattern, such as back-
3691 references, recursion, and conditions, can be made by name as well as
3692 by number.
3693
3694 Names consist of up to 32 alphanumeric characters and underscores.
3695 Named capturing parentheses are still allocated numbers as well as
3696 names, exactly as if the names were not present. The PCRE API provides
3697 function calls for extracting the name-to-number translation table from
3698 a compiled pattern. There is also a convenience function for extracting
3699 a captured substring by name.
3700
3701 By default, a name must be unique within a pattern, but it is possible
3702 to relax this constraint by setting the PCRE_DUPNAMES option at compile
3703 time. This can be useful for patterns where only one instance of the
3704 named parentheses can match. Suppose you want to match the name of a
3705 weekday, either as a 3-letter abbreviation or as the full name, and in
3706 both cases you want to extract the abbreviation. This pattern (ignoring
3707 the line breaks) does the job:
3708
3709 (?<DN>Mon|Fri|Sun)(?:day)?|
3710 (?<DN>Tue)(?:sday)?|
3711 (?<DN>Wed)(?:nesday)?|
3712 (?<DN>Thu)(?:rsday)?|
3713 (?<DN>Sat)(?:urday)?
3714
3715 There are five capturing substrings, but only one is ever set after a
3716 match. (An alternative way of solving this problem is to use a "branch
3717 reset" subpattern, as described in the previous section.)
3718
3719 The convenience function for extracting the data by name returns the
3720 substring for the first (and in this example, the only) subpattern of
3721 that name that matched. This saves searching to find which numbered
3722 subpattern it was. If you make a reference to a non-unique named sub-
3723 pattern from elsewhere in the pattern, the one that corresponds to the
3724 lowest number is used. For further details of the interfaces for han-
3725 dling named subpatterns, see the pcreapi documentation.
3726
3727
3728 REPETITION
3729
3730 Repetition is specified by quantifiers, which can follow any of the
3731 following items:
3732
3733 a literal data character
3734 the dot metacharacter
3735 the \C escape sequence
3736 the \X escape sequence (in UTF-8 mode with Unicode properties)
3737 the \R escape sequence
3738 an escape such as \d that matches a single character
3739 a character class
3740 a back reference (see next section)
3741 a parenthesized subpattern (unless it is an assertion)
3742
3743 The general repetition quantifier specifies a minimum and maximum num-
3744 ber of permitted matches, by giving the two numbers in curly brackets
3745 (braces), separated by a comma. The numbers must be less than 65536,
3746 and the first must be less than or equal to the second. For example:
3747
3748 z{2,4}
3749
3750 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
3751 special character. If the second number is omitted, but the comma is
3752 present, there is no upper limit; if the second number and the comma
3753 are both omitted, the quantifier specifies an exact number of required
3754 matches. Thus
3755
3756 [aeiou]{3,}
3757
3758 matches at least 3 successive vowels, but may match many more, while
3759
3760 \d{8}
3761
3762 matches exactly 8 digits. An opening curly bracket that appears in a
3763 position where a quantifier is not allowed, or one that does not match
3764 the syntax of a quantifier, is taken as a literal character. For exam-
3765 ple, {,6} is not a quantifier, but a literal string of four characters.
3766
3767 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to
3768 individual bytes. Thus, for example, \x{100}{2} matches two UTF-8 char-
3769 acters, each of which is represented by a two-byte sequence. Similarly,
3770 when Unicode property support is available, \X{3} matches three Unicode
3771 extended sequences, each of which may be several bytes long (and they
3772 may be of different lengths).
3773
3774 The quantifier {0} is permitted, causing the expression to behave as if
3775 the previous item and the quantifier were not present.
3776
3777 For convenience, the three most common quantifiers have single-charac-
3778 ter abbreviations:
3779
3780 * is equivalent to {0,}
3781 + is equivalent to {1,}
3782 ? is equivalent to {0,1}
3783
3784 It is possible to construct infinite loops by following a subpattern
3785 that can match no characters with a quantifier that has no upper limit,
3786 for example:
3787
3788 (a?)*
3789
3790 Earlier versions of Perl and PCRE used to give an error at compile time
3791 for such patterns. However, because there are cases where this can be
3792 useful, such patterns are now accepted, but if any repetition of the
3793 subpattern does in fact match no characters, the loop is forcibly bro-
3794 ken.
3795
3796 By default, the quantifiers are "greedy", that is, they match as much
3797 as possible (up to the maximum number of permitted times), without
3798 causing the rest of the pattern to fail. The classic example of where
3799 this gives problems is in trying to match comments in C programs. These
3800 appear between /* and */ and within the comment, individual * and /
3801 characters may appear. An attempt to match C comments by applying the
3802 pattern
3803
3804 /\*.*\*/
3805
3806 to the string
3807
3808 /* first comment */ not comment /* second comment */
3809
3810 fails, because it matches the entire string owing to the greediness of
3811 the .* item.
3812
3813 However, if a quantifier is followed by a question mark, it ceases to
3814 be greedy, and instead matches the minimum number of times possible, so
3815 the pattern
3816
3817 /\*.*?\*/
3818
3819 does the right thing with the C comments. The meaning of the various
3820 quantifiers is not otherwise changed, just the preferred number of
3821 matches. Do not confuse this use of question mark with its use as a
3822 quantifier in its own right. Because it has two uses, it can sometimes
3823 appear doubled, as in
3824
3825 \d??\d
3826
3827 which matches one digit by preference, but can match two if that is the
3828 only way the rest of the pattern matches.
3829
3830 If the PCRE_UNGREEDY option is set (an option that is not available in
3831 Perl), the quantifiers are not greedy by default, but individual ones
3832 can be made greedy by following them with a question mark. In other
3833 words, it inverts the default behaviour.
3834
3835 When a parenthesized subpattern is quantified with a minimum repeat
3836 count that is greater than 1 or with a limited maximum, more memory is
3837 required for the compiled pattern, in proportion to the size of the
3838 minimum or maximum.
3839
3840 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
3841 alent to Perl's /s) is set, thus allowing the dot to match newlines,
3842 the pattern is implicitly anchored, because whatever follows will be
3843 tried against every character position in the subject string, so there
3844 is no point in retrying the overall match at any position after the
3845 first. PCRE normally treats such a pattern as though it were preceded
3846 by \A.
3847
3848 In cases where it is known that the subject string contains no new-
3849 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
3850 mization, or alternatively using ^ to indicate anchoring explicitly.
3851
3852 However, there is one situation where the optimization cannot be used.
3853 When .* is inside capturing parentheses that are the subject of a
3854 backreference elsewhere in the pattern, a match at the start may fail
3855 where a later one succeeds. Consider, for example:
3856
3857 (.*)abc\1
3858
3859 If the subject is "xyz123abc123" the match point is the fourth charac-
3860 ter. For this reason, such a pattern is not implicitly anchored.
3861
3862 When a capturing subpattern is repeated, the value captured is the sub-
3863 string that matched the final iteration. For example, after
3864
3865 (tweedle[dume]{3}\s*)+
3866
3867 has matched "tweedledum tweedledee" the value of the captured substring
3868 is "tweedledee". However, if there are nested capturing subpatterns,
3869 the corresponding captured values may have been set in previous itera-
3870 tions. For example, after
3871
3872 /(a|(b))+/
3873
3874 matches "aba" the value of the second captured substring is "b".
3875
3876
3877 ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS
3878
3879 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
3880 repetition, failure of what follows normally causes the repeated item
3881 to be re-evaluated to see if a different number of repeats allows the
3882 rest of the pattern to match. Sometimes it is useful to prevent this,
3883 either to change the nature of the match, or to cause it fail earlier
3884 than it otherwise might, when the author of the pattern knows there is
3885 no point in carrying on.
3886
3887 Consider, for example, the pattern \d+foo when applied to the subject
3888 line
3889
3890 123456bar
3891
3892 After matching all 6 digits and then failing to match "foo", the normal
3893 action of the matcher is to try again with only 5 digits matching the
3894 \d+ item, and then with 4, and so on, before ultimately failing.
3895 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
3896 the means for specifying that once a subpattern has matched, it is not
3897 to be re-evaluated in this way.
3898
3899 If we use atomic grouping for the previous example, the matcher gives
3900 up immediately on failing to match "foo" the first time. The notation
3901 is a kind of special parenthesis, starting with (?> as in this example:
3902
3903 (?>\d+)foo
3904
3905 This kind of parenthesis "locks up" the part of the pattern it con-
3906 tains once it has matched, and a failure further into the pattern is
3907 prevented from backtracking into it. Backtracking past it to previous
3908 items, however, works as normal.
3909
3910 An alternative description is that a subpattern of this type matches
3911 the string of characters that an identical standalone pattern would
3912 match, if anchored at the current point in the subject string.
3913
3914 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
3915 such as the above example can be thought of as a maximizing repeat that
3916 must swallow everything it can. So, while both \d+ and \d+? are pre-
3917 pared to adjust the number of digits they match in order to make the
3918 rest of the pattern match, (?>\d+) can only match an entire sequence of
3919 digits.
3920
3921 Atomic groups in general can of course contain arbitrarily complicated
3922 subpatterns, and can be nested. However, when the subpattern for an
3923 atomic group is just a single repeated item, as in the example above, a
3924 simpler notation, called a "possessive quantifier" can be used. This
3925 consists of an additional + character following a quantifier. Using
3926 this notation, the previous example can be rewritten as
3927
3928 \d++foo
3929
3930 Note that a possessive quantifier can be used with an entire group, for
3931 example:
3932
3933 (abc|xyz){2,3}+
3934
3935 Possessive quantifiers are always greedy; the setting of the
3936 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
3937 simpler forms of atomic group. However, there is no difference in the
3938 meaning of a possessive quantifier and the equivalent atomic group,
3939 though there may be a performance difference; possessive quantifiers
3940 should be slightly faster.
3941
3942 The possessive quantifier syntax is an extension to the Perl 5.8 syn-
3943 tax. Jeffrey Friedl originated the idea (and the name) in the first
3944 edition of his book. Mike McCloskey liked it, so implemented it when he
3945 built Sun's Java package, and PCRE copied it from there. It ultimately
3946 found its way into Perl at release 5.10.
3947
3948 PCRE has an optimization that automatically "possessifies" certain sim-
3949 ple pattern constructs. For example, the sequence A+B is treated as
3950 A++B because there is no point in backtracking into a sequence of A's
3951 when B must follow.
3952
3953 When a pattern contains an unlimited repeat inside a subpattern that
3954 can itself be repeated an unlimited number of times, the use of an
3955 atomic group is the only way to avoid some failing matches taking a
3956 very long time indeed. The pattern
3957
3958 (\D+|<\d+>)*[!?]
3959
3960 matches an unlimited number of substrings that either consist of non-
3961 digits, or digits enclosed in <>, followed by either ! or ?. When it
3962 matches, it runs quickly. However, if it is applied to
3963
3964 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
3965
3966 it takes a long time before reporting failure. This is because the
3967 string can be divided between the internal \D+ repeat and the external
3968 * repeat in a large number of ways, and all have to be tried. (The
3969 example uses [!?] rather than a single character at the end, because
3970 both PCRE and Perl have an optimization that allows for fast failure
3971 when a single character is used. They remember the last single charac-
3972 ter that is required for a match, and fail early if it is not present
3973 in the string.) If the pattern is changed so that it uses an atomic
3974 group, like this:
3975
3976 ((?>\D+)|<\d+>)*[!?]
3977
3978 sequences of non-digits cannot be broken, and failure happens quickly.
