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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 197 - (show annotations) (download)
Tue Jul 31 10:50:18 2007 UTC (7 years, 3 months ago) by ph10
File MIME type: text/plain
File size: 275917 byte(s)
Abolish the NULLWSLIMIT error at the expense of using more stack when an 
unlimited repeat could match an empty string. Also, doc tidies for a test 
release.

1 -----------------------------------------------------------------------------
2 This file contains a concatenation of the PCRE man pages, converted to plain
3 text format for ease of searching with a text editor, or for use on systems
4 that do not have a man page processor. The small individual files that give
5 synopses of each function in the library have not been included. 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.
49
50 Some features of PCRE can be included, excluded, or changed when the
51 library is built. The pcre_config() function makes it possible for a
52 client to discover which features are available. The features them-
53 selves are described in the pcrebuild page. Documentation about build-
54 ing PCRE for various operating systems can be found in the README file
55 in the source distribution.
56
57 The library contains a number of undocumented internal functions and
58 data tables that are used by more than one of the exported external
59 functions, but which are not intended for use by external callers.
60 Their names all begin with "_pcre_", which hopefully will not provoke
61 any name clashes. In some environments, it is possible to control which
62 external symbols are exported when a shared library is built, and in
63 these cases the undocumented symbols are not exported.
64
65
66 USER DOCUMENTATION
67
68 The user documentation for PCRE comprises a number of different sec-
69 tions. In the "man" format, each of these is a separate "man page". In
70 the HTML format, each is a separate page, linked from the index page.
71 In the plain text format, all the sections are concatenated, for ease
72 of searching. The sections are as follows:
73
74 pcre this document
75 pcre-config show PCRE installation configuration information
76 pcreapi details of PCRE's native C API
77 pcrebuild options for building PCRE
78 pcrecallout details of the callout feature
79 pcrecompat discussion of Perl compatibility
80 pcrecpp details of the C++ wrapper
81 pcregrep description of the pcregrep command
82 pcrematching discussion of the two matching algorithms
83 pcrepartial details of the partial matching facility
84 pcrepattern syntax and semantics of supported
85 regular expressions
86 pcreperform discussion of performance issues
87 pcreposix the POSIX-compatible C API
88 pcreprecompile details of saving and re-using precompiled patterns
89 pcresample discussion of the sample program
90 pcrestack discussion of stack usage
91 pcretest description of the pcretest testing command
92
93 In addition, in the "man" and HTML formats, there is a short page for
94 each C library function, listing its arguments and results.
95
96
97 LIMITATIONS
98
99 There are some size limitations in PCRE but it is hoped that they will
100 never in practice be relevant.
101
102 The maximum length of a compiled pattern is 65539 (sic) bytes if PCRE
103 is compiled with the default internal linkage size of 2. If you want to
104 process regular expressions that are truly enormous, you can compile
105 PCRE with an internal linkage size of 3 or 4 (see the README file in
106 the source distribution and the pcrebuild documentation for details).
107 In these cases the limit is substantially larger. However, the speed
108 of execution is slower.
109
110 All values in repeating quantifiers must be less than 65536. The maxi-
111 mum compiled length of subpattern with an explicit repeat count is
112 30000 bytes. The maximum number of capturing subpatterns is 65535.
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 If a non-capturing subpattern with an unlimited repetition quantifier
118 can match an empty string, there is a limit of 1000 on the number of
119 times it can be repeated while not matching an empty string - if it
120 does match an empty string, the loop is immediately broken.
121
122 The maximum length of name for a named subpattern is 32 characters, and
123 the maximum number of named subpatterns is 10000.
124
125 The maximum length of a subject string is the largest positive number
126 that an integer variable can hold. However, when using the traditional
127 matching function, PCRE uses recursion to handle subpatterns and indef-
128 inite repetition. This means that the available stack space may limit
129 the size of a subject string that can be processed by certain patterns.
130 For a discussion of stack issues, see the pcrestack documentation.
131
132
133 UTF-8 AND UNICODE PROPERTY SUPPORT
134
135 From release 3.3, PCRE has had some support for character strings
136 encoded in the UTF-8 format. For release 4.0 this was greatly extended
137 to cover most common requirements, and in release 5.0 additional sup-
138 port for Unicode general category properties was added.
139
140 In order process UTF-8 strings, you must build PCRE to include UTF-8
141 support in the code, and, in addition, you must call pcre_compile()
142 with the PCRE_UTF8 option flag. When you do this, both the pattern and
143 any subject strings that are matched against it are treated as UTF-8
144 strings instead of just strings of bytes.
145
146 If you compile PCRE with UTF-8 support, but do not use it at run time,
147 the library will be a bit bigger, but the additional run time overhead
148 is limited to testing the PCRE_UTF8 flag occasionally, so should not be
149 very big.
150
151 If PCRE is built with Unicode character property support (which implies
152 UTF-8 support), the escape sequences \p{..}, \P{..}, and \X are sup-
153 ported. The available properties that can be tested are limited to the
154 general category properties such as Lu for an upper case letter or Nd
155 for a decimal number, the Unicode script names such as Arabic or Han,
156 and the derived properties Any and L&. A full list is given in the
157 pcrepattern documentation. Only the short names for properties are sup-
158 ported. For example, \p{L} matches a letter. Its Perl synonym, \p{Let-
159 ter}, is not supported. Furthermore, in Perl, many properties may
160 optionally be prefixed by "Is", for compatibility with Perl 5.6. PCRE
161 does not support this.
162
163 The following comments apply when PCRE is running in UTF-8 mode:
164
165 1. When you set the PCRE_UTF8 flag, the strings passed as patterns and
166 subjects are checked for validity on entry to the relevant functions.
167 If an invalid UTF-8 string is passed, an error return is given. In some
168 situations, you may already know that your strings are valid, and
169 therefore want to skip these checks in order to improve performance. If
170 you set the PCRE_NO_UTF8_CHECK flag at compile time or at run time,
171 PCRE assumes that the pattern or subject it is given (respectively)
172 contains only valid UTF-8 codes. In this case, it does not diagnose an
173 invalid UTF-8 string. If you pass an invalid UTF-8 string to PCRE when
174 PCRE_NO_UTF8_CHECK is set, the results are undefined. Your program may
175 crash.
176
177 2. An unbraced hexadecimal escape sequence (such as \xb3) matches a
178 two-byte UTF-8 character if the value is greater than 127.
179
180 3. Octal numbers up to \777 are recognized, and match two-byte UTF-8
181 characters for values greater than \177.
182
183 4. Repeat quantifiers apply to complete UTF-8 characters, not to indi-
184 vidual bytes, for example: \x{100}{3}.
185
186 5. The dot metacharacter matches one UTF-8 character instead of a sin-
187 gle byte.
188
189 6. The escape sequence \C can be used to match a single byte in UTF-8
190 mode, but its use can lead to some strange effects. This facility is
191 not available in the alternative matching function, pcre_dfa_exec().
192
193 7. The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly
194 test characters of any code value, but the characters that PCRE recog-
195 nizes as digits, spaces, or word characters remain the same set as
196 before, all with values less than 256. This remains true even when PCRE
197 includes Unicode property support, because to do otherwise would slow
198 down PCRE in many common cases. If you really want to test for a wider
199 sense of, say, "digit", you must use Unicode property tests such as
200 \p{Nd}.
201
202 8. Similarly, characters that match the POSIX named character classes
203 are all low-valued characters.
204
205 9. However, the Perl 5.10 horizontal and vertical whitespace matching
206 escapes (\h, \H, \v, and \V) do match all the appropriate Unicode char-
207 acters.
208
209 10. Case-insensitive matching applies only to characters whose values
210 are less than 128, unless PCRE is built with Unicode property support.
211 Even when Unicode property support is available, PCRE still uses its
212 own character tables when checking the case of low-valued characters,
213 so as not to degrade performance. The Unicode property information is
214 used only for characters with higher values. Even when Unicode property
215 support is available, PCRE supports case-insensitive matching only when
216 there is a one-to-one mapping between a letter's cases. There are a
217 small number of many-to-one mappings in Unicode; these are not sup-
218 ported by PCRE.
219
220
221 AUTHOR
222
223 Philip Hazel
224 University Computing Service
225 Cambridge CB2 3QH, England.
226
227 Putting an actual email address here seems to have been a spam magnet,
228 so I've taken it away. If you want to email me, use my two initials,
229 followed by the two digits 10, at the domain cam.ac.uk.
230
231
232 REVISION
233
234 Last updated: 30 July 2007
235 Copyright (c) 1997-2007 University of Cambridge.
236 ------------------------------------------------------------------------------
237
238
239 PCREBUILD(3) PCREBUILD(3)
240
241
242 NAME
243 PCRE - Perl-compatible regular expressions
244
245
246 PCRE BUILD-TIME OPTIONS
247
248 This document describes the optional features of PCRE that can be
249 selected when the library is compiled. They are all selected, or dese-
250 lected, by providing options to the configure script that is run before
251 the make command. The complete list of options for configure (which
252 includes the standard ones such as the selection of the installation
253 directory) can be obtained by running
254
255 ./configure --help
256
257 The following sections include descriptions of options whose names
258 begin with --enable or --disable. These settings specify changes to the
259 defaults for the configure command. Because of the way that configure
260 works, --enable and --disable always come in pairs, so the complemen-
261 tary option always exists as well, but as it specifies the default, it
262 is not described.
263
264
265 C++ SUPPORT
266
267 By default, the configure script will search for a C++ compiler and C++
268 header files. If it finds them, it automatically builds the C++ wrapper
269 library for PCRE. You can disable this by adding
270
271 --disable-cpp
272
273 to the configure command.
274
275
276 UTF-8 SUPPORT
277
278 To build PCRE with support for UTF-8 character strings, add
279
280 --enable-utf8
281
282 to the configure command. Of itself, this does not make PCRE treat
283 strings as UTF-8. As well as compiling PCRE with this option, you also
284 have have to set the PCRE_UTF8 option when you call the pcre_compile()
285 function.
286
287
288 UNICODE CHARACTER PROPERTY SUPPORT
289
290 UTF-8 support allows PCRE to process character values greater than 255
291 in the strings that it handles. On its own, however, it does not pro-
292 vide any facilities for accessing the properties of such characters. If
293 you want to be able to use the pattern escapes \P, \p, and \X, which
294 refer to Unicode character properties, you must add
295
296 --enable-unicode-properties
297
298 to the configure command. This implies UTF-8 support, even if you have
299 not explicitly requested it.
300
301 Including Unicode property support adds around 30K of tables to the
302 PCRE library. Only the general category properties such as Lu and Nd
303 are supported. Details are given in the pcrepattern documentation.
304
305
306 CODE VALUE OF NEWLINE
307
308 By default, PCRE interprets character 10 (linefeed, LF) as indicating
309 the end of a line. This is the normal newline character on Unix-like
310 systems. You can compile PCRE to use character 13 (carriage return, CR)
311 instead, by adding
312
313 --enable-newline-is-cr
314
315 to the configure command. There is also a --enable-newline-is-lf
316 option, which explicitly specifies linefeed as the newline character.
317
318 Alternatively, you can specify that line endings are to be indicated by
319 the two character sequence CRLF. If you want this, add
320
321 --enable-newline-is-crlf
322
323 to the configure command. There is a fourth option, specified by
324
325 --enable-newline-is-anycrlf
326
327 which causes PCRE to recognize any of the three sequences CR, LF, or
328 CRLF as indicating a line ending. Finally, a fifth option, specified by
329
330 --enable-newline-is-any
331
332 causes PCRE to recognize any Unicode newline sequence.
333
334 Whatever line ending convention is selected when PCRE is built can be
335 overridden when the library functions are called. At build time it is
336 conventional to use the standard for your operating system.
337
338
339 BUILDING SHARED AND STATIC LIBRARIES
340
341 The PCRE building process uses libtool to build both shared and static
342 Unix libraries by default. You can suppress one of these by adding one
343 of
344
345 --disable-shared
346 --disable-static
347
348 to the configure command, as required.
349
350
351 POSIX MALLOC USAGE
352
353 When PCRE is called through the POSIX interface (see the pcreposix doc-
354 umentation), additional working storage is required for holding the
355 pointers to capturing substrings, because PCRE requires three integers
356 per substring, whereas the POSIX interface provides only two. If the
357 number of expected substrings is small, the wrapper function uses space
358 on the stack, because this is faster than using malloc() for each call.
359 The default threshold above which the stack is no longer used is 10; it
360 can be changed by adding a setting such as
361
362 --with-posix-malloc-threshold=20
363
364 to the configure command.
365
366
367 HANDLING VERY LARGE PATTERNS
368
369 Within a compiled pattern, offset values are used to point from one
370 part to another (for example, from an opening parenthesis to an alter-
371 nation metacharacter). By default, two-byte values are used for these
372 offsets, leading to a maximum size for a compiled pattern of around
373 64K. This is sufficient to handle all but the most gigantic patterns.
374 Nevertheless, some people do want to process enormous patterns, so it
375 is possible to compile PCRE to use three-byte or four-byte offsets by
376 adding a setting such as
377
378 --with-link-size=3
379
380 to the configure command. The value given must be 2, 3, or 4. Using
381 longer offsets slows down the operation of PCRE because it has to load
382 additional bytes when handling them.
383
384
385 AVOIDING EXCESSIVE STACK USAGE
386
387 When matching with the pcre_exec() function, PCRE implements backtrack-
388 ing by making recursive calls to an internal function called match().
389 In environments where the size of the stack is limited, this can se-
390 verely limit PCRE's operation. (The Unix environment does not usually
391 suffer from this problem, but it may sometimes be necessary to increase
392 the maximum stack size. There is a discussion in the pcrestack docu-
393 mentation.) An alternative approach to recursion that uses memory from
394 the heap to remember data, instead of using recursive function calls,
395 has been implemented to work round the problem of limited stack size.
396 If you want to build a version of PCRE that works this way, add
397
398 --disable-stack-for-recursion
399
400 to the configure command. With this configuration, PCRE will use the
401 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
402 ment functions. By default these point to malloc() and free(), but you
403 can replace the pointers so that your own functions are used.
404
405 Separate functions are provided rather than using pcre_malloc and
406 pcre_free because the usage is very predictable: the block sizes
407 requested are always the same, and the blocks are always freed in
408 reverse order. A calling program might be able to implement optimized
409 functions that perform better than malloc() and free(). PCRE runs
410 noticeably more slowly when built in this way. This option affects only
411 the pcre_exec() function; it is not relevant for the the
412 pcre_dfa_exec() function.
413
414
415 LIMITING PCRE RESOURCE USAGE
416
417 Internally, PCRE has a function called match(), which it calls repeat-
418 edly (sometimes recursively) when matching a pattern with the
419 pcre_exec() function. By controlling the maximum number of times this
420 function may be called during a single matching operation, a limit can
421 be placed on the resources used by a single call to pcre_exec(). The
422 limit can be changed at run time, as described in the pcreapi documen-
423 tation. The default is 10 million, but this can be changed by adding a
424 setting such as
425
426 --with-match-limit=500000
427
428 to the configure command. This setting has no effect on the
429 pcre_dfa_exec() matching function.
430
431 In some environments it is desirable to limit the depth of recursive
432 calls of match() more strictly than the total number of calls, in order
433 to restrict the maximum amount of stack (or heap, if --disable-stack-
434 for-recursion is specified) that is used. A second limit controls this;
435 it defaults to the value that is set for --with-match-limit, which
436 imposes no additional constraints. However, you can set a lower limit
437 by adding, for example,
438
439 --with-match-limit-recursion=10000
440
441 to the configure command. This value can also be overridden at run
442 time.
443
444
445 CREATING CHARACTER TABLES AT BUILD TIME
446
447 PCRE uses fixed tables for processing characters whose code values are
448 less than 256. By default, PCRE is built with a set of tables that are
449 distributed in the file pcre_chartables.c.dist. These tables are for
450 ASCII codes only. If you add
451
452 --enable-rebuild-chartables
453
454 to the configure command, the distributed tables are no longer used.
455 Instead, a program called dftables is compiled and run. This outputs
456 the source for new set of tables, created in the default locale of your
457 C runtime system. (This method of replacing the tables does not work if
458 you are cross compiling, because dftables is run on the local host. If
459 you need to create alternative tables when cross compiling, you will
460 have to do so "by hand".)
461
462
463 USING EBCDIC CODE
464
465 PCRE assumes by default that it will run in an environment where the
466 character code is ASCII (or Unicode, which is a superset of ASCII).
467 This is the case for most computer operating systems. PCRE can, how-
468 ever, be compiled to run in an EBCDIC environment by adding
469
470 --enable-ebcdic
471
472 to the configure command. This setting implies --enable-rebuild-charta-
473 bles. You should only use it if you know that you are in an EBCDIC
474 environment (for example, an IBM mainframe operating system).
475
476
477 SEE ALSO
478
479 pcreapi(3), pcre_config(3).
480
481
482 AUTHOR
483
484 Philip Hazel
485 University Computing Service
486 Cambridge CB2 3QH, England.
487
488
489 REVISION
490
491 Last updated: 30 July 2007
492 Copyright (c) 1997-2007 University of Cambridge.
493 ------------------------------------------------------------------------------
494
495
496 PCREMATCHING(3) PCREMATCHING(3)
497
498
499 NAME
500 PCRE - Perl-compatible regular expressions
501
502
503 PCRE MATCHING ALGORITHMS
504
505 This document describes the two different algorithms that are available
506 in PCRE for matching a compiled regular expression against a given sub-
507 ject string. The "standard" algorithm is the one provided by the
508 pcre_exec() function. This works in the same was as Perl's matching
509 function, and provides a Perl-compatible matching operation.
510
511 An alternative algorithm is provided by the pcre_dfa_exec() function;
512 this operates in a different way, and is not Perl-compatible. It has
513 advantages and disadvantages compared with the standard algorithm, and
514 these are described below.
515
516 When there is only one possible way in which a given subject string can
517 match a pattern, the two algorithms give the same answer. A difference
518 arises, however, when there are multiple possibilities. For example, if
519 the pattern
520
521 ^<.*>
522
523 is matched against the string
524
525 <something> <something else> <something further>
526
527 there are three possible answers. The standard algorithm finds only one
528 of them, whereas the alternative algorithm finds all three.
529
530
531 REGULAR EXPRESSIONS AS TREES
532
533 The set of strings that are matched by a regular expression can be rep-
534 resented as a tree structure. An unlimited repetition in the pattern
535 makes the tree of infinite size, but it is still a tree. Matching the
536 pattern to a given subject string (from a given starting point) can be
537 thought of as a search of the tree. There are two ways to search a
538 tree: depth-first and breadth-first, and these correspond to the two
539 matching algorithms provided by PCRE.
540
541
542 THE STANDARD MATCHING ALGORITHM
543
544 In the terminology of Jeffrey Friedl's book "Mastering Regular Expres-
545 sions", the standard algorithm is an "NFA algorithm". It conducts a
546 depth-first search of the pattern tree. That is, it proceeds along a
547 single path through the tree, checking that the subject matches what is
548 required. When there is a mismatch, the algorithm tries any alterna-
549 tives at the current point, and if they all fail, it backs up to the
550 previous branch point in the tree, and tries the next alternative
551 branch at that level. This often involves backing up (moving to the
552 left) in the subject string as well. The order in which repetition
553 branches are tried is controlled by the greedy or ungreedy nature of
554 the quantifier.
555
556 If a leaf node is reached, a matching string has been found, and at
557 that point the algorithm stops. Thus, if there is more than one possi-
558 ble match, this algorithm returns the first one that it finds. Whether
559 this is the shortest, the longest, or some intermediate length depends
560 on the way the greedy and ungreedy repetition quantifiers are specified
561 in the pattern.
562
563 Because it ends up with a single path through the tree, it is rela-
564 tively straightforward for this algorithm to keep track of the sub-
565 strings that are matched by portions of the pattern in parentheses.
566 This provides support for capturing parentheses and back references.
567
568
569 THE ALTERNATIVE MATCHING ALGORITHM
570
571 This algorithm conducts a breadth-first search of the tree. Starting
572 from the first matching point in the subject, it scans the subject
573 string from left to right, once, character by character, and as it does
574 this, it remembers all the paths through the tree that represent valid
575 matches. In Friedl's terminology, this is a kind of "DFA algorithm",
576 though it is not implemented as a traditional finite state machine (it
577 keeps multiple states active simultaneously).
578
579 The scan continues until either the end of the subject is reached, or
580 there are no more unterminated paths. At this point, terminated paths
581 represent the different matching possibilities (if there are none, the
582 match has failed). Thus, if there is more than one possible match,
583 this algorithm finds all of them, and in particular, it finds the long-
584 est. In PCRE, there is an option to stop the algorithm after the first
585 match (which is necessarily the shortest) has been found.
586
587 Note that all the matches that are found start at the same point in the
588 subject. If the pattern
589
590 cat(er(pillar)?)
591
592 is matched against the string "the caterpillar catchment", the result
593 will be the three strings "cat", "cater", and "caterpillar" that start
594 at the fourth character of the subject. The algorithm does not automat-
595 ically move on to find matches that start at later positions.
596
597 There are a number of features of PCRE regular expressions that are not
598 supported by the alternative matching algorithm. They are as follows:
599
600 1. Because the algorithm finds all possible matches, the greedy or
601 ungreedy nature of repetition quantifiers is not relevant. Greedy and
602 ungreedy quantifiers are treated in exactly the same way. However, pos-
603 sessive quantifiers can make a difference when what follows could also
604 match what is quantified, for example in a pattern like this:
605
606 ^a++\w!
607
608 This pattern matches "aaab!" but not "aaa!", which would be matched by
609 a non-possessive quantifier. Similarly, if an atomic group is present,
610 it is matched as if it were a standalone pattern at the current point,
611 and the longest match is then "locked in" for the rest of the overall
612 pattern.
613
614 2. When dealing with multiple paths through the tree simultaneously, it
615 is not straightforward to keep track of captured substrings for the
616 different matching possibilities, and PCRE's implementation of this
617 algorithm does not attempt to do this. This means that no captured sub-
618 strings are available.
619
620 3. Because no substrings are captured, back references within the pat-
621 tern are not supported, and cause errors if encountered.
622
623 4. For the same reason, conditional expressions that use a backrefer-
624 ence as the condition or test for a specific group recursion are not
625 supported.
626
627 5. Because many paths through the tree may be active, the \K escape
628 sequence, which resets the start of the match when encountered (but may
629 be on some paths and not on others), is not supported. It causes an
630 error if encountered.
631
632 6. Callouts are supported, but the value of the capture_top field is
633 always 1, and the value of the capture_last field is always -1.
634
635 7. The \C escape sequence, which (in the standard algorithm) matches a
636 single byte, even in UTF-8 mode, is not supported because the alterna-
637 tive algorithm moves through the subject string one character at a
638 time, for all active paths through the tree.
639
640
641 ADVANTAGES OF THE ALTERNATIVE ALGORITHM
642
643 Using the alternative matching algorithm provides the following advan-
644 tages:
645
646 1. All possible matches (at a single point in the subject) are automat-
647 ically found, and in particular, the longest match is found. To find
648 more than one match using the standard algorithm, you have to do kludgy
649 things with callouts.
650
651 2. There is much better support for partial matching. The restrictions
652 on the content of the pattern that apply when using the standard algo-
653 rithm for partial matching do not apply to the alternative algorithm.
654 For non-anchored patterns, the starting position of a partial match is
655 available.
656
657 3. Because the alternative algorithm scans the subject string just
658 once, and never needs to backtrack, it is possible to pass very long
659 subject strings to the matching function in several pieces, checking
660 for partial matching each time.
661
662
663 DISADVANTAGES OF THE ALTERNATIVE ALGORITHM
664
665 The alternative algorithm suffers from a number of disadvantages:
666
667 1. It is substantially slower than the standard algorithm. This is
668 partly because it has to search for all possible matches, but is also
669 because it is less susceptible to optimization.
670
671 2. Capturing parentheses and back references are not supported.
672
673 3. Although atomic groups are supported, their use does not provide the
674 performance advantage that it does for the standard algorithm.
675
676
677 AUTHOR
678
679 Philip Hazel
680 University Computing Service
681 Cambridge CB2 3QH, England.
682
683
684 REVISION
685
686 Last updated: 29 May 2007
687 Copyright (c) 1997-2007 University of Cambridge.
688 ------------------------------------------------------------------------------
689
690
691 PCREAPI(3) PCREAPI(3)
692
693
694 NAME
695 PCRE - Perl-compatible regular expressions
696
697
698 PCRE NATIVE API
699
700 #include <pcre.h>
701
702 pcre *pcre_compile(const char *pattern, int options,
703 const char **errptr, int *erroffset,
704 const unsigned char *tableptr);
705
706 pcre *pcre_compile2(const char *pattern, int options,
707 int *errorcodeptr,
708 const char **errptr, int *erroffset,
709 const unsigned char *tableptr);
710
711 pcre_extra *pcre_study(const pcre *code, int options,
712 const char **errptr);
713
714 int pcre_exec(const pcre *code, const pcre_extra *extra,
715 const char *subject, int length, int startoffset,
716 int options, int *ovector, int ovecsize);
717
718 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
719 const char *subject, int length, int startoffset,
720 int options, int *ovector, int ovecsize,
721 int *workspace, int wscount);
722
723 int pcre_copy_named_substring(const pcre *code,
724 const char *subject, int *ovector,
725 int stringcount, const char *stringname,
726 char *buffer, int buffersize);
727
728 int pcre_copy_substring(const char *subject, int *ovector,
729 int stringcount, int stringnumber, char *buffer,
730 int buffersize);
731
732 int pcre_get_named_substring(const pcre *code,
733 const char *subject, int *ovector,
734 int stringcount, const char *stringname,
735 const char **stringptr);
736
737 int pcre_get_stringnumber(const pcre *code,
738 const char *name);
739
740 int pcre_get_stringtable_entries(const pcre *code,
741 const char *name, char **first, char **last);
742
743 int pcre_get_substring(const char *subject, int *ovector,
744 int stringcount, int stringnumber,
745 const char **stringptr);
746
747 int pcre_get_substring_list(const char *subject,
748 int *ovector, int stringcount, const char ***listptr);
749
750 void pcre_free_substring(const char *stringptr);
751
752 void pcre_free_substring_list(const char **stringptr);
753
754 const unsigned char *pcre_maketables(void);
755
756 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
757 int what, void *where);
758
759 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
760
761 int pcre_refcount(pcre *code, int adjust);
762
763 int pcre_config(int what, void *where);
764
765 char *pcre_version(void);
766
767 void *(*pcre_malloc)(size_t);
768
769 void (*pcre_free)(void *);
770
771 void *(*pcre_stack_malloc)(size_t);
772
773 void (*pcre_stack_free)(void *);
774
775 int (*pcre_callout)(pcre_callout_block *);
776
777
778 PCRE API OVERVIEW
779
780 PCRE has its own native API, which is described in this document. There
781 are also some wrapper functions that correspond to the POSIX regular
782 expression API. These are described in the pcreposix documentation.
783 Both of these APIs define a set of C function calls. A C++ wrapper is
784 distributed with PCRE. It is documented in the pcrecpp page.
785
786 The native API C function prototypes are defined in the header file
787 pcre.h, and on Unix systems the library itself is called libpcre. It
788 can normally be accessed by adding -lpcre to the command for linking an
789 application that uses PCRE. The header file defines the macros
790 PCRE_MAJOR and PCRE_MINOR to contain the major and minor release num-
791 bers for the library. Applications can use these to include support
792 for different releases of PCRE.
793
794 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
795 pcre_exec() are used for compiling and matching regular expressions in
796 a Perl-compatible manner. A sample program that demonstrates the sim-
797 plest way of using them is provided in the file called pcredemo.c in
798 the source distribution. The pcresample documentation describes how to
799 run it.
800
801 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
802 ble, is also provided. This uses a different algorithm for the match-
803 ing. The alternative algorithm finds all possible matches (at a given
804 point in the subject), and scans the subject just once. However, this
805 algorithm does not return captured substrings. A description of the two
806 matching algorithms and their advantages and disadvantages is given in
807 the pcrematching documentation.
808
809 In addition to the main compiling and matching functions, there are
810 convenience functions for extracting captured substrings from a subject
811 string that is matched by pcre_exec(). They are:
812
813 pcre_copy_substring()
814 pcre_copy_named_substring()
815 pcre_get_substring()
816 pcre_get_named_substring()
817 pcre_get_substring_list()
818 pcre_get_stringnumber()
819 pcre_get_stringtable_entries()
820
821 pcre_free_substring() and pcre_free_substring_list() are also provided,
822 to free the memory used for extracted strings.
823
824 The function pcre_maketables() is used to build a set of character
825 tables in the current locale for passing to pcre_compile(),
826 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
827 provided for specialist use. Most commonly, no special tables are
828 passed, in which case internal tables that are generated when PCRE is
829 built are used.
830
831 The function pcre_fullinfo() is used to find out information about a
832 compiled pattern; pcre_info() is an obsolete version that returns only
833 some of the available information, but is retained for backwards com-
834 patibility. The function pcre_version() returns a pointer to a string
835 containing the version of PCRE and its date of release.
836
837 The function pcre_refcount() maintains a reference count in a data
838 block containing a compiled pattern. This is provided for the benefit
839 of object-oriented applications.
840
841 The global variables pcre_malloc and pcre_free initially contain the
842 entry points of the standard malloc() and free() functions, respec-
843 tively. PCRE calls the memory management functions via these variables,
844 so a calling program can replace them if it wishes to intercept the
845 calls. This should be done before calling any PCRE functions.
846
847 The global variables pcre_stack_malloc and pcre_stack_free are also
848 indirections to memory management functions. These special functions
849 are used only when PCRE is compiled to use the heap for remembering
850 data, instead of recursive function calls, when running the pcre_exec()
851 function. See the pcrebuild documentation for details of how to do
852 this. It is a non-standard way of building PCRE, for use in environ-
853 ments that have limited stacks. Because of the greater use of memory
854 management, it runs more slowly. Separate functions are provided so
855 that special-purpose external code can be used for this case. When
856 used, these functions are always called in a stack-like manner (last
857 obtained, first freed), and always for memory blocks of the same size.
858 There is a discussion about PCRE's stack usage in the pcrestack docu-
859 mentation.
860
861 The global variable pcre_callout initially contains NULL. It can be set
862 by the caller to a "callout" function, which PCRE will then call at
863 specified points during a matching operation. Details are given in the
864 pcrecallout documentation.
865
866
867 NEWLINES
868
869 PCRE supports five different conventions for indicating line breaks in
870 strings: a single CR (carriage return) character, a single LF (line-
871 feed) character, the two-character sequence CRLF, any of the three pre-
872 ceding, or any Unicode newline sequence. The Unicode newline sequences
873 are the three just mentioned, plus the single characters VT (vertical
874 tab, U+000B), FF (formfeed, U+000C), NEL (next line, U+0085), LS (line
875 separator, U+2028), and PS (paragraph separator, U+2029).
876
877 Each of the first three conventions is used by at least one operating
878 system as its standard newline sequence. When PCRE is built, a default
879 can be specified. The default default is LF, which is the Unix stan-
880 dard. When PCRE is run, the default can be overridden, either when a
881 pattern is compiled, or when it is matched.
882
883 In the PCRE documentation the word "newline" is used to mean "the char-
884 acter or pair of characters that indicate a line break". The choice of
885 newline convention affects the handling of the dot, circumflex, and
886 dollar metacharacters, the handling of #-comments in /x mode, and, when
887 CRLF is a recognized line ending sequence, the match position advance-
888 ment for a non-anchored pattern. The choice of newline convention does
889 not affect the interpretation of the \n or \r escape sequences.
890
891
892 MULTITHREADING
893
894 The PCRE functions can be used in multi-threading applications, with
895 the proviso that the memory management functions pointed to by
896 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
897 callout function pointed to by pcre_callout, are shared by all threads.
898
899 The compiled form of a regular expression is not altered during match-
900 ing, so the same compiled pattern can safely be used by several threads
901 at once.
902
903
904 SAVING PRECOMPILED PATTERNS FOR LATER USE
905
906 The compiled form of a regular expression can be saved and re-used at a
907 later time, possibly by a different program, and even on a host other
908 than the one on which it was compiled. Details are given in the
909 pcreprecompile documentation. However, compiling a regular expression
910 with one version of PCRE for use with a different version is not guar-
911 anteed to work and may cause crashes.
912
913
914 CHECKING BUILD-TIME OPTIONS
915
916 int pcre_config(int what, void *where);
917
918 The function pcre_config() makes it possible for a PCRE client to dis-
919 cover which optional features have been compiled into the PCRE library.
920 The pcrebuild documentation has more details about these optional fea-
921 tures.
922
923 The first argument for pcre_config() is an integer, specifying which
924 information is required; the second argument is a pointer to a variable
925 into which the information is placed. The following information is
926 available:
927
928 PCRE_CONFIG_UTF8
929
930 The output is an integer that is set to one if UTF-8 support is avail-
931 able; otherwise it is set to zero.
932
933 PCRE_CONFIG_UNICODE_PROPERTIES
934
935 The output is an integer that is set to one if support for Unicode
936 character properties is available; otherwise it is set to zero.
937
938 PCRE_CONFIG_NEWLINE
939
940 The output is an integer whose value specifies the default character
941 sequence that is recognized as meaning "newline". The four values that
942 are supported are: 10 for LF, 13 for CR, 3338 for CRLF, -2 for ANYCRLF,
943 and -1 for ANY. The default should normally be the standard sequence
944 for your operating system.
945
946 PCRE_CONFIG_LINK_SIZE
947
948 The output is an integer that contains the number of bytes used for
949 internal linkage in compiled regular expressions. The value is 2, 3, or
950 4. Larger values allow larger regular expressions to be compiled, at
951 the expense of slower matching. The default value of 2 is sufficient
952 for all but the most massive patterns, since it allows the compiled
953 pattern to be up to 64K in size.