3979
3980
3981 BACK REFERENCES
3982
3983 Outside a character class, a backslash followed by a digit greater than
3984 0 (and possibly further digits) is a back reference to a capturing sub-
3985 pattern earlier (that is, to its left) in the pattern, provided there
3986 have been that many previous capturing left parentheses.
3987
3988 However, if the decimal number following the backslash is less than 10,
3989 it is always taken as a back reference, and causes an error only if
3990 there are not that many capturing left parentheses in the entire pat-
3991 tern. In other words, the parentheses that are referenced need not be
3992 to the left of the reference for numbers less than 10. A "forward back
3993 reference" of this type can make sense when a repetition is involved
3994 and the subpattern to the right has participated in an earlier itera-
3995 tion.
3996
3997 It is not possible to have a numerical "forward back reference" to a
3998 subpattern whose number is 10 or more using this syntax because a
3999 sequence such as \50 is interpreted as a character defined in octal.
4000 See the subsection entitled "Non-printing characters" above for further
4001 details of the handling of digits following a backslash. There is no
4002 such problem when named parentheses are used. A back reference to any
4003 subpattern is possible using named parentheses (see below).
4004
4005 Another way of avoiding the ambiguity inherent in the use of digits
4006 following a backslash is to use the \g escape sequence, which is a fea-
4007 ture introduced in Perl 5.10. This escape must be followed by an
4008 unsigned number or a negative number, optionally enclosed in braces.
4009 These examples are all identical:
4010
4011 (ring), \1
4012 (ring), \g1
4013 (ring), \g{1}
4014
4015 An unsigned number specifies an absolute reference without the ambigu-
4016 ity that is present in the older syntax. It is also useful when literal
4017 digits follow the reference. A negative number is a relative reference.
4018 Consider this example:
4019
4020 (abc(def)ghi)\g{-1}
4021
4022 The sequence \g{-1} is a reference to the most recently started captur-
4023 ing subpattern before \g, that is, is it equivalent to \2. Similarly,
4024 \g{-2} would be equivalent to \1. The use of relative references can be
4025 helpful in long patterns, and also in patterns that are created by
4026 joining together fragments that contain references within themselves.
4027
4028 A back reference matches whatever actually matched the capturing sub-
4029 pattern in the current subject string, rather than anything matching
4030 the subpattern itself (see "Subpatterns as subroutines" below for a way
4031 of doing that). So the pattern
4032
4033 (sens|respons)e and \1ibility
4034
4035 matches "sense and sensibility" and "response and responsibility", but
4036 not "sense and responsibility". If caseful matching is in force at the
4037 time of the back reference, the case of letters is relevant. For exam-
4038 ple,
4039
4040 ((?i)rah)\s+\1
4041
4042 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
4043 original capturing subpattern is matched caselessly.
4044
4045 There are several different ways of writing back references to named
4046 subpatterns. The .NET syntax \k{name} and the Perl syntax \k<name> or
4047 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's
4048 unified back reference syntax, in which \g can be used for both numeric
4049 and named references, is also supported. We could rewrite the above
4050 example in any of the following ways:
4051
4052 (?<p1>(?i)rah)\s+\k<p1>
4053 (?'p1'(?i)rah)\s+\k{p1}
4054 (?P<p1>(?i)rah)\s+(?P=p1)
4055 (?<p1>(?i)rah)\s+\g{p1}
4056
4057 A subpattern that is referenced by name may appear in the pattern
4058 before or after the reference.
4059
4060 There may be more than one back reference to the same subpattern. If a
4061 subpattern has not actually been used in a particular match, any back
4062 references to it always fail. For example, the pattern
4063
4064 (a|(bc))\2
4065
4066 always fails if it starts to match "a" rather than "bc". Because there
4067 may be many capturing parentheses in a pattern, all digits following
4068 the backslash are taken as part of a potential back reference number.
4069 If the pattern continues with a digit character, some delimiter must be
4070 used to terminate the back reference. If the PCRE_EXTENDED option is
4071 set, this can be whitespace. Otherwise an empty comment (see "Com-
4072 ments" below) can be used.
4073
4074 A back reference that occurs inside the parentheses to which it refers
4075 fails when the subpattern is first used, so, for example, (a\1) never
4076 matches. However, such references can be useful inside repeated sub-
4077 patterns. For example, the pattern
4078
4079 (a|b\1)+
4080
4081 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
4082 ation of the subpattern, the back reference matches the character
4083 string corresponding to the previous iteration. In order for this to
4084 work, the pattern must be such that the first iteration does not need
4085 to match the back reference. This can be done using alternation, as in
4086 the example above, or by a quantifier with a minimum of zero.
4087
4088
4089 ASSERTIONS
4090
4091 An assertion is a test on the characters following or preceding the
4092 current matching point that does not actually consume any characters.
4093 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
4094 described above.
4095
4096 More complicated assertions are coded as subpatterns. There are two
4097 kinds: those that look ahead of the current position in the subject
4098 string, and those that look behind it. An assertion subpattern is
4099 matched in the normal way, except that it does not cause the current
4100 matching position to be changed.
4101
4102 Assertion subpatterns are not capturing subpatterns, and may not be
4103 repeated, because it makes no sense to assert the same thing several
4104 times. If any kind of assertion contains capturing subpatterns within
4105 it, these are counted for the purposes of numbering the capturing sub-
4106 patterns in the whole pattern. However, substring capturing is carried
4107 out only for positive assertions, because it does not make sense for
4108 negative assertions.
4109
4110 Lookahead assertions
4111
4112 Lookahead assertions start with (?= for positive assertions and (?! for
4113 negative assertions. For example,
4114
4115 \w+(?=;)
4116
4117 matches a word followed by a semicolon, but does not include the semi-
4118 colon in the match, and
4119
4120 foo(?!bar)
4121
4122 matches any occurrence of "foo" that is not followed by "bar". Note
4123 that the apparently similar pattern
4124
4125 (?!foo)bar
4126
4127 does not find an occurrence of "bar" that is preceded by something
4128 other than "foo"; it finds any occurrence of "bar" whatsoever, because
4129 the assertion (?!foo) is always true when the next three characters are
4130 "bar". A lookbehind assertion is needed to achieve the other effect.
4131
4132 If you want to force a matching failure at some point in a pattern, the
4133 most convenient way to do it is with (?!) because an empty string
4134 always matches, so an assertion that requires there not to be an empty
4135 string must always fail.
4136
4137 Lookbehind assertions
4138
4139 Lookbehind assertions start with (?<= for positive assertions and (?<!
4140 for negative assertions. For example,
4141
4142 (?<!foo)bar
4143
4144 does find an occurrence of "bar" that is not preceded by "foo". The
4145 contents of a lookbehind assertion are restricted such that all the
4146 strings it matches must have a fixed length. However, if there are sev-
4147 eral top-level alternatives, they do not all have to have the same
4148 fixed length. Thus
4149
4150 (?<=bullock|donkey)
4151
4152 is permitted, but
4153
4154 (?<!dogs?|cats?)
4155
4156 causes an error at compile time. Branches that match different length
4157 strings are permitted only at the top level of a lookbehind assertion.
4158 This is an extension compared with Perl (at least for 5.8), which
4159 requires all branches to match the same length of string. An assertion
4160 such as
4161
4162 (?<=ab(c|de))
4163
4164 is not permitted, because its single top-level branch can match two
4165 different lengths, but it is acceptable if rewritten to use two top-
4166 level branches:
4167
4168 (?<=abc|abde)
4169
4170 In some cases, the Perl 5.10 escape sequence \K (see above) can be used
4171 instead of a lookbehind assertion; this is not restricted to a fixed-
4172 length.
4173
4174 The implementation of lookbehind assertions is, for each alternative,
4175 to temporarily move the current position back by the fixed length and
4176 then try to match. If there are insufficient characters before the cur-
4177 rent position, the assertion fails.
4178
4179 PCRE does not allow the \C escape (which matches a single byte in UTF-8
4180 mode) to appear in lookbehind assertions, because it makes it impossi-
4181 ble to calculate the length of the lookbehind. The \X and \R escapes,
4182 which can match different numbers of bytes, are also not permitted.
4183
4184 Possessive quantifiers can be used in conjunction with lookbehind
4185 assertions to specify efficient matching at the end of the subject
4186 string. Consider a simple pattern such as
4187
4188 abcd$
4189
4190 when applied to a long string that does not match. Because matching
4191 proceeds from left to right, PCRE will look for each "a" in the subject
4192 and then see if what follows matches the rest of the pattern. If the
4193 pattern is specified as
4194
4195 ^.*abcd$
4196
4197 the initial .* matches the entire string at first, but when this fails
4198 (because there is no following "a"), it backtracks to match all but the
4199 last character, then all but the last two characters, and so on. Once
4200 again the search for "a" covers the entire string, from right to left,
4201 so we are no better off. However, if the pattern is written as
4202
4203 ^.*+(?<=abcd)
4204
4205 there can be no backtracking for the .*+ item; it can match only the
4206 entire string. The subsequent lookbehind assertion does a single test
4207 on the last four characters. If it fails, the match fails immediately.
4208 For long strings, this approach makes a significant difference to the
4209 processing time.
4210
4211 Using multiple assertions
4212
4213 Several assertions (of any sort) may occur in succession. For example,
4214
4215 (?<=\d{3})(?<!999)foo
4216
4217 matches "foo" preceded by three digits that are not "999". Notice that
4218 each of the assertions is applied independently at the same point in
4219 the subject string. First there is a check that the previous three
4220 characters are all digits, and then there is a check that the same
4221 three characters are not "999". This pattern does not match "foo" pre-
4222 ceded by six characters, the first of which are digits and the last
4223 three of which are not "999". For example, it doesn't match "123abc-
4224 foo". A pattern to do that is
4225
4226 (?<=\d{3}...)(?<!999)foo
4227
4228 This time the first assertion looks at the preceding six characters,
4229 checking that the first three are digits, and then the second assertion
4230 checks that the preceding three characters are not "999".
4231
4232 Assertions can be nested in any combination. For example,
4233
4234 (?<=(?<!foo)bar)baz
4235
4236 matches an occurrence of "baz" that is preceded by "bar" which in turn
4237 is not preceded by "foo", while
4238
4239 (?<=\d{3}(?!999)...)foo
4240
4241 is another pattern that matches "foo" preceded by three digits and any
4242 three characters that are not "999".
4243
4244
4245 CONDITIONAL SUBPATTERNS
4246
4247 It is possible to cause the matching process to obey a subpattern con-
4248 ditionally or to choose between two alternative subpatterns, depending
4249 on the result of an assertion, or whether a previous capturing subpat-
4250 tern matched or not. The two possible forms of conditional subpattern
4251 are
4252
4253 (?(condition)yes-pattern)
4254 (?(condition)yes-pattern|no-pattern)
4255
4256 If the condition is satisfied, the yes-pattern is used; otherwise the
4257 no-pattern (if present) is used. If there are more than two alterna-
4258 tives in the subpattern, a compile-time error occurs.
4259
4260 There are four kinds of condition: references to subpatterns, refer-
4261 ences to recursion, a pseudo-condition called DEFINE, and assertions.
4262
4263 Checking for a used subpattern by number
4264
4265 If the text between the parentheses consists of a sequence of digits,
4266 the condition is true if the capturing subpattern of that number has
4267 previously matched. An alternative notation is to precede the digits
4268 with a plus or minus sign. In this case, the subpattern number is rela-
4269 tive rather than absolute. The most recently opened parentheses can be
4270 referenced by (?(-1), the next most recent by (?(-2), and so on. In
4271 looping constructs it can also make sense to refer to subsequent groups
4272 with constructs such as (?(+2).
4273
4274 Consider the following pattern, which contains non-significant white
4275 space to make it more readable (assume the PCRE_EXTENDED option) and to
4276 divide it into three parts for ease of discussion:
4277
4278 ( \( )? [^()]+ (?(1) \) )
4279
4280 The first part matches an optional opening parenthesis, and if that
4281 character is present, sets it as the first captured substring. The sec-
4282 ond part matches one or more characters that are not parentheses. The
4283 third part is a conditional subpattern that tests whether the first set
4284 of parentheses matched or not. If they did, that is, if subject started
4285 with an opening parenthesis, the condition is true, and so the yes-pat-
4286 tern is executed and a closing parenthesis is required. Otherwise,
4287 since no-pattern is not present, the subpattern matches nothing. In
4288 other words, this pattern matches a sequence of non-parentheses,
4289 optionally enclosed in parentheses.
4290
4291 If you were embedding this pattern in a larger one, you could use a
4292 relative reference:
4293
4294 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
4295
4296 This makes the fragment independent of the parentheses in the larger
4297 pattern.
4298
4299 Checking for a used subpattern by name
4300
4301 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
4302 used subpattern by name. For compatibility with earlier versions of
4303 PCRE, which had this facility before Perl, the syntax (?(name)...) is
4304 also recognized. However, there is a possible ambiguity with this syn-
4305 tax, because subpattern names may consist entirely of digits. PCRE
4306 looks first for a named subpattern; if it cannot find one and the name
4307 consists entirely of digits, PCRE looks for a subpattern of that num-
4308 ber, which must be greater than zero. Using subpattern names that con-
4309 sist entirely of digits is not recommended.
4310
4311 Rewriting the above example to use a named subpattern gives this:
4312
4313 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
4314
4315
4316 Checking for pattern recursion
4317
4318 If the condition is the string (R), and there is no subpattern with the
4319 name R, the condition is true if a recursive call to the whole pattern
4320 or any subpattern has been made. If digits or a name preceded by amper-
4321 sand follow the letter R, for example:
4322
4323 (?(R3)...) or (?(R&name)...)
4324
4325 the condition is true if the most recent recursion is into the subpat-
4326 tern whose number or name is given. This condition does not check the
4327 entire recursion stack.
4328
4329 At "top level", all these recursion test conditions are false. Recur-
4330 sive patterns are described below.
4331
4332 Defining subpatterns for use by reference only
4333
4334 If the condition is the string (DEFINE), and there is no subpattern
4335 with the name DEFINE, the condition is always false. In this case,
4336 there may be only one alternative in the subpattern. It is always
4337 skipped if control reaches this point in the pattern; the idea of
4338 DEFINE is that it can be used to define "subroutines" that can be ref-
4339 erenced from elsewhere. (The use of "subroutines" is described below.)
4340 For example, a pattern to match an IPv4 address could be written like
4341 this (ignore whitespace and line breaks):
4342
4343 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
4344 \b (?&byte) (\.(?&byte)){3} \b
4345
4346 The first part of the pattern is a DEFINE group inside which a another
4347 group named "byte" is defined. This matches an individual component of
4348 an IPv4 address (a number less than 256). When matching takes place,
4349 this part of the pattern is skipped because DEFINE acts like a false
4350 condition.
4351
4352 The rest of the pattern uses references to the named group to match the
4353 four dot-separated components of an IPv4 address, insisting on a word
4354 boundary at each end.
4355
4356 Assertion conditions
4357
4358 If the condition is not in any of the above formats, it must be an
4359 assertion. This may be a positive or negative lookahead or lookbehind
4360 assertion. Consider this pattern, again containing non-significant
4361 white space, and with the two alternatives on the second line:
4362
4363 (?(?=[^a-z]*[a-z])
4364 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
4365
4366 The condition is a positive lookahead assertion that matches an
4367 optional sequence of non-letters followed by a letter. In other words,
4368 it tests for the presence of at least one letter in the subject. If a
4369 letter is found, the subject is matched against the first alternative;
4370 otherwise it is matched against the second. This pattern matches
4371 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
4372 letters and dd are digits.
4373
4374
4375 COMMENTS
4376
4377 The sequence (?# marks the start of a comment that continues up to the
4378 next closing parenthesis. Nested parentheses are not permitted. The
4379 characters that make up a comment play no part in the pattern matching
4380 at all.
4381
4382 If the PCRE_EXTENDED option is set, an unescaped # character outside a
4383 character class introduces a comment that continues to immediately
4384 after the next newline in the pattern.
4385
4386
4387 RECURSIVE PATTERNS
4388
4389 Consider the problem of matching a string in parentheses, allowing for
4390 unlimited nested parentheses. Without the use of recursion, the best
4391 that can be done is to use a pattern that matches up to some fixed
4392 depth of nesting. It is not possible to handle an arbitrary nesting
4393 depth.
4394
4395 For some time, Perl has provided a facility that allows regular expres-
4396 sions to recurse (amongst other things). It does this by interpolating
4397 Perl code in the expression at run time, and the code can refer to the
4398 expression itself. A Perl pattern using code interpolation to solve the
4399 parentheses problem can be created like this:
4400
4401 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
4402
4403 The (?p{...}) item interpolates Perl code at run time, and in this case
4404 refers recursively to the pattern in which it appears.
4405
4406 Obviously, PCRE cannot support the interpolation of Perl code. Instead,
4407 it supports special syntax for recursion of the entire pattern, and
4408 also for individual subpattern recursion. After its introduction in
4409 PCRE and Python, this kind of recursion was introduced into Perl at
4410 release 5.10.
4411
4412 A special item that consists of (? followed by a number greater than
4413 zero and a closing parenthesis is a recursive call of the subpattern of
4414 the given number, provided that it occurs inside that subpattern. (If
4415 not, it is a "subroutine" call, which is described in the next sec-
4416 tion.) The special item (?R) or (?0) is a recursive call of the entire
4417 regular expression.
4418
4419 In PCRE (like Python, but unlike Perl), a recursive subpattern call is
4420 always treated as an atomic group. That is, once it has matched some of
4421 the subject string, it is never re-entered, even if it contains untried
4422 alternatives and there is a subsequent matching failure.
4423
4424 This PCRE pattern solves the nested parentheses problem (assume the
4425 PCRE_EXTENDED option is set so that white space is ignored):
4426
4427 \( ( (?>[^()]+) | (?R) )* \)
4428
4429 First it matches an opening parenthesis. Then it matches any number of
4430 substrings which can either be a sequence of non-parentheses, or a
4431 recursive match of the pattern itself (that is, a correctly parenthe-
4432 sized substring). Finally there is a closing parenthesis.
4433
4434 If this were part of a larger pattern, you would not want to recurse
4435 the entire pattern, so instead you could use this:
4436
4437 ( \( ( (?>[^()]+) | (?1) )* \) )
4438
4439 We have put the pattern into parentheses, and caused the recursion to
4440 refer to them instead of the whole pattern.
4441
4442 In a larger pattern, keeping track of parenthesis numbers can be
4443 tricky. This is made easier by the use of relative references. (A Perl
4444 5.10 feature.) Instead of (?1) in the pattern above you can write
4445 (?-2) to refer to the second most recently opened parentheses preceding
4446 the recursion. In other words, a negative number counts capturing
4447 parentheses leftwards from the point at which it is encountered.
4448
4449 It is also possible to refer to subsequently opened parentheses, by
4450 writing references such as (?+2). However, these cannot be recursive
4451 because the reference is not inside the parentheses that are refer-
4452 enced. They are always "subroutine" calls, as described in the next
4453 section.
4454
4455 An alternative approach is to use named parentheses instead. The Perl
4456 syntax for this is (?&name); PCRE's earlier syntax (?P>name) is also
4457 supported. We could rewrite the above example as follows:
4458
4459 (?<pn> \( ( (?>[^()]+) | (?&pn) )* \) )
4460
4461 If there is more than one subpattern with the same name, the earliest
4462 one is used.
4463
4464 This particular example pattern that we have been looking at contains
4465 nested unlimited repeats, and so the use of atomic grouping for match-
4466 ing strings of non-parentheses is important when applying the pattern
4467 to strings that do not match. For example, when this pattern is applied
4468 to
4469
4470 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
4471
4472 it yields "no match" quickly. However, if atomic grouping is not used,
4473 the match runs for a very long time indeed because there are so many
4474 different ways the + and * repeats can carve up the subject, and all
4475 have to be tested before failure can be reported.
4476
4477 At the end of a match, the values set for any capturing subpatterns are
4478 those from the outermost level of the recursion at which the subpattern
4479 value is set. If you want to obtain intermediate values, a callout
4480 function can be used (see below and the pcrecallout documentation). If
4481 the pattern above is matched against
4482
4483 (ab(cd)ef)
4484
4485 the value for the capturing parentheses is "ef", which is the last
4486 value taken on at the top level. If additional parentheses are added,
4487 giving
4488
4489 \( ( ( (?>[^()]+) | (?R) )* ) \)
4490 ^ ^
4491 ^ ^
4492
4493 the string they capture is "ab(cd)ef", the contents of the top level
4494 parentheses. If there are more than 15 capturing parentheses in a pat-
4495 tern, PCRE has to obtain extra memory to store data during a recursion,
4496 which it does by using pcre_malloc, freeing it via pcre_free after-
4497 wards. If no memory can be obtained, the match fails with the
4498 PCRE_ERROR_NOMEMORY error.
4499
4500 Do not confuse the (?R) item with the condition (R), which tests for
4501 recursion. Consider this pattern, which matches text in angle brack-
4502 ets, allowing for arbitrary nesting. Only digits are allowed in nested
4503 brackets (that is, when recursing), whereas any characters are permit-
4504 ted at the outer level.
4505
4506 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
4507
4508 In this pattern, (?(R) is the start of a conditional subpattern, with
4509 two different alternatives for the recursive and non-recursive cases.
4510 The (?R) item is the actual recursive call.
4511
4512
4513 SUBPATTERNS AS SUBROUTINES
4514
4515 If the syntax for a recursive subpattern reference (either by number or
4516 by name) is used outside the parentheses to which it refers, it oper-
4517 ates like a subroutine in a programming language. The "called" subpat-
4518 tern may be defined before or after the reference. A numbered reference
4519 can be absolute or relative, as in these examples:
4520
4521 (...(absolute)...)...(?2)...
4522 (...(relative)...)...(?-1)...
4523 (...(?+1)...(relative)...
4524
4525 An earlier example pointed out that the pattern
4526
4527 (sens|respons)e and \1ibility
4528
4529 matches "sense and sensibility" and "response and responsibility", but
4530 not "sense and responsibility". If instead the pattern
4531
4532 (sens|respons)e and (?1)ibility
4533
4534 is used, it does match "sense and responsibility" as well as the other
4535 two strings. Another example is given in the discussion of DEFINE
4536 above.
4537
4538 Like recursive subpatterns, a "subroutine" call is always treated as an
4539 atomic group. That is, once it has matched some of the subject string,
4540 it is never re-entered, even if it contains untried alternatives and
4541 there is a subsequent matching failure.
4542
4543 When a subpattern is used as a subroutine, processing options such as
4544 case-independence are fixed when the subpattern is defined. They cannot
4545 be changed for different calls. For example, consider this pattern:
4546
4547 (abc)(?i:(?-1))
4548
4549 It matches "abcabc". It does not match "abcABC" because the change of
4550 processing option does not affect the called subpattern.
4551
4552
4553 CALLOUTS
4554
4555 Perl has a feature whereby using the sequence (?{...}) causes arbitrary
4556 Perl code to be obeyed in the middle of matching a regular expression.
4557 This makes it possible, amongst other things, to extract different sub-
4558 strings that match the same pair of parentheses when there is a repeti-
4559 tion.