954
955 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
956
957 The output is an integer that contains the threshold above which the
958 POSIX interface uses malloc() for output vectors. Further details are
959 given in the pcreposix documentation.
960
961 PCRE_CONFIG_MATCH_LIMIT
962
963 The output is an integer that gives the default limit for the number of
964 internal matching function calls in a pcre_exec() execution. Further
965 details are given with pcre_exec() below.
966
967 PCRE_CONFIG_MATCH_LIMIT_RECURSION
968
969 The output is an integer that gives the default limit for the depth of
970 recursion when calling the internal matching function in a pcre_exec()
971 execution. Further details are given with pcre_exec() below.
972
973 PCRE_CONFIG_STACKRECURSE
974
975 The output is an integer that is set to one if internal recursion when
976 running pcre_exec() is implemented by recursive function calls that use
977 the stack to remember their state. This is the usual way that PCRE is
978 compiled. The output is zero if PCRE was compiled to use blocks of data
979 on the heap instead of recursive function calls. In this case,
980 pcre_stack_malloc and pcre_stack_free are called to manage memory
981 blocks on the heap, thus avoiding the use of the stack.
982
983
984 COMPILING A PATTERN
985
986 pcre *pcre_compile(const char *pattern, int options,
987 const char **errptr, int *erroffset,
988 const unsigned char *tableptr);
989
990 pcre *pcre_compile2(const char *pattern, int options,
991 int *errorcodeptr,
992 const char **errptr, int *erroffset,
993 const unsigned char *tableptr);
994
995 Either of the functions pcre_compile() or pcre_compile2() can be called
996 to compile a pattern into an internal form. The only difference between
997 the two interfaces is that pcre_compile2() has an additional argument,
998 errorcodeptr, via which a numerical error code can be returned.
999
1000 The pattern is a C string terminated by a binary zero, and is passed in
1001 the pattern argument. A pointer to a single block of memory that is
1002 obtained via pcre_malloc is returned. This contains the compiled code
1003 and related data. The pcre type is defined for the returned block; this
1004 is a typedef for a structure whose contents are not externally defined.
1005 It is up to the caller to free the memory (via pcre_free) when it is no
1006 longer required.
1007
1008 Although the compiled code of a PCRE regex is relocatable, that is, it
1009 does not depend on memory location, the complete pcre data block is not
1010 fully relocatable, because it may contain a copy of the tableptr argu-
1011 ment, which is an address (see below).
1012
1013 The options argument contains various bit settings that affect the com-
1014 pilation. It should be zero if no options are required. The available
1015 options are described below. Some of them, in particular, those that
1016 are compatible with Perl, can also be set and unset from within the
1017 pattern (see the detailed description in the pcrepattern documenta-
1018 tion). For these options, the contents of the options argument speci-
1019 fies their initial settings at the start of compilation and execution.
1020 The PCRE_ANCHORED and PCRE_NEWLINE_xxx options can be set at the time
1021 of matching as well as at compile time.
1022
1023 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
1024 if compilation of a pattern fails, pcre_compile() returns NULL, and
1025 sets the variable pointed to by errptr to point to a textual error mes-
1026 sage. This is a static string that is part of the library. You must not
1027 try to free it. The offset from the start of the pattern to the charac-
1028 ter where the error was discovered is placed in the variable pointed to
1029 by erroffset, which must not be NULL. If it is, an immediate error is
1030 given.
1031
1032 If pcre_compile2() is used instead of pcre_compile(), and the error-
1033 codeptr argument is not NULL, a non-zero error code number is returned
1034 via this argument in the event of an error. This is in addition to the
1035 textual error message. Error codes and messages are listed below.
1036
1037 If the final argument, tableptr, is NULL, PCRE uses a default set of
1038 character tables that are built when PCRE is compiled, using the
1039 default C locale. Otherwise, tableptr must be an address that is the
1040 result of a call to pcre_maketables(). This value is stored with the
1041 compiled pattern, and used again by pcre_exec(), unless another table
1042 pointer is passed to it. For more discussion, see the section on locale
1043 support below.
1044
1045 This code fragment shows a typical straightforward call to pcre_com-
1046 pile():
1047
1048 pcre *re;
1049 const char *error;
1050 int erroffset;
1051 re = pcre_compile(
1052 "^A.*Z", /* the pattern */
1053 0, /* default options */
1054 &error, /* for error message */
1055 &erroffset, /* for error offset */
1056 NULL); /* use default character tables */
1057
1058 The following names for option bits are defined in the pcre.h header
1059 file:
1060
1061 PCRE_ANCHORED
1062
1063 If this bit is set, the pattern is forced to be "anchored", that is, it
1064 is constrained to match only at the first matching point in the string
1065 that is being searched (the "subject string"). This effect can also be
1066 achieved by appropriate constructs in the pattern itself, which is the
1067 only way to do it in Perl.
1068
1069 PCRE_AUTO_CALLOUT
1070
1071 If this bit is set, pcre_compile() automatically inserts callout items,
1072 all with number 255, before each pattern item. For discussion of the
1073 callout facility, see the pcrecallout documentation.
1074
1075 PCRE_CASELESS
1076
1077 If this bit is set, letters in the pattern match both upper and lower
1078 case letters. It is equivalent to Perl's /i option, and it can be
1079 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
1080 always understands the concept of case for characters whose values are
1081 less than 128, so caseless matching is always possible. For characters
1082 with higher values, the concept of case is supported if PCRE is com-
1083 piled with Unicode property support, but not otherwise. If you want to
1084 use caseless matching for characters 128 and above, you must ensure
1085 that PCRE is compiled with Unicode property support as well as with
1086 UTF-8 support.
1087
1088 PCRE_DOLLAR_ENDONLY
1089
1090 If this bit is set, a dollar metacharacter in the pattern matches only
1091 at the end of the subject string. Without this option, a dollar also
1092 matches immediately before a newline at the end of the string (but not
1093 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
1094 if PCRE_MULTILINE is set. There is no equivalent to this option in
1095 Perl, and no way to set it within a pattern.
1096
1097 PCRE_DOTALL
1098
1099 If this bit is set, a dot metacharater in the pattern matches all char-
1100 acters, including those that indicate newline. Without it, a dot does
1101 not match when the current position is at a newline. This option is
1102 equivalent to Perl's /s option, and it can be changed within a pattern
1103 by a (?s) option setting. A negative class such as [^a] always matches
1104 newline characters, independent of the setting of this option.
1105
1106 PCRE_DUPNAMES
1107
1108 If this bit is set, names used to identify capturing subpatterns need
1109 not be unique. This can be helpful for certain types of pattern when it
1110 is known that only one instance of the named subpattern can ever be
1111 matched. There are more details of named subpatterns below; see also
1112 the pcrepattern documentation.
1113
1114 PCRE_EXTENDED
1115
1116 If this bit is set, whitespace data characters in the pattern are
1117 totally ignored except when escaped or inside a character class. White-
1118 space does not include the VT character (code 11). In addition, charac-
1119 ters between an unescaped # outside a character class and the next new-
1120 line, inclusive, are also ignored. This is equivalent to Perl's /x
1121 option, and it can be changed within a pattern by a (?x) option set-
1122 ting.
1123
1124 This option makes it possible to include comments inside complicated
1125 patterns. Note, however, that this applies only to data characters.
1126 Whitespace characters may never appear within special character
1127 sequences in a pattern, for example within the sequence (?( which
1128 introduces a conditional subpattern.
1129
1130 PCRE_EXTRA
1131
1132 This option was invented in order to turn on additional functionality
1133 of PCRE that is incompatible with Perl, but it is currently of very
1134 little use. When set, any backslash in a pattern that is followed by a
1135 letter that has no special meaning causes an error, thus reserving
1136 these combinations for future expansion. By default, as in Perl, a
1137 backslash followed by a letter with no special meaning is treated as a
1138 literal. (Perl can, however, be persuaded to give a warning for this.)
1139 There are at present no other features controlled by this option. It
1140 can also be set by a (?X) option setting within a pattern.
1141
1142 PCRE_FIRSTLINE
1143
1144 If this option is set, an unanchored pattern is required to match
1145 before or at the first newline in the subject string, though the
1146 matched text may continue over the newline.
1147
1148 PCRE_MULTILINE
1149
1150 By default, PCRE treats the subject string as consisting of a single
1151 line of characters (even if it actually contains newlines). The "start
1152 of line" metacharacter (^) matches only at the start of the string,
1153 while the "end of line" metacharacter ($) matches only at the end of
1154 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1155 is set). This is the same as Perl.
1156
1157 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1158 constructs match immediately following or immediately before internal
1159 newlines in the subject string, respectively, as well as at the very
1160 start and end. This is equivalent to Perl's /m option, and it can be
1161 changed within a pattern by a (?m) option setting. If there are no new-
1162 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1163 setting PCRE_MULTILINE has no effect.
1164
1165 PCRE_NEWLINE_CR
1166 PCRE_NEWLINE_LF
1167 PCRE_NEWLINE_CRLF
1168 PCRE_NEWLINE_ANYCRLF
1169 PCRE_NEWLINE_ANY
1170
1171 These options override the default newline definition that was chosen
1172 when PCRE was built. Setting the first or the second specifies that a
1173 newline is indicated by a single character (CR or LF, respectively).
1174 Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
1175 two-character CRLF sequence. Setting PCRE_NEWLINE_ANYCRLF specifies
1176 that any of the three preceding sequences should be recognized. Setting
1177 PCRE_NEWLINE_ANY specifies that any Unicode newline sequence should be
1178 recognized. The Unicode newline sequences are the three just mentioned,
1179 plus the single characters VT (vertical tab, U+000B), FF (formfeed,
1180 U+000C), NEL (next line, U+0085), LS (line separator, U+2028), and PS
1181 (paragraph separator, U+2029). The last two are recognized only in
1182 UTF-8 mode.
1183
1184 The newline setting in the options word uses three bits that are
1185 treated as a number, giving eight possibilities. Currently only six are
1186 used (default plus the five values above). This means that if you set
1187 more than one newline option, the combination may or may not be sensi-
1188 ble. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equivalent to
1189 PCRE_NEWLINE_CRLF, but other combinations may yield unused numbers and
1190 cause an error.
1191
1192 The only time that a line break is specially recognized when compiling
1193 a pattern is if PCRE_EXTENDED is set, and an unescaped # outside a
1194 character class is encountered. This indicates a comment that lasts
1195 until after the next line break sequence. In other circumstances, line
1196 break sequences are treated as literal data, except that in
1197 PCRE_EXTENDED mode, both CR and LF are treated as whitespace characters
1198 and are therefore ignored.
1199
1200 The newline option that is set at compile time becomes the default that
1201 is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1202
1203 PCRE_NO_AUTO_CAPTURE
1204
1205 If this option is set, it disables the use of numbered capturing paren-
1206 theses in the pattern. Any opening parenthesis that is not followed by
1207 ? behaves as if it were followed by ?: but named parentheses can still
1208 be used for capturing (and they acquire numbers in the usual way).
1209 There is no equivalent of this option in Perl.
1210
1211 PCRE_UNGREEDY
1212
1213 This option inverts the "greediness" of the quantifiers so that they
1214 are not greedy by default, but become greedy if followed by "?". It is
1215 not compatible with Perl. It can also be set by a (?U) option setting
1216 within the pattern.
1217
1218 PCRE_UTF8
1219
1220 This option causes PCRE to regard both the pattern and the subject as
1221 strings of UTF-8 characters instead of single-byte character strings.
1222 However, it is available only when PCRE is built to include UTF-8 sup-
1223 port. If not, the use of this option provokes an error. Details of how
1224 this option changes the behaviour of PCRE are given in the section on
1225 UTF-8 support in the main pcre page.
1226
1227 PCRE_NO_UTF8_CHECK
1228
1229 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1230 automatically checked. If an invalid UTF-8 sequence of bytes is found,
1231 pcre_compile() returns an error. If you already know that your pattern
1232 is valid, and you want to skip this check for performance reasons, you
1233 can set the PCRE_NO_UTF8_CHECK option. When it is set, the effect of
1234 passing an invalid UTF-8 string as a pattern is undefined. It may cause
1235 your program to crash. Note that this option can also be passed to
1236 pcre_exec() and pcre_dfa_exec(), to suppress the UTF-8 validity check-
1237 ing of subject strings.
1238
1239
1240 COMPILATION ERROR CODES
1241
1242 The following table lists the error codes than may be returned by
1243 pcre_compile2(), along with the error messages that may be returned by
1244 both compiling functions. As PCRE has developed, some error codes have
1245 fallen out of use. To avoid confusion, they have not been re-used.
1246
1247 0 no error
1248 1 \ at end of pattern
1249 2 \c at end of pattern
1250 3 unrecognized character follows \
1251 4 numbers out of order in {} quantifier
1252 5 number too big in {} quantifier
1253 6 missing terminating ] for character class
1254 7 invalid escape sequence in character class
1255 8 range out of order in character class
1256 9 nothing to repeat
1257 10 [this code is not in use]
1258 11 internal error: unexpected repeat
1259 12 unrecognized character after (?
1260 13 POSIX named classes are supported only within a class
1261 14 missing )
1262 15 reference to non-existent subpattern
1263 16 erroffset passed as NULL
1264 17 unknown option bit(s) set
1265 18 missing ) after comment
1266 19 [this code is not in use]
1267 20 regular expression too large
1268 21 failed to get memory
1269 22 unmatched parentheses
1270 23 internal error: code overflow
1271 24 unrecognized character after (?<
1272 25 lookbehind assertion is not fixed length
1273 26 malformed number or name after (?(
1274 27 conditional group contains more than two branches
1275 28 assertion expected after (?(
1276 29 (?R or (?[+-]digits must be followed by )
1277 30 unknown POSIX class name
1278 31 POSIX collating elements are not supported
1279 32 this version of PCRE is not compiled with PCRE_UTF8 support
1280 33 [this code is not in use]
1281 34 character value in \x{...} sequence is too large
1282 35 invalid condition (?(0)
1283 36 \C not allowed in lookbehind assertion
1284 37 PCRE does not support \L, \l, \N, \U, or \u
1285 38 number after (?C is > 255
1286 39 closing ) for (?C expected
1287 40 recursive call could loop indefinitely
1288 41 unrecognized character after (?P
1289 42 syntax error in subpattern name (missing terminator)
1290 43 two named subpatterns have the same name
1291 44 invalid UTF-8 string
1292 45 support for \P, \p, and \X has not been compiled
1293 46 malformed \P or \p sequence
1294 47 unknown property name after \P or \p
1295 48 subpattern name is too long (maximum 32 characters)
1296 49 too many named subpatterns (maximum 10,000)
1297 50 repeated subpattern is too long
1298 51 octal value is greater than \377 (not in UTF-8 mode)
1299 52 internal error: overran compiling workspace
1300 53 internal error: previously-checked referenced subpattern not
1301 found
1302 54 DEFINE group contains more than one branch
1303 55 repeating a DEFINE group is not allowed
1304 56 inconsistent NEWLINE options"
1305 57 \g is not followed by a braced name or an optionally braced
1306 non-zero number
1307 58 (?+ or (?- or (?(+ or (?(- must be followed by a non-zero number
1308
1309
1310 STUDYING A PATTERN
1311
1312 pcre_extra *pcre_study(const pcre *code, int options
1313 const char **errptr);
1314
1315 If a compiled pattern is going to be used several times, it is worth
1316 spending more time analyzing it in order to speed up the time taken for
1317 matching. The function pcre_study() takes a pointer to a compiled pat-
1318 tern as its first argument. If studying the pattern produces additional
1319 information that will help speed up matching, pcre_study() returns a
1320 pointer to a pcre_extra block, in which the study_data field points to
1321 the results of the study.
1322
1323 The returned value from pcre_study() can be passed directly to
1324 pcre_exec(). However, a pcre_extra block also contains other fields
1325 that can be set by the caller before the block is passed; these are
1326 described below in the section on matching a pattern.
1327
1328 If studying the pattern does not produce any additional information
1329 pcre_study() returns NULL. In that circumstance, if the calling program
1330 wants to pass any of the other fields to pcre_exec(), it must set up
1331 its own pcre_extra block.
1332
1333 The second argument of pcre_study() contains option bits. At present,
1334 no options are defined, and this argument should always be zero.
1335
1336 The third argument for pcre_study() is a pointer for an error message.
1337 If studying succeeds (even if no data is returned), the variable it
1338 points to is set to NULL. Otherwise it is set to point to a textual
1339 error message. This is a static string that is part of the library. You
1340 must not try to free it. You should test the error pointer for NULL
1341 after calling pcre_study(), to be sure that it has run successfully.
1342
1343 This is a typical call to pcre_study():
1344
1345 pcre_extra *pe;
1346 pe = pcre_study(
1347 re, /* result of pcre_compile() */
1348 0, /* no options exist */
1349 &error); /* set to NULL or points to a message */
1350
1351 At present, studying a pattern is useful only for non-anchored patterns
1352 that do not have a single fixed starting character. A bitmap of possi-
1353 ble starting bytes is created.
1354
1355
1356 LOCALE SUPPORT
1357
1358 PCRE handles caseless matching, and determines whether characters are
1359 letters, digits, or whatever, by reference to a set of tables, indexed
1360 by character value. When running in UTF-8 mode, this applies only to
1361 characters with codes less than 128. Higher-valued codes never match
1362 escapes such as \w or \d, but can be tested with \p if PCRE is built
1363 with Unicode character property support. The use of locales with Uni-
1364 code is discouraged. If you are handling characters with codes greater
1365 than 128, you should either use UTF-8 and Unicode, or use locales, but
1366 not try to mix the two.
1367
1368 PCRE contains an internal set of tables that are used when the final
1369 argument of pcre_compile() is NULL. These are sufficient for many
1370 applications. Normally, the internal tables recognize only ASCII char-
1371 acters. However, when PCRE is built, it is possible to cause the inter-
1372 nal tables to be rebuilt in the default "C" locale of the local system,
1373 which may cause them to be different.
1374
1375 The internal tables can always be overridden by tables supplied by the
1376 application that calls PCRE. These may be created in a different locale
1377 from the default. As more and more applications change to using Uni-
1378 code, the need for this locale support is expected to die away.
1379
1380 External tables are built by calling the pcre_maketables() function,
1381 which has no arguments, in the relevant locale. The result can then be
1382 passed to pcre_compile() or pcre_exec() as often as necessary. For
1383 example, to build and use tables that are appropriate for the French
1384 locale (where accented characters with values greater than 128 are
1385 treated as letters), the following code could be used:
1386
1387 setlocale(LC_CTYPE, "fr_FR");
1388 tables = pcre_maketables();
1389 re = pcre_compile(..., tables);
1390
1391 The locale name "fr_FR" is used on Linux and other Unix-like systems;
1392 if you are using Windows, the name for the French locale is "french".
1393
1394 When pcre_maketables() runs, the tables are built in memory that is
1395 obtained via pcre_malloc. It is the caller's responsibility to ensure
1396 that the memory containing the tables remains available for as long as
1397 it is needed.
1398
1399 The pointer that is passed to pcre_compile() is saved with the compiled
1400 pattern, and the same tables are used via this pointer by pcre_study()
1401 and normally also by pcre_exec(). Thus, by default, for any single pat-
1402 tern, compilation, studying and matching all happen in the same locale,
1403 but different patterns can be compiled in different locales.
1404
1405 It is possible to pass a table pointer or NULL (indicating the use of
1406 the internal tables) to pcre_exec(). Although not intended for this
1407 purpose, this facility could be used to match a pattern in a different
1408 locale from the one in which it was compiled. Passing table pointers at
1409 run time is discussed below in the section on matching a pattern.
1410
1411
1412 INFORMATION ABOUT A PATTERN
1413
1414 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1415 int what, void *where);
1416
1417 The pcre_fullinfo() function returns information about a compiled pat-
1418 tern. It replaces the obsolete pcre_info() function, which is neverthe-
1419 less retained for backwards compability (and is documented below).
1420
1421 The first argument for pcre_fullinfo() is a pointer to the compiled
1422 pattern. The second argument is the result of pcre_study(), or NULL if
1423 the pattern was not studied. The third argument specifies which piece
1424 of information is required, and the fourth argument is a pointer to a
1425 variable to receive the data. The yield of the function is zero for
1426 success, or one of the following negative numbers:
1427
1428 PCRE_ERROR_NULL the argument code was NULL
1429 the argument where was NULL
1430 PCRE_ERROR_BADMAGIC the "magic number" was not found
1431 PCRE_ERROR_BADOPTION the value of what was invalid
1432
1433 The "magic number" is placed at the start of each compiled pattern as
1434 an simple check against passing an arbitrary memory pointer. Here is a
1435 typical call of pcre_fullinfo(), to obtain the length of the compiled
1436 pattern:
1437
1438 int rc;
1439 size_t length;
1440 rc = pcre_fullinfo(
1441 re, /* result of pcre_compile() */
1442 pe, /* result of pcre_study(), or NULL */
1443 PCRE_INFO_SIZE, /* what is required */
1444 &length); /* where to put the data */
1445
1446 The possible values for the third argument are defined in pcre.h, and
1447 are as follows:
1448
1449 PCRE_INFO_BACKREFMAX
1450
1451 Return the number of the highest back reference in the pattern. The
1452 fourth argument should point to an int variable. Zero is returned if
1453 there are no back references.
1454
1455 PCRE_INFO_CAPTURECOUNT
1456
1457 Return the number of capturing subpatterns in the pattern. The fourth
1458 argument should point to an int variable.
1459
1460 PCRE_INFO_DEFAULT_TABLES
1461
1462 Return a pointer to the internal default character tables within PCRE.
1463 The fourth argument should point to an unsigned char * variable. This
1464 information call is provided for internal use by the pcre_study() func-
1465 tion. External callers can cause PCRE to use its internal tables by
1466 passing a NULL table pointer.
1467
1468 PCRE_INFO_FIRSTBYTE
1469
1470 Return information about the first byte of any matched string, for a
1471 non-anchored pattern. The fourth argument should point to an int vari-
1472 able. (This option used to be called PCRE_INFO_FIRSTCHAR; the old name
1473 is still recognized for backwards compatibility.)
1474
1475 If there is a fixed first byte, for example, from a pattern such as
1476 (cat|cow|coyote), its value is returned. Otherwise, if either
1477
1478 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
1479 branch starts with "^", or
1480
1481 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
1482 set (if it were set, the pattern would be anchored),
1483
1484 -1 is returned, indicating that the pattern matches only at the start
1485 of a subject string or after any newline within the string. Otherwise
1486 -2 is returned. For anchored patterns, -2 is returned.
1487
1488 PCRE_INFO_FIRSTTABLE
1489
1490 If the pattern was studied, and this resulted in the construction of a
1491 256-bit table indicating a fixed set of bytes for the first byte in any
1492 matching string, a pointer to the table is returned. Otherwise NULL is
1493 returned. The fourth argument should point to an unsigned char * vari-
1494 able.
1495
1496 PCRE_INFO_JCHANGED
1497
1498 Return 1 if the (?J) option setting is used in the pattern, otherwise
1499 0. The fourth argument should point to an int variable. The (?J) inter-
1500 nal option setting changes the local PCRE_DUPNAMES option.
1501
1502 PCRE_INFO_LASTLITERAL
1503
1504 Return the value of the rightmost literal byte that must exist in any
1505 matched string, other than at its start, if such a byte has been
1506 recorded. The fourth argument should point to an int variable. If there
1507 is no such byte, -1 is returned. For anchored patterns, a last literal
1508 byte is recorded only if it follows something of variable length. For
1509 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
1510 /^a\dz\d/ the returned value is -1.
1511
1512 PCRE_INFO_NAMECOUNT
1513 PCRE_INFO_NAMEENTRYSIZE
1514 PCRE_INFO_NAMETABLE
1515
1516 PCRE supports the use of named as well as numbered capturing parenthe-
1517 ses. The names are just an additional way of identifying the parenthe-
1518 ses, which still acquire numbers. Several convenience functions such as
1519 pcre_get_named_substring() are provided for extracting captured sub-
1520 strings by name. It is also possible to extract the data directly, by
1521 first converting the name to a number in order to access the correct
1522 pointers in the output vector (described with pcre_exec() below). To do
1523 the conversion, you need to use the name-to-number map, which is
1524 described by these three values.
1525
1526 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
1527 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
1528 of each entry; both of these return an int value. The entry size
1529 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
1530 a pointer to the first entry of the table (a pointer to char). The
1531 first two bytes of each entry are the number of the capturing parenthe-
1532 sis, most significant byte first. The rest of the entry is the corre-
1533 sponding name, zero terminated. The names are in alphabetical order.
1534 When PCRE_DUPNAMES is set, duplicate names are in order of their paren-
1535 theses numbers. For example, consider the following pattern (assume
1536 PCRE_EXTENDED is set, so white space - including newlines - is
1537 ignored):
1538
1539 (?<date> (?<year>(\d\d)?\d\d) -
1540 (?<month>\d\d) - (?<day>\d\d) )
1541
1542 There are four named subpatterns, so the table has four entries, and
1543 each entry in the table is eight bytes long. The table is as follows,
1544 with non-printing bytes shows in hexadecimal, and undefined bytes shown
1545 as ??:
1546
1547 00 01 d a t e 00 ??
1548 00 05 d a y 00 ?? ??
1549 00 04 m o n t h 00
1550 00 02 y e a r 00 ??
1551
1552 When writing code to extract data from named subpatterns using the
1553 name-to-number map, remember that the length of the entries is likely
1554 to be different for each compiled pattern.
1555
1556 PCRE_INFO_OKPARTIAL
1557
1558 Return 1 if the pattern can be used for partial matching, otherwise 0.
1559 The fourth argument should point to an int variable. The pcrepartial
1560 documentation lists the restrictions that apply to patterns when par-
1561 tial matching is used.
1562
1563 PCRE_INFO_OPTIONS
1564
1565 Return a copy of the options with which the pattern was compiled. The
1566 fourth argument should point to an unsigned long int variable. These
1567 option bits are those specified in the call to pcre_compile(), modified
1568 by any top-level option settings at the start of the pattern itself. In
1569 other words, they are the options that will be in force when matching
1570 starts. For example, if the pattern /(?im)abc(?-i)d/ is compiled with
1571 the PCRE_EXTENDED option, the result is PCRE_CASELESS, PCRE_MULTILINE,
1572 and PCRE_EXTENDED.
1573
1574 A pattern is automatically anchored by PCRE if all of its top-level
1575 alternatives begin with one of the following:
1576
1577 ^ unless PCRE_MULTILINE is set
1578 \A always
1579 \G always
1580 .* if PCRE_DOTALL is set and there are no back
1581 references to the subpattern in which .* appears
1582
1583 For such patterns, the PCRE_ANCHORED bit is set in the options returned
1584 by pcre_fullinfo().
1585
1586 PCRE_INFO_SIZE
1587
1588 Return the size of the compiled pattern, that is, the value that was
1589 passed as the argument to pcre_malloc() when PCRE was getting memory in
1590 which to place the compiled data. The fourth argument should point to a
1591 size_t variable.
1592
1593 PCRE_INFO_STUDYSIZE
1594
1595 Return the size of the data block pointed to by the study_data field in
1596 a pcre_extra block. That is, it is the value that was passed to
1597 pcre_malloc() when PCRE was getting memory into which to place the data
1598 created by pcre_study(). The fourth argument should point to a size_t
1599 variable.
1600
1601
1602 OBSOLETE INFO FUNCTION
1603
1604 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
1605
1606 The pcre_info() function is now obsolete because its interface is too
1607 restrictive to return all the available data about a compiled pattern.
1608 New programs should use pcre_fullinfo() instead. The yield of
1609 pcre_info() is the number of capturing subpatterns, or one of the fol-
1610 lowing negative numbers:
1611
1612 PCRE_ERROR_NULL the argument code was NULL
1613 PCRE_ERROR_BADMAGIC the "magic number" was not found
1614
1615 If the optptr argument is not NULL, a copy of the options with which
1616 the pattern was compiled is placed in the integer it points to (see
1617 PCRE_INFO_OPTIONS above).
1618
1619 If the pattern is not anchored and the firstcharptr argument is not
1620 NULL, it is used to pass back information about the first character of
1621 any matched string (see PCRE_INFO_FIRSTBYTE above).
1622
1623
1624 REFERENCE COUNTS
1625
1626 int pcre_refcount(pcre *code, int adjust);
1627
1628 The pcre_refcount() function is used to maintain a reference count in
1629 the data block that contains a compiled pattern. It is provided for the
1630 benefit of applications that operate in an object-oriented manner,
1631 where different parts of the application may be using the same compiled
1632 pattern, but you want to free the block when they are all done.
1633
1634 When a pattern is compiled, the reference count field is initialized to
1635 zero. It is changed only by calling this function, whose action is to
1636 add the adjust value (which may be positive or negative) to it. The
1637 yield of the function is the new value. However, the value of the count
1638 is constrained to lie between 0 and 65535, inclusive. If the new value
1639 is outside these limits, it is forced to the appropriate limit value.
1640
1641 Except when it is zero, the reference count is not correctly preserved
1642 if a pattern is compiled on one host and then transferred to a host
1643 whose byte-order is different. (This seems a highly unlikely scenario.)
1644
1645
1646 MATCHING A PATTERN: THE TRADITIONAL FUNCTION
1647
1648 int pcre_exec(const pcre *code, const pcre_extra *extra,
1649 const char *subject, int length, int startoffset,
1650 int options, int *ovector, int ovecsize);
1651
1652 The function pcre_exec() is called to match a subject string against a
1653 compiled pattern, which is passed in the code argument. If the pattern
1654 has been studied, the result of the study should be passed in the extra
1655 argument. This function is the main matching facility of the library,
1656 and it operates in a Perl-like manner. For specialist use there is also
1657 an alternative matching function, which is described below in the sec-
1658 tion about the pcre_dfa_exec() function.
1659
1660 In most applications, the pattern will have been compiled (and option-
1661 ally studied) in the same process that calls pcre_exec(). However, it
1662 is possible to save compiled patterns and study data, and then use them
1663 later in different processes, possibly even on different hosts. For a
1664 discussion about this, see the pcreprecompile documentation.
1665
1666 Here is an example of a simple call to pcre_exec():
1667
1668 int rc;
1669 int ovector[30];
1670 rc = pcre_exec(
1671 re, /* result of pcre_compile() */
1672 NULL, /* we didn't study the pattern */
1673 "some string", /* the subject string */
1674 11, /* the length of the subject string */
1675 0, /* start at offset 0 in the subject */
1676 0, /* default options */
1677 ovector, /* vector of integers for substring information */
1678 30); /* number of elements (NOT size in bytes) */
1679
1680 Extra data for pcre_exec()
1681
1682 If the extra argument is not NULL, it must point to a pcre_extra data
1683 block. The pcre_study() function returns such a block (when it doesn't
1684 return NULL), but you can also create one for yourself, and pass addi-
1685 tional information in it. The pcre_extra block contains the following
1686 fields (not necessarily in this order):
1687
1688 unsigned long int flags;
1689 void *study_data;
1690 unsigned long int match_limit;
1691 unsigned long int match_limit_recursion;
1692 void *callout_data;
1693 const unsigned char *tables;
1694
1695 The flags field is a bitmap that specifies which of the other fields
1696 are set. The flag bits are:
1697
1698 PCRE_EXTRA_STUDY_DATA
1699 PCRE_EXTRA_MATCH_LIMIT
1700 PCRE_EXTRA_MATCH_LIMIT_RECURSION
1701 PCRE_EXTRA_CALLOUT_DATA
1702 PCRE_EXTRA_TABLES
1703
1704 Other flag bits should be set to zero. The study_data field is set in
1705 the pcre_extra block that is returned by pcre_study(), together with
1706 the appropriate flag bit. You should not set this yourself, but you may
1707 add to the block by setting the other fields and their corresponding
1708 flag bits.
1709
1710 The match_limit field provides a means of preventing PCRE from using up
1711 a vast amount of resources when running patterns that are not going to
1712 match, but which have a very large number of possibilities in their
1713 search trees. The classic example is the use of nested unlimited
1714 repeats.
1715
1716 Internally, PCRE uses a function called match() which it calls repeat-
1717 edly (sometimes recursively). The limit set by match_limit is imposed
1718 on the number of times this function is called during a match, which
1719 has the effect of limiting the amount of backtracking that can take
1720 place. For patterns that are not anchored, the count restarts from zero
1721 for each position in the subject string.
1722
1723 The default value for the limit can be set when PCRE is built; the
1724 default default is 10 million, which handles all but the most extreme
1725 cases. You can override the default by suppling pcre_exec() with a
1726 pcre_extra block in which match_limit is set, and
1727 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
1728 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
1729
1730 The match_limit_recursion field is similar to match_limit, but instead
1731 of limiting the total number of times that match() is called, it limits
1732 the depth of recursion. The recursion depth is a smaller number than
1733 the total number of calls, because not all calls to match() are recur-
1734 sive. This limit is of use only if it is set smaller than match_limit.
1735
1736 Limiting the recursion depth limits the amount of stack that can be
1737 used, or, when PCRE has been compiled to use memory on the heap instead
1738 of the stack, the amount of heap memory that can be used.
1739
1740 The default value for match_limit_recursion can be set when PCRE is
1741 built; the default default is the same value as the default for
1742 match_limit. You can override the default by suppling pcre_exec() with
1743 a pcre_extra block in which match_limit_recursion is set, and
1744 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
1745 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
1746
1747 The pcre_callout field is used in conjunction with the "callout" fea-
1748 ture, which is described in the pcrecallout documentation.