4560
4561 PCRE provides a similar feature, but of course it cannot obey arbitrary
4562 Perl code. The feature is called "callout". The caller of PCRE provides
4563 an external function by putting its entry point in the global variable
4564 pcre_callout. By default, this variable contains NULL, which disables
4565 all calling out.
4566
4567 Within a regular expression, (?C) indicates the points at which the
4568 external function is to be called. If you want to identify different
4569 callout points, you can put a number less than 256 after the letter C.
4570 The default value is zero. For example, this pattern has two callout
4571 points:
4572
4573 (?C1)abc(?C2)def
4574
4575 If the PCRE_AUTO_CALLOUT flag is passed to pcre_compile(), callouts are
4576 automatically installed before each item in the pattern. They are all
4577 numbered 255.
4578
4579 During matching, when PCRE reaches a callout point (and pcre_callout is
4580 set), the external function is called. It is provided with the number
4581 of the callout, the position in the pattern, and, optionally, one item
4582 of data originally supplied by the caller of pcre_exec(). The callout
4583 function may cause matching to proceed, to backtrack, or to fail alto-
4584 gether. A complete description of the interface to the callout function
4585 is given in the pcrecallout documentation.
4586
4587
4588 BACTRACKING CONTROL
4589
4590 Perl 5.10 introduced a number of "Special Backtracking Control Verbs",
4591 which are described in the Perl documentation as "experimental and sub-
4592 ject to change or removal in a future version of Perl". It goes on to
4593 say: "Their usage in production code should be noted to avoid problems
4594 during upgrades." The same remarks apply to the PCRE features described
4595 in this section.
4596
4597 Since these verbs are specifically related to backtracking, they can be
4598 used only when the pattern is to be matched using pcre_exec(), which
4599 uses a backtracking algorithm. They cause an error if encountered by
4600 pcre_dfa_exec().
4601
4602 The new verbs make use of what was previously invalid syntax: an open-
4603 ing parenthesis followed by an asterisk. In Perl, they are generally of
4604 the form (*VERB:ARG) but PCRE does not support the use of arguments, so
4605 its general form is just (*VERB). Any number of these verbs may occur
4606 in a pattern. There are two kinds:
4607
4608 Verbs that act immediately
4609
4610 The following verbs act as soon as they are encountered:
4611
4612 (*ACCEPT)
4613
4614 This verb causes the match to end successfully, skipping the remainder
4615 of the pattern. When inside a recursion, only the innermost pattern is
4616 ended immediately. PCRE differs from Perl in what happens if the
4617 (*ACCEPT) is inside capturing parentheses. In Perl, the data so far is
4618 captured: in PCRE no data is captured. For example:
4619
4620 A(A|B(*ACCEPT)|C)D
4621
4622 This matches "AB", "AAD", or "ACD", but when it matches "AB", no data
4623 is captured.
4624
4625 (*FAIL) or (*F)
4626
4627 This verb causes the match to fail, forcing backtracking to occur. It
4628 is equivalent to (?!) but easier to read. The Perl documentation notes
4629 that it is probably useful only when combined with (?{}) or (??{}).
4630 Those are, of course, Perl features that are not present in PCRE. The
4631 nearest equivalent is the callout feature, as for example in this pat-
4632 tern:
4633
4634 a+(?C)(*FAIL)
4635
4636 A match with the string "aaaa" always fails, but the callout is taken
4637 before each backtrack happens (in this example, 10 times).
4638
4639 Verbs that act after backtracking
4640
4641 The following verbs do nothing when they are encountered. Matching con-
4642 tinues with what follows, but if there is no subsequent match, a fail-
4643 ure is forced. The verbs differ in exactly what kind of failure
4644 occurs.
4645
4646 (*COMMIT)
4647
4648 This verb causes the whole match to fail outright if the rest of the
4649 pattern does not match. Even if the pattern is unanchored, no further
4650 attempts to find a match by advancing the start point take place. Once
4651 (*COMMIT) has been passed, pcre_exec() is committed to finding a match
4652 at the current starting point, or not at all. For example:
4653
4654 a+(*COMMIT)b
4655
4656 This matches "xxaab" but not "aacaab". It can be thought of as a kind
4657 of dynamic anchor, or "I've started, so I must finish."
4658
4659 (*PRUNE)
4660
4661 This verb causes the match to fail at the current position if the rest
4662 of the pattern does not match. If the pattern is unanchored, the normal
4663 "bumpalong" advance to the next starting character then happens. Back-
4664 tracking can occur as usual to the left of (*PRUNE), or when matching
4665 to the right of (*PRUNE), but if there is no match to the right, back-
4666 tracking cannot cross (*PRUNE). In simple cases, the use of (*PRUNE)
4667 is just an alternative to an atomic group or possessive quantifier, but
4668 there are some uses of (*PRUNE) that cannot be expressed in any other
4669 way.
4670
4671 (*SKIP)
4672
4673 This verb is like (*PRUNE), except that if the pattern is unanchored,
4674 the "bumpalong" advance is not to the next character, but to the posi-
4675 tion in the subject where (*SKIP) was encountered. (*SKIP) signifies
4676 that whatever text was matched leading up to it cannot be part of a
4677 successful match. Consider:
4678
4679 a+(*SKIP)b
4680
4681 If the subject is "aaaac...", after the first match attempt fails
4682 (starting at the first character in the string), the starting point
4683 skips on to start the next attempt at "c". Note that a possessive quan-
4684 tifer does not have the same effect in this example; although it would
4685 suppress backtracking during the first match attempt, the second
4686 attempt would start at the second character instead of skipping on to
4687 "c".
4688
4689 (*THEN)
4690
4691 This verb causes a skip to the next alternation if the rest of the pat-
4692 tern does not match. That is, it cancels pending backtracking, but only
4693 within the current alternation. Its name comes from the observation
4694 that it can be used for a pattern-based if-then-else block:
4695
4696 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
4697
4698 If the COND1 pattern matches, FOO is tried (and possibly further items
4699 after the end of the group if FOO succeeds); on failure the matcher
4700 skips to the second alternative and tries COND2, without backtracking
4701 into COND1. If (*THEN) is used outside of any alternation, it acts
4702 exactly like (*PRUNE).
4703
4704
4705 SEE ALSO
4706
4707 pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
4708
4709
4710 AUTHOR
4711
4712 Philip Hazel
4713 University Computing Service
4714 Cambridge CB2 3QH, England.
4715
4716
4717 REVISION
4718
4719 Last updated: 09 August 2007
4720 Copyright (c) 1997-2007 University of Cambridge.
4721 ------------------------------------------------------------------------------
4722
4723
4724 PCRESYNTAX(3) PCRESYNTAX(3)
4725
4726
4727 NAME
4728 PCRE - Perl-compatible regular expressions
4729
4730
4731 PCRE REGULAR EXPRESSION SYNTAX SUMMARY
4732
4733 The full syntax and semantics of the regular expressions that are sup-
4734 ported by PCRE are described in the pcrepattern documentation. This
4735 document contains just a quick-reference summary of the syntax.
4736
4737
4738 QUOTING
4739
4740 \x where x is non-alphanumeric is a literal x
4741 \Q...\E treat enclosed characters as literal
4742
4743
4744 CHARACTERS
4745
4746 \a alarm, that is, the BEL character (hex 07)
4747 \cx "control-x", where x is any character
4748 \e escape (hex 1B)
4749 \f formfeed (hex 0C)
4750 \n newline (hex 0A)
4751 \r carriage return (hex 0D)
4752 \t tab (hex 09)
4753 \ddd character with octal code ddd, or backreference
4754 \xhh character with hex code hh
4755 \x{hhh..} character with hex code hhh..
4756
4757
4758 CHARACTER TYPES
4759
4760 . any character except newline;
4761 in dotall mode, any character whatsoever
4762 \C one byte, even in UTF-8 mode (best avoided)
4763 \d a decimal digit
4764 \D a character that is not a decimal digit
4765 \h a horizontal whitespace character
4766 \H a character that is not a horizontal whitespace character
4767 \p{xx} a character with the xx property
4768 \P{xx} a character without the xx property
4769 \R a newline sequence
4770 \s a whitespace character
4771 \S a character that is not a whitespace character
4772 \v a vertical whitespace character
4773 \V a character that is not a vertical whitespace character
4774 \w a "word" character
4775 \W a "non-word" character
4776 \X an extended Unicode sequence
4777
4778 In PCRE, \d, \D, \s, \S, \w, and \W recognize only ASCII characters.
4779
4780
4781 GENERAL CATEGORY PROPERTY CODES FOR \p and \P
4782
4783 C Other
4784 Cc Control
4785 Cf Format
4786 Cn Unassigned
4787 Co Private use
4788 Cs Surrogate
4789
4790 L Letter
4791 Ll Lower case letter
4792 Lm Modifier letter
4793 Lo Other letter
4794 Lt Title case letter
4795 Lu Upper case letter
4796 L& Ll, Lu, or Lt
4797
4798 M Mark
4799 Mc Spacing mark
4800 Me Enclosing mark
4801 Mn Non-spacing mark
4802
4803 N Number
4804 Nd Decimal number
4805 Nl Letter number
4806 No Other number
4807
4808 P Punctuation
4809 Pc Connector punctuation
4810 Pd Dash punctuation
4811 Pe Close punctuation
4812 Pf Final punctuation
4813 Pi Initial punctuation
4814 Po Other punctuation
4815 Ps Open punctuation
4816
4817 S Symbol
4818 Sc Currency symbol
4819 Sk Modifier symbol
4820 Sm Mathematical symbol
4821 So Other symbol
4822
4823 Z Separator
4824 Zl Line separator
4825 Zp Paragraph separator
4826 Zs Space separator
4827
4828
4829 SCRIPT NAMES FOR \p AND \P
4830
4831 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
4832 Buhid, Canadian_Aboriginal, Cherokee, Common, Coptic, Cuneiform,
4833 Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian, Glagolitic,
4834 Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
4835 gana, Inherited, Kannada, Katakana, Kharoshthi, Khmer, Lao, Latin,
4836 Limbu, Linear_B, Malayalam, Mongolian, Myanmar, New_Tai_Lue, Nko,
4837 Ogham, Old_Italic, Old_Persian, Oriya, Osmanya, Phags_Pa, Phoenician,
4838 Runic, Shavian, Sinhala, Syloti_Nagri, Syriac, Tagalog, Tagbanwa,
4839 Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Yi.
4840
4841
4842 CHARACTER CLASSES
4843
4844 [...] positive character class
4845 [^...] negative character class
4846 [x-y] range (can be used for hex characters)
4847 [[:xxx:]] positive POSIX named set
4848 [[^:xxx:]] negative POSIX named set
4849
4850 alnum alphanumeric
4851 alpha alphabetic
4852 ascii 0-127
4853 blank space or tab
4854 cntrl control character
4855 digit decimal digit
4856 graph printing, excluding space
4857 lower lower case letter
4858 print printing, including space
4859 punct printing, excluding alphanumeric
4860 space whitespace
4861 upper upper case letter
4862 word same as \w
4863 xdigit hexadecimal digit
4864
4865 In PCRE, POSIX character set names recognize only ASCII characters. You
4866 can use \Q...\E inside a character class.
4867
4868
4869 QUANTIFIERS
4870
4871 ? 0 or 1, greedy
4872 ?+ 0 or 1, possessive
4873 ?? 0 or 1, lazy
4874 * 0 or more, greedy
4875 *+ 0 or more, possessive
4876 *? 0 or more, lazy
4877 + 1 or more, greedy
4878 ++ 1 or more, possessive
4879 +? 1 or more, lazy
4880 {n} exactly n
4881 {n,m} at least n, no more than m, greedy
4882 {n,m}+ at least n, no more than m, possessive
4883 {n,m}? at least n, no more than m, lazy
4884 {n,} n or more, greedy
4885 {n,}+ n or more, possessive
4886 {n,}? n or more, lazy
4887
4888
4889 ANCHORS AND SIMPLE ASSERTIONS
4890
4891 \b word boundary
4892 \B not a word boundary
4893 ^ start of subject
4894 also after internal newline in multiline mode
4895 \A start of subject
4896 $ end of subject
4897 also before newline at end of subject
4898 also before internal newline in multiline mode
4899 \Z end of subject
4900 also before newline at end of subject
4901 \z end of subject
4902 \G first matching position in subject
4903
4904
4905 MATCH POINT RESET
4906
4907 \K reset start of match
4908
4909
4910 ALTERNATION
4911
4912 expr|expr|expr...
4913
4914
4915 CAPTURING
4916
4917 (...) capturing group
4918 (?<name>...) named capturing group (Perl)
4919 (?'name'...) named capturing group (Perl)
4920 (?P<name>...) named capturing group (Python)
4921 (?:...) non-capturing group
4922 (?|...) non-capturing group; reset group numbers for
4923 capturing groups in each alternative
4924
4925
4926 ATOMIC GROUPS
4927
4928 (?>...) atomic, non-capturing group
4929
4930
4931 COMMENT
4932
4933 (?#....) comment (not nestable)
4934
4935
4936 OPTION SETTING
4937
4938 (?i) caseless
4939 (?J) allow duplicate names
4940 (?m) multiline
4941 (?s) single line (dotall)
4942 (?U) default ungreedy (lazy)
4943 (?x) extended (ignore white space)
4944 (?-...) unset option(s)
4945
4946
4947 LOOKAHEAD AND LOOKBEHIND ASSERTIONS
4948
4949 (?=...) positive look ahead
4950 (?!...) negative look ahead
4951 (?<=...) positive look behind
4952 (?<!...) negative look behind
4953
4954 Each top-level branch of a look behind must be of a fixed length.
4955
4956
4957 BACKREFERENCES
4958
4959 \n reference by number (can be ambiguous)
4960 \gn reference by number
4961 \g{n} reference by number
4962 \g{-n} relative reference by number
4963 \k<name> reference by name (Perl)
4964 \k'name' reference by name (Perl)
4965 \g{name} reference by name (Perl)
4966 \k{name} reference by name (.NET)
4967 (?P=name) reference by name (Python)
4968
4969
4970 SUBROUTINE REFERENCES (POSSIBLY RECURSIVE)
4971
4972 (?R) recurse whole pattern
4973 (?n) call subpattern by absolute number
4974 (?+n) call subpattern by relative number
4975 (?-n) call subpattern by relative number
4976 (?&name) call subpattern by name (Perl)
4977 (?P>name) call subpattern by name (Python)
4978
4979
4980 CONDITIONAL PATTERNS
4981
4982 (?(condition)yes-pattern)
4983 (?(condition)yes-pattern|no-pattern)
4984
4985 (?(n)... absolute reference condition
4986 (?(+n)... relative reference condition
4987 (?(-n)... relative reference condition
4988 (?(<name>)... named reference condition (Perl)
4989 (?('name')... named reference condition (Perl)
4990 (?(name)... named reference condition (PCRE)
4991 (?(R)... overall recursion condition
4992 (?(Rn)... specific group recursion condition
4993 (?(R&name)... specific recursion condition
4994 (?(DEFINE)... define subpattern for reference
4995 (?(assert)... assertion condition
4996
4997
4998 BACKTRACKING CONTROL
4999
5000 The following act immediately they are reached:
5001
5002 (*ACCEPT) force successful match
5003 (*FAIL) force backtrack; synonym (*F)
5004
5005 The following act only when a subsequent match failure causes a back-
5006 track to reach them. They all force a match failure, but they differ in
5007 what happens afterwards. Those that advance the start-of-match point do
5008 so only if the pattern is not anchored.
5009
5010 (*COMMIT) overall failure, no advance of starting point
5011 (*PRUNE) advance to next starting character
5012 (*SKIP) advance start to current matching position
5013 (*THEN) local failure, backtrack to next alternation
5014
5015
5016 CALLOUTS
5017
5018 (?C) callout
5019 (?Cn) callout with data n
5020
5021
5022 SEE ALSO
5023
5024 pcrepattern(3), pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
5025
5026
5027 AUTHOR
5028
5029 Philip Hazel
5030 University Computing Service
5031 Cambridge CB2 3QH, England.
5032
5033
5034 REVISION
5035
5036 Last updated: 08 August 2007
5037 Copyright (c) 1997-2007 University of Cambridge.
5038 ------------------------------------------------------------------------------
5039
5040
5041 PCREPARTIAL(3) PCREPARTIAL(3)
5042
5043
5044 NAME
5045 PCRE - Perl-compatible regular expressions
5046
5047
5048 PARTIAL MATCHING IN PCRE
5049
5050 In normal use of PCRE, if the subject string that is passed to
5051 pcre_exec() or pcre_dfa_exec() matches as far as it goes, but is too
5052 short to match the entire pattern, PCRE_ERROR_NOMATCH is returned.
5053 There are circumstances where it might be helpful to distinguish this
5054 case from other cases in which there is no match.
5055
5056 Consider, for example, an application where a human is required to type
5057 in data for a field with specific formatting requirements. An example
5058 might be a date in the form ddmmmyy, defined by this pattern:
5059
5060 ^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$
5061
5062 If the application sees the user's keystrokes one by one, and can check
5063 that what has been typed so far is potentially valid, it is able to
5064 raise an error as soon as a mistake is made, possibly beeping and not
5065 reflecting the character that has been typed. This immediate feedback
5066 is likely to be a better user interface than a check that is delayed
5067 until the entire string has been entered.
5068
5069 PCRE supports the concept of partial matching by means of the PCRE_PAR-
5070 TIAL option, which can be set when calling pcre_exec() or
5071 pcre_dfa_exec(). When this flag is set for pcre_exec(), the return code
5072 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if at any time
5073 during the matching process the last part of the subject string matched
5074 part of the pattern. Unfortunately, for non-anchored matching, it is
5075 not possible to obtain the position of the start of the partial match.
5076 No captured data is set when PCRE_ERROR_PARTIAL is returned.
5077
5078 When PCRE_PARTIAL is set for pcre_dfa_exec(), the return code
5079 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end of
5080 the subject is reached, there have been no complete matches, but there
5081 is still at least one matching possibility. The portion of the string
5082 that provided the partial match is set as the first matching string.
5083
5084 Using PCRE_PARTIAL disables one of PCRE's optimizations. PCRE remembers
5085 the last literal byte in a pattern, and abandons matching immediately
5086 if such a byte is not present in the subject string. This optimization
5087 cannot be used for a subject string that might match only partially.
5088
5089
5090 RESTRICTED PATTERNS FOR PCRE_PARTIAL
5091
5092 Because of the way certain internal optimizations are implemented in
5093 the pcre_exec() function, the PCRE_PARTIAL option cannot be used with
5094 all patterns. These restrictions do not apply when pcre_dfa_exec() is
5095 used. For pcre_exec(), repeated single characters such as
5096
5097 a{2,4}
5098
5099 and repeated single metasequences such as
5100
5101 \d+
5102
5103 are not permitted if the maximum number of occurrences is greater than
5104 one. Optional items such as \d? (where the maximum is one) are permit-
5105 ted. Quantifiers with any values are permitted after parentheses, so
5106 the invalid examples above can be coded thus:
5107
5108 (a){2,4}
5109 (\d)+
5110
5111 These constructions run more slowly, but for the kinds of application
5112 that are envisaged for this facility, this is not felt to be a major
5113 restriction.
5114
5115 If PCRE_PARTIAL is set for a pattern that does not conform to the
5116 restrictions, pcre_exec() returns the error code PCRE_ERROR_BADPARTIAL
5117 (-13). You can use the PCRE_INFO_OKPARTIAL call to pcre_fullinfo() to
5118 find out if a compiled pattern can be used for partial matching.
5119
5120
5121 EXAMPLE OF PARTIAL MATCHING USING PCRETEST
5122
5123 If the escape sequence \P is present in a pcretest data line, the
5124 PCRE_PARTIAL flag is used for the match. Here is a run of pcretest that
5125 uses the date example quoted above:
5126
5127 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5128 data> 25jun04\P
5129 0: 25jun04
5130 1: jun
5131 data> 25dec3\P
5132 Partial match
5133 data> 3ju\P
5134 Partial match
5135 data> 3juj\P
5136 No match
5137 data> j\P
5138 No match
5139
5140 The first data string is matched completely, so pcretest shows the
5141 matched substrings. The remaining four strings do not match the com-
5142 plete pattern, but the first two are partial matches. The same test,
5143 using pcre_dfa_exec() matching (by means of the \D escape sequence),
5144 produces the following output:
5145
5146 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5147 data> 25jun04\P\D
5148 0: 25jun04
5149 data> 23dec3\P\D
5150 Partial match: 23dec3
5151 data> 3ju\P\D
5152 Partial match: 3ju
5153 data> 3juj\P\D
5154 No match
5155 data> j\P\D
5156 No match
5157
5158 Notice that in this case the portion of the string that was matched is
5159 made available.
5160
5161
5162 MULTI-SEGMENT MATCHING WITH pcre_dfa_exec()
5163
5164 When a partial match has been found using pcre_dfa_exec(), it is possi-
5165 ble to continue the match by providing additional subject data and
5166 calling pcre_dfa_exec() again with the same compiled regular expres-
5167 sion, this time setting the PCRE_DFA_RESTART option. You must also pass
5168 the same working space as before, because this is where details of the
5169 previous partial match are stored. Here is an example using pcretest,
5170 using the \R escape sequence to set the PCRE_DFA_RESTART option (\P and
5171 \D are as above):
5172
5173 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
5174 data> 23ja\P\D
5175 Partial match: 23ja
5176 data> n05\R\D
5177 0: n05
5178
5179 The first call has "23ja" as the subject, and requests partial match-
5180 ing; the second call has "n05" as the subject for the continued
5181 (restarted) match. Notice that when the match is complete, only the
5182 last part is shown; PCRE does not retain the previously partially-
5183 matched string. It is up to the calling program to do that if it needs
5184 to.
5185
5186 You can set PCRE_PARTIAL with PCRE_DFA_RESTART to continue partial
5187 matching over multiple segments. This facility can be used to pass very
5188 long subject strings to pcre_dfa_exec(). However, some care is needed
5189 for certain types of pattern.
5190
5191 1. If the pattern contains tests for the beginning or end of a line,
5192 you need to pass the PCRE_NOTBOL or PCRE_NOTEOL options, as appropri-
5193 ate, when the subject string for any call does not contain the begin-
5194 ning or end of a line.
5195
5196 2. If the pattern contains backward assertions (including \b or \B),
5197 you need to arrange for some overlap in the subject strings to allow
5198 for this. For example, you could pass the subject in chunks that are
5199 500 bytes long, but in a buffer of 700 bytes, with the starting offset
5200 set to 200 and the previous 200 bytes at the start of the buffer.