1749
1750 The tables field is used to pass a character tables pointer to
1751 pcre_exec(); this overrides the value that is stored with the compiled
1752 pattern. A non-NULL value is stored with the compiled pattern only if
1753 custom tables were supplied to pcre_compile() via its tableptr argu-
1754 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
1755 PCRE's internal tables to be used. This facility is helpful when re-
1756 using patterns that have been saved after compiling with an external
1757 set of tables, because the external tables might be at a different
1758 address when pcre_exec() is called. See the pcreprecompile documenta-
1759 tion for a discussion of saving compiled patterns for later use.
1760
1761 Option bits for pcre_exec()
1762
1763 The unused bits of the options argument for pcre_exec() must be zero.
1764 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
1765 PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK and
1766 PCRE_PARTIAL.
1767
1768 PCRE_ANCHORED
1769
1770 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
1771 matching position. If a pattern was compiled with PCRE_ANCHORED, or
1772 turned out to be anchored by virtue of its contents, it cannot be made
1773 unachored at matching time.
1774
1775 PCRE_NEWLINE_CR
1776 PCRE_NEWLINE_LF
1777 PCRE_NEWLINE_CRLF
1778 PCRE_NEWLINE_ANYCRLF
1779 PCRE_NEWLINE_ANY
1780
1781 These options override the newline definition that was chosen or
1782 defaulted when the pattern was compiled. For details, see the descrip-
1783 tion of pcre_compile() above. During matching, the newline choice
1784 affects the behaviour of the dot, circumflex, and dollar metacharac-
1785 ters. It may also alter the way the match position is advanced after a
1786 match failure for an unanchored pattern. When PCRE_NEWLINE_CRLF,
1787 PCRE_NEWLINE_ANYCRLF, or PCRE_NEWLINE_ANY is set, and a match attempt
1788 fails when the current position is at a CRLF sequence, the match posi-
1789 tion is advanced by two characters instead of one, in other words, to
1790 after the CRLF.
1791
1792 PCRE_NOTBOL
1793
1794 This option specifies that first character of the subject string is not
1795 the beginning of a line, so the circumflex metacharacter should not
1796 match before it. Setting this without PCRE_MULTILINE (at compile time)
1797 causes circumflex never to match. This option affects only the behav-
1798 iour of the circumflex metacharacter. It does not affect \A.
1799
1800 PCRE_NOTEOL
1801
1802 This option specifies that the end of the subject string is not the end
1803 of a line, so the dollar metacharacter should not match it nor (except
1804 in multiline mode) a newline immediately before it. Setting this with-
1805 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
1806 option affects only the behaviour of the dollar metacharacter. It does
1807 not affect \Z or \z.
1808
1809 PCRE_NOTEMPTY
1810
1811 An empty string is not considered to be a valid match if this option is
1812 set. If there are alternatives in the pattern, they are tried. If all
1813 the alternatives match the empty string, the entire match fails. For
1814 example, if the pattern
1815
1816 a?b?
1817
1818 is applied to a string not beginning with "a" or "b", it matches the
1819 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
1820 match is not valid, so PCRE searches further into the string for occur-
1821 rences of "a" or "b".
1822
1823 Perl has no direct equivalent of PCRE_NOTEMPTY, but it does make a spe-
1824 cial case of a pattern match of the empty string within its split()
1825 function, and when using the /g modifier. It is possible to emulate
1826 Perl's behaviour after matching a null string by first trying the match
1827 again at the same offset with PCRE_NOTEMPTY and PCRE_ANCHORED, and then
1828 if that fails by advancing the starting offset (see below) and trying
1829 an ordinary match again. There is some code that demonstrates how to do
1830 this in the pcredemo.c sample program.
1831
1832 PCRE_NO_UTF8_CHECK
1833
1834 When PCRE_UTF8 is set at compile time, the validity of the subject as a
1835 UTF-8 string is automatically checked when pcre_exec() is subsequently
1836 called. The value of startoffset is also checked to ensure that it
1837 points to the start of a UTF-8 character. If an invalid UTF-8 sequence
1838 of bytes is found, pcre_exec() returns the error PCRE_ERROR_BADUTF8. If
1839 startoffset contains an invalid value, PCRE_ERROR_BADUTF8_OFFSET is
1840 returned.
1841
1842 If you already know that your subject is valid, and you want to skip
1843 these checks for performance reasons, you can set the
1844 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
1845 do this for the second and subsequent calls to pcre_exec() if you are
1846 making repeated calls to find all the matches in a single subject
1847 string. However, you should be sure that the value of startoffset
1848 points to the start of a UTF-8 character. When PCRE_NO_UTF8_CHECK is
1849 set, the effect of passing an invalid UTF-8 string as a subject, or a
1850 value of startoffset that does not point to the start of a UTF-8 char-
1851 acter, is undefined. Your program may crash.
1852
1853 PCRE_PARTIAL
1854
1855 This option turns on the partial matching feature. If the subject
1856 string fails to match the pattern, but at some point during the match-
1857 ing process the end of the subject was reached (that is, the subject
1858 partially matches the pattern and the failure to match occurred only
1859 because there were not enough subject characters), pcre_exec() returns
1860 PCRE_ERROR_PARTIAL instead of PCRE_ERROR_NOMATCH. When PCRE_PARTIAL is
1861 used, there are restrictions on what may appear in the pattern. These
1862 are discussed in the pcrepartial documentation.
1863
1864 The string to be matched by pcre_exec()
1865
1866 The subject string is passed to pcre_exec() as a pointer in subject, a
1867 length in length, and a starting byte offset in startoffset. In UTF-8
1868 mode, the byte offset must point to the start of a UTF-8 character.
1869 Unlike the pattern string, the subject may contain binary zero bytes.
1870 When the starting offset is zero, the search for a match starts at the
1871 beginning of the subject, and this is by far the most common case.
1872
1873 A non-zero starting offset is useful when searching for another match
1874 in the same subject by calling pcre_exec() again after a previous suc-
1875 cess. Setting startoffset differs from just passing over a shortened
1876 string and setting PCRE_NOTBOL in the case of a pattern that begins
1877 with any kind of lookbehind. For example, consider the pattern
1878
1879 \Biss\B
1880
1881 which finds occurrences of "iss" in the middle of words. (\B matches
1882 only if the current position in the subject is not a word boundary.)
1883 When applied to the string "Mississipi" the first call to pcre_exec()
1884 finds the first occurrence. If pcre_exec() is called again with just
1885 the remainder of the subject, namely "issipi", it does not match,
1886 because \B is always false at the start of the subject, which is deemed
1887 to be a word boundary. However, if pcre_exec() is passed the entire
1888 string again, but with startoffset set to 4, it finds the second occur-
1889 rence of "iss" because it is able to look behind the starting point to
1890 discover that it is preceded by a letter.
1891
1892 If a non-zero starting offset is passed when the pattern is anchored,
1893 one attempt to match at the given offset is made. This can only succeed
1894 if the pattern does not require the match to be at the start of the
1895 subject.
1896
1897 How pcre_exec() returns captured substrings
1898
1899 In general, a pattern matches a certain portion of the subject, and in
1900 addition, further substrings from the subject may be picked out by
1901 parts of the pattern. Following the usage in Jeffrey Friedl's book,
1902 this is called "capturing" in what follows, and the phrase "capturing
1903 subpattern" is used for a fragment of a pattern that picks out a sub-
1904 string. PCRE supports several other kinds of parenthesized subpattern
1905 that do not cause substrings to be captured.
1906
1907 Captured substrings are returned to the caller via a vector of integer
1908 offsets whose address is passed in ovector. The number of elements in
1909 the vector is passed in ovecsize, which must be a non-negative number.
1910 Note: this argument is NOT the size of ovector in bytes.
1911
1912 The first two-thirds of the vector is used to pass back captured sub-
1913 strings, each substring using a pair of integers. The remaining third
1914 of the vector is used as workspace by pcre_exec() while matching cap-
1915 turing subpatterns, and is not available for passing back information.
1916 The length passed in ovecsize should always be a multiple of three. If
1917 it is not, it is rounded down.
1918
1919 When a match is successful, information about captured substrings is
1920 returned in pairs of integers, starting at the beginning of ovector,
1921 and continuing up to two-thirds of its length at the most. The first
1922 element of a pair is set to the offset of the first character in a sub-
1923 string, and the second is set to the offset of the first character
1924 after the end of a substring. The first pair, ovector[0] and ovec-
1925 tor[1], identify the portion of the subject string matched by the
1926 entire pattern. The next pair is used for the first capturing subpat-
1927 tern, and so on. The value returned by pcre_exec() is one more than the
1928 highest numbered pair that has been set. For example, if two substrings
1929 have been captured, the returned value is 3. If there are no capturing
1930 subpatterns, the return value from a successful match is 1, indicating
1931 that just the first pair of offsets has been set.
1932
1933 If a capturing subpattern is matched repeatedly, it is the last portion
1934 of the string that it matched that is returned.
1935
1936 If the vector is too small to hold all the captured substring offsets,
1937 it is used as far as possible (up to two-thirds of its length), and the
1938 function returns a value of zero. In particular, if the substring off-
1939 sets are not of interest, pcre_exec() may be called with ovector passed
1940 as NULL and ovecsize as zero. However, if the pattern contains back
1941 references and the ovector is not big enough to remember the related
1942 substrings, PCRE has to get additional memory for use during matching.
1943 Thus it is usually advisable to supply an ovector.
1944
1945 The pcre_info() function can be used to find out how many capturing
1946 subpatterns there are in a compiled pattern. The smallest size for
1947 ovector that will allow for n captured substrings, in addition to the
1948 offsets of the substring matched by the whole pattern, is (n+1)*3.
1949
1950 It is possible for capturing subpattern number n+1 to match some part
1951 of the subject when subpattern n has not been used at all. For example,
1952 if the string "abc" is matched against the pattern (a|(z))(bc) the
1953 return from the function is 4, and subpatterns 1 and 3 are matched, but
1954 2 is not. When this happens, both values in the offset pairs corre-
1955 sponding to unused subpatterns are set to -1.
1956
1957 Offset values that correspond to unused subpatterns at the end of the
1958 expression are also set to -1. For example, if the string "abc" is
1959 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
1960 matched. The return from the function is 2, because the highest used
1961 capturing subpattern number is 1. However, you can refer to the offsets
1962 for the second and third capturing subpatterns if you wish (assuming
1963 the vector is large enough, of course).
1964
1965 Some convenience functions are provided for extracting the captured
1966 substrings as separate strings. These are described below.
1967
1968 Error return values from pcre_exec()
1969
1970 If pcre_exec() fails, it returns a negative number. The following are
1971 defined in the header file:
1972
1973 PCRE_ERROR_NOMATCH (-1)
1974
1975 The subject string did not match the pattern.
1976
1977 PCRE_ERROR_NULL (-2)
1978
1979 Either code or subject was passed as NULL, or ovector was NULL and
1980 ovecsize was not zero.
1981
1982 PCRE_ERROR_BADOPTION (-3)
1983
1984 An unrecognized bit was set in the options argument.
1985
1986 PCRE_ERROR_BADMAGIC (-4)
1987
1988 PCRE stores a 4-byte "magic number" at the start of the compiled code,
1989 to catch the case when it is passed a junk pointer and to detect when a
1990 pattern that was compiled in an environment of one endianness is run in
1991 an environment with the other endianness. This is the error that PCRE
1992 gives when the magic number is not present.
1993
1994 PCRE_ERROR_UNKNOWN_OPCODE (-5)
1995
1996 While running the pattern match, an unknown item was encountered in the
1997 compiled pattern. This error could be caused by a bug in PCRE or by
1998 overwriting of the compiled pattern.
1999
2000 PCRE_ERROR_NOMEMORY (-6)
2001
2002 If a pattern contains back references, but the ovector that is passed
2003 to pcre_exec() is not big enough to remember the referenced substrings,
2004 PCRE gets a block of memory at the start of matching to use for this
2005 purpose. If the call via pcre_malloc() fails, this error is given. The
2006 memory is automatically freed at the end of matching.
2007
2008 PCRE_ERROR_NOSUBSTRING (-7)
2009
2010 This error is used by the pcre_copy_substring(), pcre_get_substring(),
2011 and pcre_get_substring_list() functions (see below). It is never
2012 returned by pcre_exec().
2013
2014 PCRE_ERROR_MATCHLIMIT (-8)
2015
2016 The backtracking limit, as specified by the match_limit field in a
2017 pcre_extra structure (or defaulted) was reached. See the description
2018 above.
2019
2020 PCRE_ERROR_CALLOUT (-9)
2021
2022 This error is never generated by pcre_exec() itself. It is provided for
2023 use by callout functions that want to yield a distinctive error code.
2024 See the pcrecallout documentation for details.
2025
2026 PCRE_ERROR_BADUTF8 (-10)
2027
2028 A string that contains an invalid UTF-8 byte sequence was passed as a
2029 subject.
2030
2031 PCRE_ERROR_BADUTF8_OFFSET (-11)
2032
2033 The UTF-8 byte sequence that was passed as a subject was valid, but the
2034 value of startoffset did not point to the beginning of a UTF-8 charac-
2035 ter.
2036
2037 PCRE_ERROR_PARTIAL (-12)
2038
2039 The subject string did not match, but it did match partially. See the
2040 pcrepartial documentation for details of partial matching.
2041
2042 PCRE_ERROR_BADPARTIAL (-13)
2043
2044 The PCRE_PARTIAL option was used with a compiled pattern containing
2045 items that are not supported for partial matching. See the pcrepartial
2046 documentation for details of partial matching.
2047
2048 PCRE_ERROR_INTERNAL (-14)
2049
2050 An unexpected internal error has occurred. This error could be caused
2051 by a bug in PCRE or by overwriting of the compiled pattern.
2052
2053 PCRE_ERROR_BADCOUNT (-15)
2054
2055 This error is given if the value of the ovecsize argument is negative.
2056
2057 PCRE_ERROR_RECURSIONLIMIT (-21)
2058
2059 The internal recursion limit, as specified by the match_limit_recursion
2060 field in a pcre_extra structure (or defaulted) was reached. See the
2061 description above.
2062
2063 PCRE_ERROR_BADNEWLINE (-23)
2064
2065 An invalid combination of PCRE_NEWLINE_xxx options was given.
2066
2067 Error numbers -16 to -20 and -22 are not used by pcre_exec().
2068
2069
2070 EXTRACTING CAPTURED SUBSTRINGS BY NUMBER
2071
2072 int pcre_copy_substring(const char *subject, int *ovector,
2073 int stringcount, int stringnumber, char *buffer,
2074 int buffersize);
2075
2076 int pcre_get_substring(const char *subject, int *ovector,
2077 int stringcount, int stringnumber,
2078 const char **stringptr);
2079
2080 int pcre_get_substring_list(const char *subject,
2081 int *ovector, int stringcount, const char ***listptr);
2082
2083 Captured substrings can be accessed directly by using the offsets
2084 returned by pcre_exec() in ovector. For convenience, the functions
2085 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
2086 string_list() are provided for extracting captured substrings as new,
2087 separate, zero-terminated strings. These functions identify substrings
2088 by number. The next section describes functions for extracting named
2089 substrings.
2090
2091 A substring that contains a binary zero is correctly extracted and has
2092 a further zero added on the end, but the result is not, of course, a C
2093 string. However, you can process such a string by referring to the
2094 length that is returned by pcre_copy_substring() and pcre_get_sub-
2095 string(). Unfortunately, the interface to pcre_get_substring_list() is
2096 not adequate for handling strings containing binary zeros, because the
2097 end of the final string is not independently indicated.
2098
2099 The first three arguments are the same for all three of these func-
2100 tions: subject is the subject string that has just been successfully
2101 matched, ovector is a pointer to the vector of integer offsets that was
2102 passed to pcre_exec(), and stringcount is the number of substrings that
2103 were captured by the match, including the substring that matched the
2104 entire regular expression. This is the value returned by pcre_exec() if
2105 it is greater than zero. If pcre_exec() returned zero, indicating that
2106 it ran out of space in ovector, the value passed as stringcount should
2107 be the number of elements in the vector divided by three.
2108
2109 The functions pcre_copy_substring() and pcre_get_substring() extract a
2110 single substring, whose number is given as stringnumber. A value of
2111 zero extracts the substring that matched the entire pattern, whereas
2112 higher values extract the captured substrings. For pcre_copy_sub-
2113 string(), the string is placed in buffer, whose length is given by
2114 buffersize, while for pcre_get_substring() a new block of memory is
2115 obtained via pcre_malloc, and its address is returned via stringptr.
2116 The yield of the function is the length of the string, not including
2117 the terminating zero, or one of these error codes:
2118
2119 PCRE_ERROR_NOMEMORY (-6)
2120
2121 The buffer was too small for pcre_copy_substring(), or the attempt to
2122 get memory failed for pcre_get_substring().
2123
2124 PCRE_ERROR_NOSUBSTRING (-7)
2125
2126 There is no substring whose number is stringnumber.
2127
2128 The pcre_get_substring_list() function extracts all available sub-
2129 strings and builds a list of pointers to them. All this is done in a
2130 single block of memory that is obtained via pcre_malloc. The address of
2131 the memory block is returned via listptr, which is also the start of
2132 the list of string pointers. The end of the list is marked by a NULL
2133 pointer. The yield of the function is zero if all went well, or the
2134 error code
2135
2136 PCRE_ERROR_NOMEMORY (-6)
2137
2138 if the attempt to get the memory block failed.
2139
2140 When any of these functions encounter a substring that is unset, which
2141 can happen when capturing subpattern number n+1 matches some part of
2142 the subject, but subpattern n has not been used at all, they return an
2143 empty string. This can be distinguished from a genuine zero-length sub-
2144 string by inspecting the appropriate offset in ovector, which is nega-
2145 tive for unset substrings.
2146
2147 The two convenience functions pcre_free_substring() and pcre_free_sub-
2148 string_list() can be used to free the memory returned by a previous
2149 call of pcre_get_substring() or pcre_get_substring_list(), respec-
2150 tively. They do nothing more than call the function pointed to by
2151 pcre_free, which of course could be called directly from a C program.
2152 However, PCRE is used in some situations where it is linked via a spe-
2153 cial interface to another programming language that cannot use
2154 pcre_free directly; it is for these cases that the functions are pro-
2155 vided.
2156
2157
2158 EXTRACTING CAPTURED SUBSTRINGS BY NAME
2159
2160 int pcre_get_stringnumber(const pcre *code,
2161 const char *name);
2162
2163 int pcre_copy_named_substring(const pcre *code,
2164 const char *subject, int *ovector,
2165 int stringcount, const char *stringname,
2166 char *buffer, int buffersize);
2167
2168 int pcre_get_named_substring(const pcre *code,
2169 const char *subject, int *ovector,
2170 int stringcount, const char *stringname,
2171 const char **stringptr);
2172
2173 To extract a substring by name, you first have to find associated num-
2174 ber. For example, for this pattern
2175
2176 (a+)b(?<xxx>\d+)...
2177
2178 the number of the subpattern called "xxx" is 2. If the name is known to
2179 be unique (PCRE_DUPNAMES was not set), you can find the number from the
2180 name by calling pcre_get_stringnumber(). The first argument is the com-
2181 piled pattern, and the second is the name. The yield of the function is
2182 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
2183 subpattern of that name.
2184
2185 Given the number, you can extract the substring directly, or use one of
2186 the functions described in the previous section. For convenience, there
2187 are also two functions that do the whole job.
2188
2189 Most of the arguments of pcre_copy_named_substring() and
2190 pcre_get_named_substring() are the same as those for the similarly
2191 named functions that extract by number. As these are described in the
2192 previous section, they are not re-described here. There are just two
2193 differences:
2194
2195 First, instead of a substring number, a substring name is given. Sec-
2196 ond, there is an extra argument, given at the start, which is a pointer
2197 to the compiled pattern. This is needed in order to gain access to the
2198 name-to-number translation table.
2199
2200 These functions call pcre_get_stringnumber(), and if it succeeds, they
2201 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
2202 ate. NOTE: If PCRE_DUPNAMES is set and there are duplicate names, the
2203 behaviour may not be what you want (see the next section).
2204
2205
2206 DUPLICATE SUBPATTERN NAMES
2207
2208 int pcre_get_stringtable_entries(const pcre *code,
2209 const char *name, char **first, char **last);
2210
2211 When a pattern is compiled with the PCRE_DUPNAMES option, names for
2212 subpatterns are not required to be unique. Normally, patterns with
2213 duplicate names are such that in any one match, only one of the named
2214 subpatterns participates. An example is shown in the pcrepattern docu-
2215 mentation. When duplicates are present, pcre_copy_named_substring() and
2216 pcre_get_named_substring() return the first substring corresponding to
2217 the given name that is set. If none are set, an empty string is
2218 returned. The pcre_get_stringnumber() function returns one of the num-
2219 bers that are associated with the name, but it is not defined which it
2220 is.
2221
2222 If you want to get full details of all captured substrings for a given
2223 name, you must use the pcre_get_stringtable_entries() function. The
2224 first argument is the compiled pattern, and the second is the name. The
2225 third and fourth are pointers to variables which are updated by the
2226 function. After it has run, they point to the first and last entries in
2227 the name-to-number table for the given name. The function itself
2228 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
2229 there are none. The format of the table is described above in the sec-
2230 tion entitled Information about a pattern. Given all the relevant
2231 entries for the name, you can extract each of their numbers, and hence
2232 the captured data, if any.
2233
2234
2235 FINDING ALL POSSIBLE MATCHES
2236
2237 The traditional matching function uses a similar algorithm to Perl,
2238 which stops when it finds the first match, starting at a given point in
2239 the subject. If you want to find all possible matches, or the longest
2240 possible match, consider using the alternative matching function (see
2241 below) instead. If you cannot use the alternative function, but still
2242 need to find all possible matches, you can kludge it up by making use
2243 of the callout facility, which is described in the pcrecallout documen-
2244 tation.
2245
2246 What you have to do is to insert a callout right at the end of the pat-
2247 tern. When your callout function is called, extract and save the cur-
2248 rent matched substring. Then return 1, which forces pcre_exec() to
2249 backtrack and try other alternatives. Ultimately, when it runs out of
2250 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
2251
2252
2253 MATCHING A PATTERN: THE ALTERNATIVE FUNCTION
2254
2255 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
2256 const char *subject, int length, int startoffset,
2257 int options, int *ovector, int ovecsize,
2258 int *workspace, int wscount);
2259
2260 The function pcre_dfa_exec() is called to match a subject string
2261 against a compiled pattern, using a matching algorithm that scans the
2262 subject string just once, and does not backtrack. This has different
2263 characteristics to the normal algorithm, and is not compatible with
2264 Perl. Some of the features of PCRE patterns are not supported. Never-
2265 theless, there are times when this kind of matching can be useful. For
2266 a discussion of the two matching algorithms, see the pcrematching docu-
2267 mentation.
2268
2269 The arguments for the pcre_dfa_exec() function are the same as for
2270 pcre_exec(), plus two extras. The ovector argument is used in a differ-
2271 ent way, and this is described below. The other common arguments are
2272 used in the same way as for pcre_exec(), so their description is not
2273 repeated here.
2274
2275 The two additional arguments provide workspace for the function. The
2276 workspace vector should contain at least 20 elements. It is used for
2277 keeping track of multiple paths through the pattern tree. More
2278 workspace will be needed for patterns and subjects where there are a
2279 lot of potential matches.
2280
2281 Here is an example of a simple call to pcre_dfa_exec():
2282
2283 int rc;
2284 int ovector[10];
2285 int wspace[20];
2286 rc = pcre_dfa_exec(
2287 re, /* result of pcre_compile() */
2288 NULL, /* we didn't study the pattern */
2289 "some string", /* the subject string */
2290 11, /* the length of the subject string */
2291 0, /* start at offset 0 in the subject */
2292 0, /* default options */
2293 ovector, /* vector of integers for substring information */
2294 10, /* number of elements (NOT size in bytes) */
2295 wspace, /* working space vector */
2296 20); /* number of elements (NOT size in bytes) */
2297
2298 Option bits for pcre_dfa_exec()
2299
2300 The unused bits of the options argument for pcre_dfa_exec() must be
2301 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
2302 LINE_xxx, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NO_UTF8_CHECK,
2303 PCRE_PARTIAL, PCRE_DFA_SHORTEST, and PCRE_DFA_RESTART. All but the last
2304 three of these are the same as for pcre_exec(), so their description is
2305 not repeated here.
2306
2307 PCRE_PARTIAL
2308
2309 This has the same general effect as it does for pcre_exec(), but the
2310 details are slightly different. When PCRE_PARTIAL is set for
2311 pcre_dfa_exec(), the return code PCRE_ERROR_NOMATCH is converted into
2312 PCRE_ERROR_PARTIAL if the end of the subject is reached, there have
2313 been no complete matches, but there is still at least one matching pos-
2314 sibility. The portion of the string that provided the partial match is
2315 set as the first matching string.
2316
2317 PCRE_DFA_SHORTEST
2318
2319 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
2320 stop as soon as it has found one match. Because of the way the alterna-
2321 tive algorithm works, this is necessarily the shortest possible match
2322 at the first possible matching point in the subject string.
2323
2324 PCRE_DFA_RESTART
2325
2326 When pcre_dfa_exec() is called with the PCRE_PARTIAL option, and
2327 returns a partial match, it is possible to call it again, with addi-
2328 tional subject characters, and have it continue with the same match.
2329 The PCRE_DFA_RESTART option requests this action; when it is set, the
2330 workspace and wscount options must reference the same vector as before
2331 because data about the match so far is left in them after a partial
2332 match. There is more discussion of this facility in the pcrepartial
2333 documentation.
2334
2335 Successful returns from pcre_dfa_exec()
2336
2337 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
2338 string in the subject. Note, however, that all the matches from one run
2339 of the function start at the same point in the subject. The shorter
2340 matches are all initial substrings of the longer matches. For example,
2341 if the pattern
2342
2343 <.*>
2344
2345 is matched against the string
2346
2347 This is <something> <something else> <something further> no more
2348
2349 the three matched strings are
2350
2351 <something>
2352 <something> <something else>
2353 <something> <something else> <something further>
2354
2355 On success, the yield of the function is a number greater than zero,
2356 which is the number of matched substrings. The substrings themselves
2357 are returned in ovector. Each string uses two elements; the first is
2358 the offset to the start, and the second is the offset to the end. In
2359 fact, all the strings have the same start offset. (Space could have
2360 been saved by giving this only once, but it was decided to retain some
2361 compatibility with the way pcre_exec() returns data, even though the
2362 meaning of the strings is different.)
2363
2364 The strings are returned in reverse order of length; that is, the long-
2365 est matching string is given first. If there were too many matches to
2366 fit into ovector, the yield of the function is zero, and the vector is
2367 filled with the longest matches.
2368
2369 Error returns from pcre_dfa_exec()
2370
2371 The pcre_dfa_exec() function returns a negative number when it fails.
2372 Many of the errors are the same as for pcre_exec(), and these are
2373 described above. There are in addition the following errors that are
2374 specific to pcre_dfa_exec():
2375
2376 PCRE_ERROR_DFA_UITEM (-16)
2377
2378 This return is given if pcre_dfa_exec() encounters an item in the pat-
2379 tern that it does not support, for instance, the use of \C or a back
2380 reference.
2381
2382 PCRE_ERROR_DFA_UCOND (-17)
2383
2384 This return is given if pcre_dfa_exec() encounters a condition item
2385 that uses a back reference for the condition, or a test for recursion
2386 in a specific group. These are not supported.
2387
2388 PCRE_ERROR_DFA_UMLIMIT (-18)
2389
2390 This return is given if pcre_dfa_exec() is called with an extra block
2391 that contains a setting of the match_limit field. This is not supported
2392 (it is meaningless).
2393
2394 PCRE_ERROR_DFA_WSSIZE (-19)
2395
2396 This return is given if pcre_dfa_exec() runs out of space in the
2397 workspace vector.
2398
2399 PCRE_ERROR_DFA_RECURSE (-20)
2400
2401 When a recursive subpattern is processed, the matching function calls
2402 itself recursively, using private vectors for ovector and workspace.
2403 This error is given if the output vector is not large enough. This
2404 should be extremely rare, as a vector of size 1000 is used.
2405
2406
2407 SEE ALSO
2408
2409 pcrebuild(3), pcrecallout(3), pcrecpp(3)(3), pcrematching(3), pcrepar-
2410 tial(3), pcreposix(3), pcreprecompile(3), pcresample(3), pcrestack(3).
2411
2412
2413 AUTHOR
2414
2415 Philip Hazel
2416 University Computing Service
2417 Cambridge CB2 3QH, England.
2418
2419
2420 REVISION
2421
2422 Last updated: 30 July 2007
2423 Copyright (c) 1997-2007 University of Cambridge.
2424 ------------------------------------------------------------------------------
2425
2426
2427 PCRECALLOUT(3) PCRECALLOUT(3)
2428
2429
2430 NAME
2431 PCRE - Perl-compatible regular expressions
2432
2433
2434 PCRE CALLOUTS
2435
2436 int (*pcre_callout)(pcre_callout_block *);
2437
2438 PCRE provides a feature called "callout", which is a means of temporar-
2439 ily passing control to the caller of PCRE in the middle of pattern
2440 matching. The caller of PCRE provides an external function by putting
2441 its entry point in the global variable pcre_callout. By default, this
2442 variable contains NULL, which disables all calling out.
2443
2444 Within a regular expression, (?C) indicates the points at which the
2445 external function is to be called. Different callout points can be
2446 identified by putting a number less than 256 after the letter C. The
2447 default value is zero. For example, this pattern has two callout
2448 points:
2449
2450 (?C1)abc(?C2)def
2451
2452 If the PCRE_AUTO_CALLOUT option bit is set when pcre_compile() is
2453 called, PCRE automatically inserts callouts, all with number 255,
2454 before each item in the pattern. For example, if PCRE_AUTO_CALLOUT is
2455 used with the pattern
2456
2457 A(\d{2}|--)
2458
2459 it is processed as if it were
2460
2461 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
2462
2463 Notice that there is a callout before and after each parenthesis and
2464 alternation bar. Automatic callouts can be used for tracking the
2465 progress of pattern matching. The pcretest command has an option that
2466 sets automatic callouts; when it is used, the output indicates how the
2467 pattern is matched. This is useful information when you are trying to
2468 optimize the performance of a particular pattern.
2469
2470
2471 MISSING CALLOUTS
2472
2473 You should be aware that, because of optimizations in the way PCRE
2474 matches patterns, callouts sometimes do not happen. For example, if the
2475 pattern is
2476
2477 ab(?C4)cd
2478
2479 PCRE knows that any matching string must contain the letter "d". If the
2480 subject string is "abyz", the lack of "d" means that matching doesn't
2481 ever start, and the callout is never reached. However, with "abyd",
2482 though the result is still no match, the callout is obeyed.
2483
2484
2485 THE CALLOUT INTERFACE
2486
2487 During matching, when PCRE reaches a callout point, the external func-
2488 tion defined by pcre_callout is called (if it is set). This applies to
2489 both the pcre_exec() and the pcre_dfa_exec() matching functions. The
2490 only argument to the callout function is a pointer to a pcre_callout
2491 block. This structure contains the following fields:
2492
2493 int version;
2494 int callout_number;
2495 int *offset_vector;
2496 const char *subject;
2497 int subject_length;
2498 int start_match;
2499 int current_position;
2500 int capture_top;
2501 int capture_last;
2502 void *callout_data;
2503 int pattern_position;
2504 int next_item_length;
2505
2506 The version field is an integer containing the version number of the
2507 block format. The initial version was 0; the current version is 1. The
2508 version number will change again in future if additional fields are
2509 added, but the intention is never to remove any of the existing fields.
2510
2511 The callout_number field contains the number of the callout, as com-
2512 piled into the pattern (that is, the number after ?C for manual call-
2513 outs, and 255 for automatically generated callouts).
2514
2515 The offset_vector field is a pointer to the vector of offsets that was
2516 passed by the caller to pcre_exec() or pcre_dfa_exec(). When
2517 pcre_exec() is used, the contents can be inspected in order to extract
2518 substrings that have been matched so far, in the same way as for
2519 extracting substrings after a match has completed. For pcre_dfa_exec()
2520 this field is not useful.
2521
2522 The subject and subject_length fields contain copies of the values that
2523 were passed to pcre_exec().
2524
2525 The start_match field normally contains the offset within the subject
2526 at which the current match attempt started. However, if the escape
2527 sequence \K has been encountered, this value is changed to reflect the
2528 modified starting point. If the pattern is not anchored, the callout
2529 function may be called several times from the same point in the pattern
2530 for different starting points in the subject.
2531
2532 The current_position field contains the offset within the subject of
2533 the current match pointer.
2534
2535 When the pcre_exec() function is used, the capture_top field contains
2536 one more than the number of the highest numbered captured substring so
2537 far. If no substrings have been captured, the value of capture_top is
2538 one. This is always the case when pcre_dfa_exec() is used, because it
2539 does not support captured substrings.
2540
2541 The capture_last field contains the number of the most recently cap-
2542 tured substring. If no substrings have been captured, its value is -1.
2543 This is always the case when pcre_dfa_exec() is used.
2544
2545 The callout_data field contains a value that is passed to pcre_exec()
2546 or pcre_dfa_exec() specifically so that it can be passed back in call-
2547 outs. It is passed in the pcre_callout field of the pcre_extra data
2548 structure. If no such data was passed, the value of callout_data in a
2549 pcre_callout block is NULL. There is a description of the pcre_extra
2550 structure in the pcreapi documentation.
2551
2552 The pattern_position field is present from version 1 of the pcre_call-
2553 out structure. It contains the offset to the next item to be matched in
2554 the pattern string.
2555
2556 The next_item_length field is present from version 1 of the pcre_call-
2557 out structure. It contains the length of the next item to be matched in
2558 the pattern string. When the callout immediately precedes an alterna-
2559 tion bar, a closing parenthesis, or the end of the pattern, the length
2560 is zero. When the callout precedes an opening parenthesis, the length
2561 is that of the entire subpattern.