5201
5202 3. Matching a subject string that is split into multiple segments does
5203 not always produce exactly the same result as matching over one single
5204 long string. The difference arises when there are multiple matching
5205 possibilities, because a partial match result is given only when there
5206 are no completed matches in a call to pcre_dfa_exec(). This means that
5207 as soon as the shortest match has been found, continuation to a new
5208 subject segment is no longer possible. Consider this pcretest example:
5209
5210 re> /dog(sbody)?/
5211 data> do\P\D
5212 Partial match: do
5213 data> gsb\R\P\D
5214 0: g
5215 data> dogsbody\D
5216 0: dogsbody
5217 1: dog
5218
5219 The pattern matches the words "dog" or "dogsbody". When the subject is
5220 presented in several parts ("do" and "gsb" being the first two) the
5221 match stops when "dog" has been found, and it is not possible to con-
5222 tinue. On the other hand, if "dogsbody" is presented as a single
5223 string, both matches are found.
5224
5225 Because of this phenomenon, it does not usually make sense to end a
5226 pattern that is going to be matched in this way with a variable repeat.
5227
5228 4. Patterns that contain alternatives at the top level which do not all
5229 start with the same pattern item may not work as expected. For example,
5230 consider this pattern:
5231
5232 1234|3789
5233
5234 If the first part of the subject is "ABC123", a partial match of the
5235 first alternative is found at offset 3. There is no partial match for
5236 the second alternative, because such a match does not start at the same
5237 point in the subject string. Attempting to continue with the string
5238 "789" does not yield a match because only those alternatives that match
5239 at one point in the subject are remembered. The problem arises because
5240 the start of the second alternative matches within the first alterna-
5241 tive. There is no problem with anchored patterns or patterns such as:
5242
5243 1234|ABCD
5244
5245 where no string can be a partial match for both alternatives.
5246
5247
5248 AUTHOR
5249
5250 Philip Hazel
5251 University Computing Service
5252 Cambridge CB2 3QH, England.
5253
5254
5255 REVISION
5256
5257 Last updated: 04 June 2007
5258 Copyright (c) 1997-2007 University of Cambridge.
5259 ------------------------------------------------------------------------------
5260
5261
5262 PCREPRECOMPILE(3) PCREPRECOMPILE(3)
5263
5264
5265 NAME
5266 PCRE - Perl-compatible regular expressions
5267
5268
5269 SAVING AND RE-USING PRECOMPILED PCRE PATTERNS
5270
5271 If you are running an application that uses a large number of regular
5272 expression patterns, it may be useful to store them in a precompiled
5273 form instead of having to compile them every time the application is
5274 run. If you are not using any private character tables (see the
5275 pcre_maketables() documentation), this is relatively straightforward.
5276 If you are using private tables, it is a little bit more complicated.
5277
5278 If you save compiled patterns to a file, you can copy them to a differ-
5279 ent host and run them there. This works even if the new host has the
5280 opposite endianness to the one on which the patterns were compiled.
5281 There may be a small performance penalty, but it should be insignifi-
5282 cant. However, compiling regular expressions with one version of PCRE
5283 for use with a different version is not guaranteed to work and may
5284 cause crashes.
5285
5286
5287 SAVING A COMPILED PATTERN
5288 The value returned by pcre_compile() points to a single block of memory
5289 that holds the compiled pattern and associated data. You can find the
5290 length of this block in bytes by calling pcre_fullinfo() with an argu-
5291 ment of PCRE_INFO_SIZE. You can then save the data in any appropriate
5292 manner. Here is sample code that compiles a pattern and writes it to a
5293 file. It assumes that the variable fd refers to a file that is open for
5294 output:
5295
5296 int erroroffset, rc, size;
5297 char *error;
5298 pcre *re;
5299
5300 re = pcre_compile("my pattern", 0, &error, &erroroffset, NULL);
5301 if (re == NULL) { ... handle errors ... }
5302 rc = pcre_fullinfo(re, NULL, PCRE_INFO_SIZE, &size);
5303 if (rc < 0) { ... handle errors ... }
5304 rc = fwrite(re, 1, size, fd);
5305 if (rc != size) { ... handle errors ... }
5306
5307 In this example, the bytes that comprise the compiled pattern are
5308 copied exactly. Note that this is binary data that may contain any of
5309 the 256 possible byte values. On systems that make a distinction
5310 between binary and non-binary data, be sure that the file is opened for
5311 binary output.
5312
5313 If you want to write more than one pattern to a file, you will have to
5314 devise a way of separating them. For binary data, preceding each pat-
5315 tern with its length is probably the most straightforward approach.
5316 Another possibility is to write out the data in hexadecimal instead of
5317 binary, one pattern to a line.
5318
5319 Saving compiled patterns in a file is only one possible way of storing
5320 them for later use. They could equally well be saved in a database, or
5321 in the memory of some daemon process that passes them via sockets to
5322 the processes that want them.
5323
5324 If the pattern has been studied, it is also possible to save the study
5325 data in a similar way to the compiled pattern itself. When studying
5326 generates additional information, pcre_study() returns a pointer to a
5327 pcre_extra data block. Its format is defined in the section on matching
5328 a pattern in the pcreapi documentation. The study_data field points to
5329 the binary study data, and this is what you must save (not the
5330 pcre_extra block itself). The length of the study data can be obtained
5331 by calling pcre_fullinfo() with an argument of PCRE_INFO_STUDYSIZE.
5332 Remember to check that pcre_study() did return a non-NULL value before
5333 trying to save the study data.
5334
5335
5336 RE-USING A PRECOMPILED PATTERN
5337
5338 Re-using a precompiled pattern is straightforward. Having reloaded it
5339 into main memory, you pass its pointer to pcre_exec() or
5340 pcre_dfa_exec() in the usual way. This should work even on another
5341 host, and even if that host has the opposite endianness to the one
5342 where the pattern was compiled.
5343
5344 However, if you passed a pointer to custom character tables when the
5345 pattern was compiled (the tableptr argument of pcre_compile()), you
5346 must now pass a similar pointer to pcre_exec() or pcre_dfa_exec(),
5347 because the value saved with the compiled pattern will obviously be
5348 nonsense. A field in a pcre_extra() block is used to pass this data, as
5349 described in the section on matching a pattern in the pcreapi documen-
5350 tation.
5351
5352 If you did not provide custom character tables when the pattern was
5353 compiled, the pointer in the compiled pattern is NULL, which causes
5354 pcre_exec() to use PCRE's internal tables. Thus, you do not need to
5355 take any special action at run time in this case.
5356
5357 If you saved study data with the compiled pattern, you need to create
5358 your own pcre_extra data block and set the study_data field to point to
5359 the reloaded study data. You must also set the PCRE_EXTRA_STUDY_DATA
5360 bit in the flags field to indicate that study data is present. Then
5361 pass the pcre_extra block to pcre_exec() or pcre_dfa_exec() in the
5362 usual way.
5363
5364
5365 COMPATIBILITY WITH DIFFERENT PCRE RELEASES
5366
5367 In general, it is safest to recompile all saved patterns when you
5368 update to a new PCRE release, though not all updates actually require
5369 this. Recompiling is definitely needed for release 7.2.
5370
5371
5372 AUTHOR
5373
5374 Philip Hazel
5375 University Computing Service
5376 Cambridge CB2 3QH, England.
5377
5378
5379 REVISION
5380
5381 Last updated: 13 June 2007
5382 Copyright (c) 1997-2007 University of Cambridge.
5383 ------------------------------------------------------------------------------
5384
5385
5386 PCREPERFORM(3) PCREPERFORM(3)
5387
5388
5389 NAME
5390 PCRE - Perl-compatible regular expressions
5391
5392
5393 PCRE PERFORMANCE
5394
5395 Two aspects of performance are discussed below: memory usage and pro-
5396 cessing time. The way you express your pattern as a regular expression
5397 can affect both of them.
5398
5399
5400 MEMORY USAGE
5401
5402 Patterns are compiled by PCRE into a reasonably efficient byte code, so
5403 that most simple patterns do not use much memory. However, there is one
5404 case where memory usage can be unexpectedly large. When a parenthesized
5405 subpattern has a quantifier with a minimum greater than 1 and/or a lim-
5406 ited maximum, the whole subpattern is repeated in the compiled code.
5407 For example, the pattern
5408
5409 (abc|def){2,4}
5410
5411 is compiled as if it were
5412
5413 (abc|def)(abc|def)((abc|def)(abc|def)?)?
5414
5415 (Technical aside: It is done this way so that backtrack points within
5416 each of the repetitions can be independently maintained.)
5417
5418 For regular expressions whose quantifiers use only small numbers, this
5419 is not usually a problem. However, if the numbers are large, and par-
5420 ticularly if such repetitions are nested, the memory usage can become
5421 an embarrassment. For example, the very simple pattern
5422
5423 ((ab){1,1000}c){1,3}
5424
5425 uses 51K bytes when compiled. When PCRE is compiled with its default
5426 internal pointer size of two bytes, the size limit on a compiled pat-
5427 tern is 64K, and this is reached with the above pattern if the outer
5428 repetition is increased from 3 to 4. PCRE can be compiled to use larger
5429 internal pointers and thus handle larger compiled patterns, but it is
5430 better to try to rewrite your pattern to use less memory if you can.
5431
5432 One way of reducing the memory usage for such patterns is to make use
5433 of PCRE's "subroutine" facility. Re-writing the above pattern as
5434
5435 ((ab)(?2){0,999}c)(?1){0,2}
5436
5437 reduces the memory requirements to 18K, and indeed it remains under 20K
5438 even with the outer repetition increased to 100. However, this pattern
5439 is not exactly equivalent, because the "subroutine" calls are treated
5440 as atomic groups into which there can be no backtracking if there is a
5441 subsequent matching failure. Therefore, PCRE cannot do this kind of
5442 rewriting automatically. Furthermore, there is a noticeable loss of
5443 speed when executing the modified pattern. Nevertheless, if the atomic
5444 grouping is not a problem and the loss of speed is acceptable, this
5445 kind of rewriting will allow you to process patterns that PCRE cannot
5446 otherwise handle.
5447
5448
5449 PROCESSING TIME
5450
5451 Certain items in regular expression patterns are processed more effi-
5452 ciently than others. It is more efficient to use a character class like
5453 [aeiou] than a set of single-character alternatives such as
5454 (a|e|i|o|u). In general, the simplest construction that provides the
5455 required behaviour is usually the most efficient. Jeffrey Friedl's book
5456 contains a lot of useful general discussion about optimizing regular
5457 expressions for efficient performance. This document contains a few
5458 observations about PCRE.
5459
5460 Using Unicode character properties (the \p, \P, and \X escapes) is
5461 slow, because PCRE has to scan a structure that contains data for over
5462 fifteen thousand characters whenever it needs a character's property.
5463 If you can find an alternative pattern that does not use character
5464 properties, it will probably be faster.
5465
5466 When a pattern begins with .* not in parentheses, or in parentheses
5467 that are not the subject of a backreference, and the PCRE_DOTALL option
5468 is set, the pattern is implicitly anchored by PCRE, since it can match
5469 only at the start of a subject string. However, if PCRE_DOTALL is not
5470 set, PCRE cannot make this optimization, because the . metacharacter
5471 does not then match a newline, and if the subject string contains new-
5472 lines, the pattern may match from the character immediately following
5473 one of them instead of from the very start. For example, the pattern
5474
5475 .*second
5476
5477 matches the subject "first\nand second" (where \n stands for a newline
5478 character), with the match starting at the seventh character. In order
5479 to do this, PCRE has to retry the match starting after every newline in
5480 the subject.
5481
5482 If you are using such a pattern with subject strings that do not con-
5483 tain newlines, the best performance is obtained by setting PCRE_DOTALL,
5484 or starting the pattern with ^.* or ^.*? to indicate explicit anchor-
5485 ing. That saves PCRE from having to scan along the subject looking for
5486 a newline to restart at.