2562
2563 The pattern_position and next_item_length fields are intended to help
2564 in distinguishing between different automatic callouts, which all have
2565 the same callout number. However, they are set for all callouts.
2566
2567
2568 RETURN VALUES
2569
2570 The external callout function returns an integer to PCRE. If the value
2571 is zero, matching proceeds as normal. If the value is greater than
2572 zero, matching fails at the current point, but the testing of other
2573 matching possibilities goes ahead, just as if a lookahead assertion had
2574 failed. If the value is less than zero, the match is abandoned, and
2575 pcre_exec() (or pcre_dfa_exec()) returns the negative value.
2576
2577 Negative values should normally be chosen from the set of
2578 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
2579 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
2580 reserved for use by callout functions; it will never be used by PCRE
2581 itself.
2582
2583
2584 AUTHOR
2585
2586 Philip Hazel
2587 University Computing Service
2588 Cambridge CB2 3QH, England.
2589
2590
2591 REVISION
2592
2593 Last updated: 29 May 2007
2594 Copyright (c) 1997-2007 University of Cambridge.
2595 ------------------------------------------------------------------------------
2596
2597
2598 PCRECOMPAT(3) PCRECOMPAT(3)
2599
2600
2601 NAME
2602 PCRE - Perl-compatible regular expressions
2603
2604
2605 DIFFERENCES BETWEEN PCRE AND PERL
2606
2607 This document describes the differences in the ways that PCRE and Perl
2608 handle regular expressions. The differences described here are mainly
2609 with respect to Perl 5.8, though PCRE versions 7.0 and later contain
2610 some features that are expected to be in the forthcoming Perl 5.10.
2611
2612 1. PCRE has only a subset of Perl's UTF-8 and Unicode support. Details
2613 of what it does have are given in the section on UTF-8 support in the
2614 main pcre page.
2615
2616 2. PCRE does not allow repeat quantifiers on lookahead assertions. Perl
2617 permits them, but they do not mean what you might think. For example,
2618 (?!a){3} does not assert that the next three characters are not "a". It
2619 just asserts that the next character is not "a" three times.
2620
2621 3. Capturing subpatterns that occur inside negative lookahead asser-
2622 tions are counted, but their entries in the offsets vector are never
2623 set. Perl sets its numerical variables from any such patterns that are
2624 matched before the assertion fails to match something (thereby succeed-
2625 ing), but only if the negative lookahead assertion contains just one
2626 branch.
2627
2628 4. Though binary zero characters are supported in the subject string,
2629 they are not allowed in a pattern string because it is passed as a nor-
2630 mal C string, terminated by zero. The escape sequence \0 can be used in
2631 the pattern to represent a binary zero.
2632
2633 5. The following Perl escape sequences are not supported: \l, \u, \L,
2634 \U, and \N. In fact these are implemented by Perl's general string-han-
2635 dling and are not part of its pattern matching engine. If any of these
2636 are encountered by PCRE, an error is generated.
2637
2638 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
2639 is built with Unicode character property support. The properties that
2640 can be tested with \p and \P are limited to the general category prop-
2641 erties such as Lu and Nd, script names such as Greek or Han, and the
2642 derived properties Any and L&.
2643
2644 7. PCRE does support the \Q...\E escape for quoting substrings. Charac-
2645 ters in between are treated as literals. This is slightly different
2646 from Perl in that $ and @ are also handled as literals inside the
2647 quotes. In Perl, they cause variable interpolation (but of course PCRE
2648 does not have variables). Note the following examples:
2649
2650 Pattern PCRE matches Perl matches
2651
2652 \Qabc$xyz\E abc$xyz abc followed by the
2653 contents of $xyz
2654 \Qabc\$xyz\E abc\$xyz abc\$xyz
2655 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2656
2657 The \Q...\E sequence is recognized both inside and outside character
2658 classes.
2659
2660 8. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
2661 constructions. However, there is support for recursive patterns. This
2662 is not available in Perl 5.8, but will be in Perl 5.10. Also, the PCRE
2663 "callout" feature allows an external function to be called during pat-
2664 tern matching. See the pcrecallout documentation for details.
2665
2666 9. Subpatterns that are called recursively or as "subroutines" are
2667 always treated as atomic groups in PCRE. This is like Python, but
2668 unlike Perl.
2669
2670 10. There are some differences that are concerned with the settings of
2671 captured strings when part of a pattern is repeated. For example,
2672 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
2673 unset, but in PCRE it is set to "b".
2674
2675 11. PCRE provides some extensions to the Perl regular expression facil-
2676 ities. Perl 5.10 will include new features that are not in earlier
2677 versions, some of which (such as named parentheses) have been in PCRE
2678 for some time. This list is with respect to Perl 5.10:
2679
2680 (a) Although lookbehind assertions must match fixed length strings,
2681 each alternative branch of a lookbehind assertion can match a different
2682 length of string. Perl requires them all to have the same length.
2683
2684 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
2685 meta-character matches only at the very end of the string.
2686
2687 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
2688 cial meaning is faulted. Otherwise, like Perl, the backslash is quietly
2689 ignored. (Perl can be made to issue a warning.)
2690
2691 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
2692 fiers is inverted, that is, by default they are not greedy, but if fol-
2693 lowed by a question mark they are.
2694
2695 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
2696 tried only at the first matching position in the subject string.
2697
2698 (f) The PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, and PCRE_NO_AUTO_CAP-
2699 TURE options for pcre_exec() have no Perl equivalents.
2700
2701 (g) The callout facility is PCRE-specific.
2702
2703 (h) The partial matching facility is PCRE-specific.
2704
2705 (i) Patterns compiled by PCRE can be saved and re-used at a later time,
2706 even on different hosts that have the other endianness.
2707
2708 (j) The alternative matching function (pcre_dfa_exec()) matches in a
2709 different way and is not Perl-compatible.
2710
2711
2712 AUTHOR
2713
2714 Philip Hazel
2715 University Computing Service
2716 Cambridge CB2 3QH, England.
2717
2718
2719 REVISION
2720
2721 Last updated: 13 June 2007
2722 Copyright (c) 1997-2007 University of Cambridge.
2723 ------------------------------------------------------------------------------
2724
2725
2726 PCREPATTERN(3) PCREPATTERN(3)
2727
2728
2729 NAME
2730 PCRE - Perl-compatible regular expressions
2731
2732
2733 PCRE REGULAR EXPRESSION DETAILS
2734
2735 The syntax and semantics of the regular expressions supported by PCRE
2736 are described below. Regular expressions are also described in the Perl
2737 documentation and in a number of books, some of which have copious
2738 examples. Jeffrey Friedl's "Mastering Regular Expressions", published
2739 by O'Reilly, covers regular expressions in great detail. This descrip-
2740 tion of PCRE's regular expressions is intended as reference material.
2741
2742 The original operation of PCRE was on strings of one-byte characters.
2743 However, there is now also support for UTF-8 character strings. To use
2744 this, you must build PCRE to include UTF-8 support, and then call
2745 pcre_compile() with the PCRE_UTF8 option. How this affects pattern
2746 matching is mentioned in several places below. There is also a summary
2747 of UTF-8 features in the section on UTF-8 support in the main pcre
2748 page.
2749
2750 The remainder of this document discusses the patterns that are sup-
2751 ported by PCRE when its main matching function, pcre_exec(), is used.
2752 From release 6.0, PCRE offers a second matching function,
2753 pcre_dfa_exec(), which matches using a different algorithm that is not
2754 Perl-compatible. Some of the features discussed below are not available
2755 when pcre_dfa_exec() is used. The advantages and disadvantages of the
2756 alternative function, and how it differs from the normal function, are
2757 discussed in the pcrematching page.
2758
2759
2760 CHARACTERS AND METACHARACTERS
2761
2762 A regular expression is a pattern that is matched against a subject
2763 string from left to right. Most characters stand for themselves in a
2764 pattern, and match the corresponding characters in the subject. As a
2765 trivial example, the pattern
2766
2767 The quick brown fox
2768
2769 matches a portion of a subject string that is identical to itself. When
2770 caseless matching is specified (the PCRE_CASELESS option), letters are
2771 matched independently of case. In UTF-8 mode, PCRE always understands
2772 the concept of case for characters whose values are less than 128, so
2773 caseless matching is always possible. For characters with higher val-
2774 ues, the concept of case is supported if PCRE is compiled with Unicode
2775 property support, but not otherwise. If you want to use caseless
2776 matching for characters 128 and above, you must ensure that PCRE is
2777 compiled with Unicode property support as well as with UTF-8 support.
2778
2779 The power of regular expressions comes from the ability to include
2780 alternatives and repetitions in the pattern. These are encoded in the
2781 pattern by the use of metacharacters, which do not stand for themselves
2782 but instead are interpreted in some special way.
2783
2784 There are two different sets of metacharacters: those that are recog-
2785 nized anywhere in the pattern except within square brackets, and those
2786 that are recognized within square brackets. Outside square brackets,
2787 the metacharacters are as follows:
2788
2789 \ general escape character with several uses
2790 ^ assert start of string (or line, in multiline mode)
2791 $ assert end of string (or line, in multiline mode)
2792 . match any character except newline (by default)
2793 [ start character class definition
2794 | start of alternative branch
2795 ( start subpattern
2796 ) end subpattern
2797 ? extends the meaning of (
2798 also 0 or 1 quantifier
2799 also quantifier minimizer
2800 * 0 or more quantifier
2801 + 1 or more quantifier
2802 also "possessive quantifier"
2803 { start min/max quantifier
2804
2805 Part of a pattern that is in square brackets is called a "character
2806 class". In a character class the only metacharacters are:
2807
2808 \ general escape character
2809 ^ negate the class, but only if the first character
2810 - indicates character range
2811 [ POSIX character class (only if followed by POSIX
2812 syntax)
2813 ] terminates the character class
2814
2815 The following sections describe the use of each of the metacharacters.
2816
2817
2818 BACKSLASH
2819
2820 The backslash character has several uses. Firstly, if it is followed by
2821 a non-alphanumeric character, it takes away any special meaning that
2822 character may have. This use of backslash as an escape character
2823 applies both inside and outside character classes.
2824
2825 For example, if you want to match a * character, you write \* in the
2826 pattern. This escaping action applies whether or not the following
2827 character would otherwise be interpreted as a metacharacter, so it is
2828 always safe to precede a non-alphanumeric with backslash to specify
2829 that it stands for itself. In particular, if you want to match a back-
2830 slash, you write \\.
2831
2832 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
2833 the pattern (other than in a character class) and characters between a
2834 # outside a character class and the next newline are ignored. An escap-
2835 ing backslash can be used to include a whitespace or # character as
2836 part of the pattern.
2837
2838 If you want to remove the special meaning from a sequence of charac-
2839 ters, you can do so by putting them between \Q and \E. This is differ-
2840 ent from Perl in that $ and @ are handled as literals in \Q...\E
2841 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
2842 tion. Note the following examples:
2843
2844 Pattern PCRE matches Perl matches
2845
2846 \Qabc$xyz\E abc$xyz abc followed by the
2847 contents of $xyz
2848 \Qabc\$xyz\E abc\$xyz abc\$xyz
2849 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
2850
2851 The \Q...\E sequence is recognized both inside and outside character
2852 classes.
2853
2854 Non-printing characters
2855
2856 A second use of backslash provides a way of encoding non-printing char-
2857 acters in patterns in a visible manner. There is no restriction on the
2858 appearance of non-printing characters, apart from the binary zero that
2859 terminates a pattern, but when a pattern is being prepared by text
2860 editing, it is usually easier to use one of the following escape
2861 sequences than the binary character it represents:
2862
2863 \a alarm, that is, the BEL character (hex 07)
2864 \cx "control-x", where x is any character
2865 \e escape (hex 1B)
2866 \f formfeed (hex 0C)
2867 \n newline (hex 0A)
2868 \r carriage return (hex 0D)
2869 \t tab (hex 09)
2870 \ddd character with octal code ddd, or backreference
2871 \xhh character with hex code hh
2872 \x{hhh..} character with hex code hhh..
2873
2874 The precise effect of \cx is as follows: if x is a lower case letter,
2875 it is converted to upper case. Then bit 6 of the character (hex 40) is
2876 inverted. Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c;
2877 becomes hex 7B.
2878
2879 After \x, from zero to two hexadecimal digits are read (letters can be
2880 in upper or lower case). Any number of hexadecimal digits may appear
2881 between \x{ and }, but the value of the character code must be less
2882 than 256 in non-UTF-8 mode, and less than 2**31 in UTF-8 mode (that is,
2883 the maximum hexadecimal value is 7FFFFFFF). If characters other than
2884 hexadecimal digits appear between \x{ and }, or if there is no termi-
2885 nating }, this form of escape is not recognized. Instead, the initial
2886 \x will be interpreted as a basic hexadecimal escape, with no following
2887 digits, giving a character whose value is zero.
2888
2889 Characters whose value is less than 256 can be defined by either of the
2890 two syntaxes for \x. There is no difference in the way they are han-
2891 dled. For example, \xdc is exactly the same as \x{dc}.
2892
2893 After \0 up to two further octal digits are read. If there are fewer
2894 than two digits, just those that are present are used. Thus the
2895 sequence \0\x\07 specifies two binary zeros followed by a BEL character
2896 (code value 7). Make sure you supply two digits after the initial zero
2897 if the pattern character that follows is itself an octal digit.
2898
2899 The handling of a backslash followed by a digit other than 0 is compli-
2900 cated. Outside a character class, PCRE reads it and any following dig-
2901 its as a decimal number. If the number is less than 10, or if there
2902 have been at least that many previous capturing left parentheses in the
2903 expression, the entire sequence is taken as a back reference. A
2904 description of how this works is given later, following the discussion
2905 of parenthesized subpatterns.
2906
2907 Inside a character class, or if the decimal number is greater than 9
2908 and there have not been that many capturing subpatterns, PCRE re-reads
2909 up to three octal digits following the backslash, and uses them to gen-
2910 erate a data character. Any subsequent digits stand for themselves. In
2911 non-UTF-8 mode, the value of a character specified in octal must be
2912 less than \400. In UTF-8 mode, values up to \777 are permitted. For
2913 example:
2914
2915 \040 is another way of writing a space
2916 \40 is the same, provided there are fewer than 40
2917 previous capturing subpatterns
2918 \7 is always a back reference
2919 \11 might be a back reference, or another way of
2920 writing a tab
2921 \011 is always a tab
2922 \0113 is a tab followed by the character "3"
2923 \113 might be a back reference, otherwise the
2924 character with octal code 113
2925 \377 might be a back reference, otherwise
2926 the byte consisting entirely of 1 bits
2927 \81 is either a back reference, or a binary zero
2928 followed by the two characters "8" and "1"
2929
2930 Note that octal values of 100 or greater must not be introduced by a
2931 leading zero, because no more than three octal digits are ever read.
2932
2933 All the sequences that define a single character value can be used both
2934 inside and outside character classes. In addition, inside a character
2935 class, the sequence \b is interpreted as the backspace character (hex
2936 08), and the sequences \R and \X are interpreted as the characters "R"
2937 and "X", respectively. Outside a character class, these sequences have
2938 different meanings (see below).
2939
2940 Absolute and relative back references
2941
2942 The sequence \g followed by a positive or negative number, optionally
2943 enclosed in braces, is an absolute or relative back reference. A named
2944 back reference can be coded as \g{name}. Back references are discussed
2945 later, following the discussion of parenthesized subpatterns.
2946
2947 Generic character types
2948
2949 Another use of backslash is for specifying generic character types. The
2950 following are always recognized:
2951
2952 \d any decimal digit
2953 \D any character that is not a decimal digit
2954 \h any horizontal whitespace character
2955 \H any character that is not a horizontal whitespace character
2956 \s any whitespace character
2957 \S any character that is not a whitespace character
2958 \v any vertical whitespace character
2959 \V any character that is not a vertical whitespace character
2960 \w any "word" character
2961 \W any "non-word" character
2962
2963 Each pair of escape sequences partitions the complete set of characters
2964 into two disjoint sets. Any given character matches one, and only one,
2965 of each pair.
2966
2967 These character type sequences can appear both inside and outside char-
2968 acter classes. They each match one character of the appropriate type.
2969 If the current matching point is at the end of the subject string, all
2970 of them fail, since there is no character to match.
2971
2972 For compatibility with Perl, \s does not match the VT character (code
2973 11). This makes it different from the the POSIX "space" class. The \s
2974 characters are HT (9), LF (10), FF (12), CR (13), and space (32). If
2975 "use locale;" is included in a Perl script, \s may match the VT charac-
2976 ter. In PCRE, it never does.
2977
2978 In UTF-8 mode, characters with values greater than 128 never match \d,
2979 \s, or \w, and always match \D, \S, and \W. This is true even when Uni-
2980 code character property support is available. These sequences retain
2981 their original meanings from before UTF-8 support was available, mainly
2982 for efficiency reasons.