5487
5488 Beware of patterns that contain nested indefinite repeats. These can
5489 take a long time to run when applied to a string that does not match.
5490 Consider the pattern fragment
5491
5492 ^(a+)*
5493
5494 This can match "aaaa" in 16 different ways, and this number increases
5495 very rapidly as the string gets longer. (The * repeat can match 0, 1,
5496 2, 3, or 4 times, and for each of those cases other than 0 or 4, the +
5497 repeats can match different numbers of times.) When the remainder of
5498 the pattern is such that the entire match is going to fail, PCRE has in
5499 principle to try every possible variation, and this can take an
5500 extremely long time, even for relatively short strings.
5501
5502 An optimization catches some of the more simple cases such as
5503
5504 (a+)*b
5505
5506 where a literal character follows. Before embarking on the standard
5507 matching procedure, PCRE checks that there is a "b" later in the sub-
5508 ject string, and if there is not, it fails the match immediately. How-
5509 ever, when there is no following literal this optimization cannot be
5510 used. You can see the difference by comparing the behaviour of
5511
5512 (a+)*\d
5513
5514 with the pattern above. The former gives a failure almost instantly
5515 when applied to a whole line of "a" characters, whereas the latter
5516 takes an appreciable time with strings longer than about 20 characters.
5517
5518 In many cases, the solution to this kind of performance issue is to use
5519 an atomic group or a possessive quantifier.
5520
5521
5522 AUTHOR
5523
5524 Philip Hazel
5525 University Computing Service
5526 Cambridge CB2 3QH, England.
5527
5528
5529 REVISION
5530
5531 Last updated: 06 March 2007
5532 Copyright (c) 1997-2007 University of Cambridge.
5533 ------------------------------------------------------------------------------
5534
5535
5536 PCREPOSIX(3) PCREPOSIX(3)
5537
5538
5539 NAME
5540 PCRE - Perl-compatible regular expressions.
5541
5542
5543 SYNOPSIS OF POSIX API
5544
5545 #include <pcreposix.h>
5546
5547 int regcomp(regex_t *preg, const char *pattern,
5548 int cflags);
5549
5550 int regexec(regex_t *preg, const char *string,
5551 size_t nmatch, regmatch_t pmatch[], int eflags);
5552
5553 size_t regerror(int errcode, const regex_t *preg,
5554 char *errbuf, size_t errbuf_size);
5555
5556 void regfree(regex_t *preg);
5557
5558
5559 DESCRIPTION
5560
5561 This set of functions provides a POSIX-style API to the PCRE regular
5562 expression package. See the pcreapi documentation for a description of
5563 PCRE's native API, which contains much additional functionality.
5564
5565 The functions described here are just wrapper functions that ultimately
5566 call the PCRE native API. Their prototypes are defined in the
5567 pcreposix.h header file, and on Unix systems the library itself is
5568 called pcreposix.a, so can be accessed by adding -lpcreposix to the
5569 command for linking an application that uses them. Because the POSIX
5570 functions call the native ones, it is also necessary to add -lpcre.
5571
5572 I have implemented only those option bits that can be reasonably mapped
5573 to PCRE native options. In addition, the option REG_EXTENDED is defined
5574 with the value zero. This has no effect, but since programs that are
5575 written to the POSIX interface often use it, this makes it easier to
5576 slot in PCRE as a replacement library. Other POSIX options are not even
5577 defined.
5578
5579 When PCRE is called via these functions, it is only the API that is
5580 POSIX-like in style. The syntax and semantics of the regular expres-
5581 sions themselves are still those of Perl, subject to the setting of
5582 various PCRE options, as described below. "POSIX-like in style" means
5583 that the API approximates to the POSIX definition; it is not fully
5584 POSIX-compatible, and in multi-byte encoding domains it is probably
5585 even less compatible.
5586
5587 The header for these functions is supplied as pcreposix.h to avoid any
5588 potential clash with other POSIX libraries. It can, of course, be
5589 renamed or aliased as regex.h, which is the "correct" name. It provides
5590 two structure types, regex_t for compiled internal forms, and reg-
5591 match_t for returning captured substrings. It also defines some con-
5592 stants whose names start with "REG_"; these are used for setting
5593 options and identifying error codes.
5594
5595
5596 COMPILING A PATTERN
5597
5598 The function regcomp() is called to compile a pattern into an internal
5599 form. The pattern is a C string terminated by a binary zero, and is
5600 passed in the argument pattern. The preg argument is a pointer to a
5601 regex_t structure that is used as a base for storing information about
5602 the compiled regular expression.
5603
5604 The argument cflags is either zero, or contains one or more of the bits
5605 defined by the following macros:
5606
5607 REG_DOTALL
5608
5609 The PCRE_DOTALL option is set when the regular expression is passed for
5610 compilation to the native function. Note that REG_DOTALL is not part of
5611 the POSIX standard.
5612
5613 REG_ICASE
5614
5615 The PCRE_CASELESS option is set when the regular expression is passed
5616 for compilation to the native function.
5617
5618 REG_NEWLINE
5619
5620 The PCRE_MULTILINE option is set when the regular expression is passed
5621 for compilation to the native function. Note that this does not mimic
5622 the defined POSIX behaviour for REG_NEWLINE (see the following sec-
5623 tion).
5624
5625 REG_NOSUB
5626
5627 The PCRE_NO_AUTO_CAPTURE option is set when the regular expression is
5628 passed for compilation to the native function. In addition, when a pat-
5629 tern that is compiled with this flag is passed to regexec() for match-
5630 ing, the nmatch and pmatch arguments are ignored, and no captured
5631 strings are returned.
5632
5633 REG_UTF8
5634
5635 The PCRE_UTF8 option is set when the regular expression is passed for
5636 compilation to the native function. This causes the pattern itself and
5637 all data strings used for matching it to be treated as UTF-8 strings.
5638 Note that REG_UTF8 is not part of the POSIX standard.
5639
5640 In the absence of these flags, no options are passed to the native
5641 function. This means the the regex is compiled with PCRE default
5642 semantics. In particular, the way it handles newline characters in the
5643 subject string is the Perl way, not the POSIX way. Note that setting
5644 PCRE_MULTILINE has only some of the effects specified for REG_NEWLINE.
5645 It does not affect the way newlines are matched by . (they aren't) or
5646 by a negative class such as [^a] (they are).
5647
5648 The yield of regcomp() is zero on success, and non-zero otherwise. The
5649 preg structure is filled in on success, and one member of the structure
5650 is public: re_nsub contains the number of capturing subpatterns in the
5651 regular expression. Various error codes are defined in the header file.
5652
5653
5654 MATCHING NEWLINE CHARACTERS
5655
5656 This area is not simple, because POSIX and Perl take different views of
5657 things. It is not possible to get PCRE to obey POSIX semantics, but
5658 then PCRE was never intended to be a POSIX engine. The following table
5659 lists the different possibilities for matching newline characters in
5660 PCRE:
5661
5662 Default Change with
5663
5664 . matches newline no PCRE_DOTALL
5665 newline matches [^a] yes not changeable
5666 $ matches \n at end yes PCRE_DOLLARENDONLY
5667 $ matches \n in middle no PCRE_MULTILINE
5668 ^ matches \n in middle no PCRE_MULTILINE
5669
5670 This is the equivalent table for POSIX:
5671
5672 Default Change with
5673
5674 . matches newline yes REG_NEWLINE
5675 newline matches [^a] yes REG_NEWLINE
5676 $ matches \n at end no REG_NEWLINE
5677 $ matches \n in middle no REG_NEWLINE
5678 ^ matches \n in middle no REG_NEWLINE
5679
5680 PCRE's behaviour is the same as Perl's, except that there is no equiva-
5681 lent for PCRE_DOLLAR_ENDONLY in Perl. In both PCRE and Perl, there is
5682 no way to stop newline from matching [^a].
5683
5684 The default POSIX newline handling can be obtained by setting
5685 PCRE_DOTALL and PCRE_DOLLAR_ENDONLY, but there is no way to make PCRE
5686 behave exactly as for the REG_NEWLINE action.
5687
5688
5689 MATCHING A PATTERN
5690
5691 The function regexec() is called to match a compiled pattern preg
5692 against a given string, which is terminated by a zero byte, subject to
5693 the options in eflags. These can be:
5694
5695 REG_NOTBOL
5696
5697 The PCRE_NOTBOL option is set when calling the underlying PCRE matching
5698 function.
5699
5700 REG_NOTEOL
5701
5702 The PCRE_NOTEOL option is set when calling the underlying PCRE matching
5703 function.
5704
5705 If the pattern was compiled with the REG_NOSUB flag, no data about any
5706 matched strings is returned. The nmatch and pmatch arguments of
5707 regexec() are ignored.
5708
5709 Otherwise,the portion of the string that was matched, and also any cap-
5710 tured substrings, are returned via the pmatch argument, which points to
5711 an array of nmatch structures of type regmatch_t, containing the mem-
5712 bers rm_so and rm_eo. These contain the offset to the first character
5713 of each substring and the offset to the first character after the end
5714 of each substring, respectively. The 0th element of the vector relates
5715 to the entire portion of string that was matched; subsequent elements
5716 relate to the capturing subpatterns of the regular expression. Unused
5717 entries in the array have both structure members set to -1.
5718
5719 A successful match yields a zero return; various error codes are
5720 defined in the header file, of which REG_NOMATCH is the "expected"
5721 failure code.
5722
5723
5724 ERROR MESSAGES
5725
5726 The regerror() function maps a non-zero errorcode from either regcomp()
5727 or regexec() to a printable message. If preg is not NULL, the error
5728 should have arisen from the use of that structure. A message terminated
5729 by a binary zero is placed in errbuf. The length of the message,
5730 including the zero, is limited to errbuf_size. The yield of the func-
5731 tion is the size of buffer needed to hold the whole message.
5732
5733
5734 MEMORY USAGE
5735
5736 Compiling a regular expression causes memory to be allocated and asso-
5737 ciated with the preg structure. The function regfree() frees all such
5738 memory, after which preg may no longer be used as a compiled expres-
5739 sion.
5740
5741
5742 AUTHOR
5743
5744 Philip Hazel
5745 University Computing Service
5746 Cambridge CB2 3QH, England.
5747
5748
5749 REVISION
5750
5751 Last updated: 06 March 2007
5752 Copyright (c) 1997-2007 University of Cambridge.
5753 ------------------------------------------------------------------------------
5754
5755
5756 PCRECPP(3) PCRECPP(3)
5757
5758
5759 NAME
5760 PCRE - Perl-compatible regular expressions.
5761
5762
5763 SYNOPSIS OF C++ WRAPPER
5764
5765 #include <pcrecpp.h>
5766
5767
5768 DESCRIPTION
5769
5770 The C++ wrapper for PCRE was provided by Google Inc. Some additional
5771 functionality was added by Giuseppe Maxia. This brief man page was con-
5772 structed from the notes in the pcrecpp.h file, which should be con-
5773 sulted for further details.
5774
5775
5776 MATCHING INTERFACE
5777
5778 The "FullMatch" operation checks that supplied text matches a supplied
5779 pattern exactly. If pointer arguments are supplied, it copies matched
5780 sub-strings that match sub-patterns into them.
5781
5782 Example: successful match
5783 pcrecpp::RE re("h.*o");
5784 re.FullMatch("hello");
5785
5786 Example: unsuccessful match (requires full match):
5787 pcrecpp::RE re("e");
5788 !re.FullMatch("hello");
5789
5790 Example: creating a temporary RE object:
5791 pcrecpp::RE("h.*o").FullMatch("hello");
5792
5793 You can pass in a "const char*" or a "string" for "text". The examples
5794 below tend to use a const char*. You can, as in the different examples
5795 above, store the RE object explicitly in a variable or use a temporary
5796 RE object. The examples below use one mode or the other arbitrarily.
5797 Either could correctly be used for any of these examples.
5798
5799 You must supply extra pointer arguments to extract matched subpieces.
5800
5801 Example: extracts "ruby" into "s" and 1234 into "i"
5802 int i;
5803 string s;
5804 pcrecpp::RE re("(\\w+):(\\d+)");
5805 re.FullMatch("ruby:1234", &s, &i);
5806
5807 Example: does not try to extract any extra sub-patterns
5808 re.FullMatch("ruby:1234", &s);
5809
5810 Example: does not try to extract into NULL
5811 re.FullMatch("ruby:1234", NULL, &i);
5812
5813 Example: integer overflow causes failure
5814 !re.FullMatch("ruby:1234567891234", NULL, &i);
5815
5816 Example: fails because there aren't enough sub-patterns:
5817 !pcrecpp::RE("\\w+:\\d+").FullMatch("ruby:1234", &s);
5818
5819 Example: fails because string cannot be stored in integer
5820 !pcrecpp::RE("(.*)").FullMatch("ruby", &i);
5821
5822 The provided pointer arguments can be pointers to any scalar numeric
5823 type, or one of:
5824
5825 string (matched piece is copied to string)
5826 StringPiece (StringPiece is mutated to point to matched piece)
5827 T (where "bool T::ParseFrom(const char*, int)" exists)
5828 NULL (the corresponding matched sub-pattern is not copied)
5829
5830 The function returns true iff all of the following conditions are sat-
5831 isfied:
5832
5833 a. "text" matches "pattern" exactly;
5834
5835 b. The number of matched sub-patterns is >= number of supplied
5836 pointers;
5837
5838 c. The "i"th argument has a suitable type for holding the
5839 string captured as the "i"th sub-pattern. If you pass in
5840 NULL for the "i"th argument, or pass fewer arguments than
5841 number of sub-patterns, "i"th captured sub-pattern is
5842 ignored.
5843
5844 CAVEAT: An optional sub-pattern that does not exist in the matched
5845 string is assigned the empty string. Therefore, the following will
5846 return false (because the empty string is not a valid number):
5847
5848 int number;
5849 pcrecpp::RE::FullMatch("abc", "[a-z]+(\\d+)?", &number);
5850
5851 The matching interface supports at most 16 arguments per call. If you
5852 need more, consider using the more general interface
5853 pcrecpp::RE::DoMatch. See pcrecpp.h for the signature for DoMatch.
5854
5855
5856 QUOTING METACHARACTERS
5857
5858 You can use the "QuoteMeta" operation to insert backslashes before all
5859 potentially meaningful characters in a string. The returned string,
5860 used as a regular expression, will exactly match the original string.
5861
5862 Example:
5863 string quoted = RE::QuoteMeta(unquoted);
5864
5865 Note that it's legal to escape a character even if it has no special
5866 meaning in a regular expression -- so this function does that. (This
5867 also makes it identical to the perl function of the same name; see
5868 "perldoc -f quotemeta".) For example, "1.5-2.0?" becomes
5869 "1\.5\-2\.0\?".
5870
5871
5872 PARTIAL MATCHES
5873
5874 You can use the "PartialMatch" operation when you want the pattern to
5875 match any substring of the text.
5876
5877 Example: simple search for a string:
5878 pcrecpp::RE("ell").PartialMatch("hello");
5879
5880 Example: find first number in a string:
5881 int number;
5882 pcrecpp::RE re("(\\d+)");
5883 re.PartialMatch("x*100 + 20", &number);
5884 assert(number == 100);
5885
5886
5887 UTF-8 AND THE MATCHING INTERFACE
5888
5889 By default, pattern and text are plain text, one byte per character.
5890 The UTF8 flag, passed to the constructor, causes both pattern and
5891 string to be treated as UTF-8 text, still a byte stream but potentially
5892 multiple bytes per character. In practice, the text is likelier to be
5893 UTF-8 than the pattern, but the match returned may depend on the UTF8
5894 flag, so always use it when matching UTF8 text. For example, "." will
5895 match one byte normally but with UTF8 set may match up to three bytes
5896 of a multi-byte character.
5897
5898 Example:
5899 pcrecpp::RE_Options options;
5900 options.set_utf8();
5901 pcrecpp::RE re(utf8_pattern, options);
5902 re.FullMatch(utf8_string);
5903
5904 Example: using the convenience function UTF8():
5905 pcrecpp::RE re(utf8_pattern, pcrecpp::UTF8());
5906 re.FullMatch(utf8_string);
5907
5908 NOTE: The UTF8 flag is ignored if pcre was not configured with the
5909 --enable-utf8 flag.
5910
5911
5912 PASSING MODIFIERS TO THE REGULAR EXPRESSION ENGINE
5913
5914 PCRE defines some modifiers to change the behavior of the regular
5915 expression engine. The C++ wrapper defines an auxiliary class,
5916 RE_Options, as a vehicle to pass such modifiers to a RE class. Cur-
5917 rently, the following modifiers are supported:
5918
5919 modifier description Perl corresponding
5920
5921 PCRE_CASELESS case insensitive match /i
5922 PCRE_MULTILINE multiple lines match /m
5923 PCRE_DOTALL dot matches newlines /s
5924 PCRE_DOLLAR_ENDONLY $ matches only at end N/A
5925 PCRE_EXTRA strict escape parsing N/A
5926 PCRE_EXTENDED ignore whitespaces /x
5927 PCRE_UTF8 handles UTF8 chars built-in
5928 PCRE_UNGREEDY reverses * and *? N/A
5929 PCRE_NO_AUTO_CAPTURE disables capturing parens N/A (*)
5930
5931 (*) Both Perl and PCRE allow non capturing parentheses by means of the
5932 "?:" modifier within the pattern itself. e.g. (?:ab|cd) does not cap-
5933 ture, while (ab|cd) does.
5934
5935 For a full account on how each modifier works, please check the PCRE
5936 API reference page.
5937
5938 For each modifier, there are two member functions whose name is made
5939 out of the modifier in lowercase, without the "PCRE_" prefix. For
5940 instance, PCRE_CASELESS is handled by
5941
5942 bool caseless()
5943
5944 which returns true if the modifier is set, and
5945
5946 RE_Options & set_caseless(bool)
5947
5948 which sets or unsets the modifier. Moreover, PCRE_EXTRA_MATCH_LIMIT can
5949 be accessed through the set_match_limit() and match_limit() member
5950 functions. Setting match_limit to a non-zero value will limit the exe-
5951 cution of pcre to keep it from doing bad things like blowing the stack
5952 or taking an eternity to return a result. A value of 5000 is good
5953 enough to stop stack blowup in a 2MB thread stack. Setting match_limit
5954 to zero disables match limiting. Alternatively, you can call
5955 match_limit_recursion() which uses PCRE_EXTRA_MATCH_LIMIT_RECURSION to
5956 limit how much PCRE recurses. match_limit() limits the number of
5957 matches PCRE does; match_limit_recursion() limits the depth of internal
5958 recursion, and therefore the amount of stack that is used.
5959
5960 Normally, to pass one or more modifiers to a RE class, you declare a
5961 RE_Options object, set the appropriate options, and pass this object to
5962 a RE constructor. Example:
5963
5964 RE_options opt;
5965 opt.set_caseless(true);
5966 if (RE("HELLO", opt).PartialMatch("hello world")) ...
5967
5968 RE_options has two constructors. The default constructor takes no argu-
5969 ments and creates a set of flags that are off by default. The optional
5970 parameter option_flags is to facilitate transfer of legacy code from C
5971 programs. This lets you do
5972
5973 RE(pattern,
5974 RE_Options(PCRE_CASELESS|PCRE_MULTILINE)).PartialMatch(str);
5975
5976 However, new code is better off doing
5977
5978 RE(pattern,
5979 RE_Options().set_caseless(true).set_multiline(true))
5980 .PartialMatch(str);
5981
5982 If you are going to pass one of the most used modifiers, there are some
5983 convenience functions that return a RE_Options class with the appropri-
5984 ate modifier already set: CASELESS(), UTF8(), MULTILINE(), DOTALL(),
5985 and EXTENDED().
5986
5987 If you need to set several options at once, and you don't want to go
5988 through the pains of declaring a RE_Options object and setting several
5989 options, there is a parallel method that give you such ability on the
5990 fly. You can concatenate several set_xxxxx() member functions, since
5991 each of them returns a reference to its class object. For example, to
5992 pass PCRE_CASELESS, PCRE_EXTENDED, and PCRE_MULTILINE to a RE with one
5993 statement, you may write:
5994
5995 RE(" ^ xyz \\s+ .* blah$",
5996 RE_Options()
5997 .set_caseless(true)
5998 .set_extended(true)
5999 .set_multiline(true)).PartialMatch(sometext);
6000
6001
6002 SCANNING TEXT INCREMENTALLY
6003
6004 The "Consume" operation may be useful if you want to repeatedly match
6005 regular expressions at the front of a string and skip over them as they
6006 match. This requires use of the "StringPiece" type, which represents a
6007 sub-range of a real string. Like RE, StringPiece is defined in the
6008 pcrecpp namespace.
6009
6010 Example: read lines of the form "var = value" from a string.
6011 string contents = ...; // Fill string somehow
6012 pcrecpp::StringPiece input(contents); // Wrap in a StringPiece
6013
6014 string var;
6015 int value;
6016 pcrecpp::RE re("(\\w+) = (\\d+)\n");
6017 while (re.Consume(&input, &var, &value)) {
6018 ...;
6019 }
6020
6021 Each successful call to "Consume" will set "var/value", and also
6022 advance "input" so it points past the matched text.
6023
6024 The "FindAndConsume" operation is similar to "Consume" but does not
6025 anchor your match at the beginning of the string. For example, you
6026 could extract all words from a string by repeatedly calling
6027
6028 pcrecpp::RE("(\\w+)").FindAndConsume(&input, &word)
6029
6030
6031 PARSING HEX/OCTAL/C-RADIX NUMBERS
6032
6033 By default, if you pass a pointer to a numeric value, the corresponding
6034 text is interpreted as a base-10 number. You can instead wrap the
6035 pointer with a call to one of the operators Hex(), Octal(), or CRadix()
6036 to interpret the text in another base. The CRadix operator interprets
6037 C-style "0" (base-8) and "0x" (base-16) prefixes, but defaults to
6038 base-10.
6039
6040 Example:
6041 int a, b, c, d;
6042 pcrecpp::RE re("(.*) (.*) (.*) (.*)");
6043 re.FullMatch("100 40 0100 0x40",
6044 pcrecpp::Octal(&a), pcrecpp::Hex(&b),
6045 pcrecpp::CRadix(&c), pcrecpp::CRadix(&d));
6046
6047 will leave 64 in a, b, c, and d.
6048
6049
6050 REPLACING PARTS OF STRINGS
6051
6052 You can replace the first match of "pattern" in "str" with "rewrite".
6053 Within "rewrite", backslash-escaped digits (\1 to \9) can be used to
6054 insert text matching corresponding parenthesized group from the pat-
6055 tern. \0 in "rewrite" refers to the entire matching text. For example:
6056
6057 string s = "yabba dabba doo";
6058 pcrecpp::RE("b+").Replace("d", &s);
6059
6060 will leave "s" containing "yada dabba doo". The result is true if the
6061 pattern matches and a replacement occurs, false otherwise.
6062
6063 GlobalReplace is like Replace except that it replaces all occurrences