2983
2984 The sequences \h, \H, \v, and \V are Perl 5.10 features. In contrast to
2985 the other sequences, these do match certain high-valued codepoints in
2986 UTF-8 mode. The horizontal space characters are:
2987
2988 U+0009 Horizontal tab
2989 U+0020 Space
2990 U+00A0 Non-break space
2991 U+1680 Ogham space mark
2992 U+180E Mongolian vowel separator
2993 U+2000 En quad
2994 U+2001 Em quad
2995 U+2002 En space
2996 U+2003 Em space
2997 U+2004 Three-per-em space
2998 U+2005 Four-per-em space
2999 U+2006 Six-per-em space
3000 U+2007 Figure space
3001 U+2008 Punctuation space
3002 U+2009 Thin space
3003 U+200A Hair space
3004 U+202F Narrow no-break space
3005 U+205F Medium mathematical space
3006 U+3000 Ideographic space
3007
3008 The vertical space characters are:
3009
3010 U+000A Linefeed
3011 U+000B Vertical tab
3012 U+000C Formfeed
3013 U+000D Carriage return
3014 U+0085 Next line
3015 U+2028 Line separator
3016 U+2029 Paragraph separator
3017
3018 A "word" character is an underscore or any character less than 256 that
3019 is a letter or digit. The definition of letters and digits is con-
3020 trolled by PCRE's low-valued character tables, and may vary if locale-
3021 specific matching is taking place (see "Locale support" in the pcreapi
3022 page). For example, in a French locale such as "fr_FR" in Unix-like
3023 systems, or "french" in Windows, some character codes greater than 128
3024 are used for accented letters, and these are matched by \w. The use of
3025 locales with Unicode is discouraged.
3026
3027 Newline sequences
3028
3029 Outside a character class, the escape sequence \R matches any Unicode
3030 newline sequence. This is a Perl 5.10 feature. In non-UTF-8 mode \R is
3031 equivalent to the following:
3032
3033 (?>\r\n|\n|\x0b|\f|\r|\x85)
3034
3035 This is an example of an "atomic group", details of which are given
3036 below. This particular group matches either the two-character sequence
3037 CR followed by LF, or one of the single characters LF (linefeed,
3038 U+000A), VT (vertical tab, U+000B), FF (formfeed, U+000C), CR (carriage
3039 return, U+000D), or NEL (next line, U+0085). The two-character sequence
3040 is treated as a single unit that cannot be split.
3041
3042 In UTF-8 mode, two additional characters whose codepoints are greater
3043 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
3044 rator, U+2029). Unicode character property support is not needed for
3045 these characters to be recognized.
3046
3047 Inside a character class, \R matches the letter "R".
3048
3049 Unicode character properties
3050
3051 When PCRE is built with Unicode character property support, three addi-
3052 tional escape sequences that match characters with specific properties
3053 are available. When not in UTF-8 mode, these sequences are of course
3054 limited to testing characters whose codepoints are less than 256, but
3055 they do work in this mode. The extra escape sequences are:
3056
3057 \p{xx} a character with the xx property
3058 \P{xx} a character without the xx property
3059 \X an extended Unicode sequence
3060
3061 The property names represented by xx above are limited to the Unicode
3062 script names, the general category properties, and "Any", which matches
3063 any character (including newline). Other properties such as "InMusical-
3064 Symbols" are not currently supported by PCRE. Note that \P{Any} does
3065 not match any characters, so always causes a match failure.
3066
3067 Sets of Unicode characters are defined as belonging to certain scripts.
3068 A character from one of these sets can be matched using a script name.
3069 For example:
3070
3071 \p{Greek}
3072 \P{Han}
3073
3074 Those that are not part of an identified script are lumped together as
3075 "Common". The current list of scripts is:
3076
3077 Arabic, Armenian, Balinese, Bengali, Bopomofo, Braille, Buginese,
3078 Buhid, Canadian_Aboriginal, Cherokee, Common, Coptic, Cuneiform,
3079 Cypriot, Cyrillic, Deseret, Devanagari, Ethiopic, Georgian, Glagolitic,
3080 Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
3081 gana, Inherited, Kannada, Katakana, Kharoshthi, Khmer, Lao, Latin,
3082 Limbu, Linear_B, Malayalam, Mongolian, Myanmar, New_Tai_Lue, Nko,
3083 Ogham, Old_Italic, Old_Persian, Oriya, Osmanya, Phags_Pa, Phoenician,
3084 Runic, Shavian, Sinhala, Syloti_Nagri, Syriac, Tagalog, Tagbanwa,
3085 Tai_Le, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Yi.
3086
3087 Each character has exactly one general category property, specified by
3088 a two-letter abbreviation. For compatibility with Perl, negation can be
3089 specified by including a circumflex between the opening brace and the
3090 property name. For example, \p{^Lu} is the same as \P{Lu}.
3091
3092 If only one letter is specified with \p or \P, it includes all the gen-
3093 eral category properties that start with that letter. In this case, in
3094 the absence of negation, the curly brackets in the escape sequence are
3095 optional; these two examples have the same effect:
3096
3097 \p{L}
3098 \pL
3099
3100 The following general category property codes are supported:
3101
3102 C Other
3103 Cc Control
3104 Cf Format
3105 Cn Unassigned
3106 Co Private use
3107 Cs Surrogate
3108
3109 L Letter
3110 Ll Lower case letter
3111 Lm Modifier letter
3112 Lo Other letter
3113 Lt Title case letter
3114 Lu Upper case letter
3115
3116 M Mark
3117 Mc Spacing mark
3118 Me Enclosing mark
3119 Mn Non-spacing mark
3120
3121 N Number
3122 Nd Decimal number
3123 Nl Letter number
3124 No Other number
3125
3126 P Punctuation
3127 Pc Connector punctuation
3128 Pd Dash punctuation
3129 Pe Close punctuation
3130 Pf Final punctuation
3131 Pi Initial punctuation
3132 Po Other punctuation
3133 Ps Open punctuation
3134
3135 S Symbol
3136 Sc Currency symbol
3137 Sk Modifier symbol
3138 Sm Mathematical symbol
3139 So Other symbol
3140
3141 Z Separator
3142 Zl Line separator
3143 Zp Paragraph separator
3144 Zs Space separator
3145
3146 The special property L& is also supported: it matches a character that
3147 has the Lu, Ll, or Lt property, in other words, a letter that is not
3148 classified as a modifier or "other".
3149
3150 The long synonyms for these properties that Perl supports (such as
3151 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
3152 any of these properties with "Is".
3153
3154 No character that is in the Unicode table has the Cn (unassigned) prop-
3155 erty. Instead, this property is assumed for any code point that is not
3156 in the Unicode table.
3157
3158 Specifying caseless matching does not affect these escape sequences.
3159 For example, \p{Lu} always matches only upper case letters.
3160
3161 The \X escape matches any number of Unicode characters that form an
3162 extended Unicode sequence. \X is equivalent to
3163
3164 (?>\PM\pM*)
3165
3166 That is, it matches a character without the "mark" property, followed
3167 by zero or more characters with the "mark" property, and treats the
3168 sequence as an atomic group (see below). Characters with the "mark"
3169 property are typically accents that affect the preceding character.
3170 None of them have codepoints less than 256, so in non-UTF-8 mode \X
3171 matches any one character.
3172
3173 Matching characters by Unicode property is not fast, because PCRE has
3174 to search a structure that contains data for over fifteen thousand
3175 characters. That is why the traditional escape sequences such as \d and
3176 \w do not use Unicode properties in PCRE.
3177
3178 Resetting the match start
3179
3180 The escape sequence \K, which is a Perl 5.10 feature, causes any previ-
3181 ously matched characters not to be included in the final matched
3182 sequence. For example, the pattern:
3183
3184 foo\Kbar
3185
3186 matches "foobar", but reports that it has matched "bar". This feature
3187 is similar to a lookbehind assertion (described below). However, in
3188 this case, the part of the subject before the real match does not have
3189 to be of fixed length, as lookbehind assertions do. The use of \K does
3190 not interfere with the setting of captured substrings. For example,
3191 when the pattern
3192
3193 (foo)\Kbar
3194
3195 matches "foobar", the first substring is still set to "foo".
3196
3197 Simple assertions
3198
3199 The final use of backslash is for certain simple assertions. An asser-
3200 tion specifies a condition that has to be met at a particular point in
3201 a match, without consuming any characters from the subject string. The
3202 use of subpatterns for more complicated assertions is described below.
3203 The backslashed assertions are:
3204
3205 \b matches at a word boundary
3206 \B matches when not at a word boundary
3207 \A matches at the start of the subject
3208 \Z matches at the end of the subject
3209 also matches before a newline at the end of the subject
3210 \z matches only at the end of the subject
3211 \G matches at the first matching position in the subject
3212
3213 These assertions may not appear in character classes (but note that \b
3214 has a different meaning, namely the backspace character, inside a char-
3215 acter class).
3216
3217 A word boundary is a position in the subject string where the current
3218 character and the previous character do not both match \w or \W (i.e.
3219 one matches \w and the other matches \W), or the start or end of the
3220 string if the first or last character matches \w, respectively.
3221
3222 The \A, \Z, and \z assertions differ from the traditional circumflex
3223 and dollar (described in the next section) in that they only ever match
3224 at the very start and end of the subject string, whatever options are
3225 set. Thus, they are independent of multiline mode. These three asser-
3226 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
3227 affect only the behaviour of the circumflex and dollar metacharacters.
3228 However, if the startoffset argument of pcre_exec() is non-zero, indi-
3229 cating that matching is to start at a point other than the beginning of
3230 the subject, \A can never match. The difference between \Z and \z is
3231 that \Z matches before a newline at the end of the string as well as at
3232 the very end, whereas \z matches only at the end.
3233
3234 The \G assertion is true only when the current matching position is at
3235 the start point of the match, as specified by the startoffset argument
3236 of pcre_exec(). It differs from \A when the value of startoffset is
3237 non-zero. By calling pcre_exec() multiple times with appropriate argu-
3238 ments, you can mimic Perl's /g option, and it is in this kind of imple-
3239 mentation where \G can be useful.
3240
3241 Note, however, that PCRE's interpretation of \G, as the start of the
3242 current match, is subtly different from Perl's, which defines it as the
3243 end of the previous match. In Perl, these can be different when the
3244 previously matched string was empty. Because PCRE does just one match
3245 at a time, it cannot reproduce this behaviour.
3246
3247 If all the alternatives of a pattern begin with \G, the expression is
3248 anchored to the starting match position, and the "anchored" flag is set
3249 in the compiled regular expression.
3250
3251
3252 CIRCUMFLEX AND DOLLAR
3253
3254 Outside a character class, in the default matching mode, the circumflex
3255 character is an assertion that is true only if the current matching
3256 point is at the start of the subject string. If the startoffset argu-
3257 ment of pcre_exec() is non-zero, circumflex can never match if the
3258 PCRE_MULTILINE option is unset. Inside a character class, circumflex
3259 has an entirely different meaning (see below).
3260
3261 Circumflex need not be the first character of the pattern if a number
3262 of alternatives are involved, but it should be the first thing in each
3263 alternative in which it appears if the pattern is ever to match that
3264 branch. If all possible alternatives start with a circumflex, that is,
3265 if the pattern is constrained to match only at the start of the sub-
3266 ject, it is said to be an "anchored" pattern. (There are also other
3267 constructs that can cause a pattern to be anchored.)
3268
3269 A dollar character is an assertion that is true only if the current
3270 matching point is at the end of the subject string, or immediately
3271 before a newline at the end of the string (by default). Dollar need not
3272 be the last character of the pattern if a number of alternatives are
3273 involved, but it should be the last item in any branch in which it
3274 appears. Dollar has no special meaning in a character class.
3275
3276 The meaning of dollar can be changed so that it matches only at the
3277 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
3278 compile time. This does not affect the \Z assertion.
3279
3280 The meanings of the circumflex and dollar characters are changed if the
3281 PCRE_MULTILINE option is set. When this is the case, a circumflex
3282 matches immediately after internal newlines as well as at the start of
3283 the subject string. It does not match after a newline that ends the
3284 string. A dollar matches before any newlines in the string, as well as
3285 at the very end, when PCRE_MULTILINE is set. When newline is specified
3286 as the two-character sequence CRLF, isolated CR and LF characters do
3287 not indicate newlines.
3288
3289 For example, the pattern /^abc$/ matches the subject string "def\nabc"
3290 (where \n represents a newline) in multiline mode, but not otherwise.
3291 Consequently, patterns that are anchored in single line mode because
3292 all branches start with ^ are not anchored in multiline mode, and a
3293 match for circumflex is possible when the startoffset argument of
3294 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
3295 PCRE_MULTILINE is set.
3296
3297 Note that the sequences \A, \Z, and \z can be used to match the start
3298 and end of the subject in both modes, and if all branches of a pattern
3299 start with \A it is always anchored, whether or not PCRE_MULTILINE is
3300 set.
3301
3302
3303 FULL STOP (PERIOD, DOT)
3304
3305 Outside a character class, a dot in the pattern matches any one charac-
3306 ter in the subject string except (by default) a character that signi-
3307 fies the end of a line. In UTF-8 mode, the matched character may be
3308 more than one byte long.
3309
3310 When a line ending is defined as a single character, dot never matches
3311 that character; when the two-character sequence CRLF is used, dot does
3312 not match CR if it is immediately followed by LF, but otherwise it
3313 matches all characters (including isolated CRs and LFs). When any Uni-
3314 code line endings are being recognized, dot does not match CR or LF or
3315 any of the other line ending characters.
3316
3317 The behaviour of dot with regard to newlines can be changed. If the
3318 PCRE_DOTALL option is set, a dot matches any one character, without
3319 exception. If the two-character sequence CRLF is present in the subject
3320 string, it takes two dots to match it.
3321
3322 The handling of dot is entirely independent of the handling of circum-
3323 flex and dollar, the only relationship being that they both involve
3324 newlines. Dot has no special meaning in a character class.
3325
3326
3327 MATCHING A SINGLE BYTE
3328
3329 Outside a character class, the escape sequence \C matches any one byte,
3330 both in and out of UTF-8 mode. Unlike a dot, it always matches any
3331 line-ending characters. The feature is provided in Perl in order to
3332 match individual bytes in UTF-8 mode. Because it breaks up UTF-8 char-
3333 acters into individual bytes, what remains in the string may be a mal-
3334 formed UTF-8 string. For this reason, the \C escape sequence is best
3335 avoided.
3336
3337 PCRE does not allow \C to appear in lookbehind assertions (described
3338 below), because in UTF-8 mode this would make it impossible to calcu-
3339 late the length of the lookbehind.
3340
3341
3342 SQUARE BRACKETS AND CHARACTER CLASSES
3343
3344 An opening square bracket introduces a character class, terminated by a
3345 closing square bracket. A closing square bracket on its own is not spe-
3346 cial. If a closing square bracket is required as a member of the class,
3347 it should be the first data character in the class (after an initial
3348 circumflex, if present) or escaped with a backslash.
3349
3350 A character class matches a single character in the subject. In UTF-8
3351 mode, the character may occupy more than one byte. A matched character
3352 must be in the set of characters defined by the class, unless the first
3353 character in the class definition is a circumflex, in which case the
3354 subject character must not be in the set defined by the class. If a
3355 circumflex is actually required as a member of the class, ensure it is
3356 not the first character, or escape it with a backslash.
3357
3358 For example, the character class [aeiou] matches any lower case vowel,
3359 while [^aeiou] matches any character that is not a lower case vowel.
3360 Note that a circumflex is just a convenient notation for specifying the
3361 characters that are in the class by enumerating those that are not. A
3362 class that starts with a circumflex is not an assertion: it still con-
3363 sumes a character from the subject string, and therefore it fails if
3364 the current pointer is at the end of the string.
3365
3366 In UTF-8 mode, characters with values greater than 255 can be included
3367 in a class as a literal string of bytes, or by using the \x{ escaping
3368 mechanism.
3369
3370 When caseless matching is set, any letters in a class represent both
3371 their upper case and lower case versions, so for example, a caseless
3372 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
3373 match "A", whereas a caseful version would. In UTF-8 mode, PCRE always
3374 understands the concept of case for characters whose values are less
3375 than 128, so caseless matching is always possible. For characters with
3376 higher values, the concept of case is supported if PCRE is compiled
3377 with Unicode property support, but not otherwise. If you want to use
3378 caseless matching for characters 128 and above, you must ensure that
3379 PCRE is compiled with Unicode property support as well as with UTF-8
3380 support.
3381
3382 Characters that might indicate line breaks are never treated in any
3383 special way when matching character classes, whatever line-ending
3384 sequence is in use, and whatever setting of the PCRE_DOTALL and
3385 PCRE_MULTILINE options is used. A class such as [^a] always matches one
3386 of these characters.
3387
3388 The minus (hyphen) character can be used to specify a range of charac-
3389 ters in a character class. For example, [d-m] matches any letter
3390 between d and m, inclusive. If a minus character is required in a
3391 class, it must be escaped with a backslash or appear in a position
3392 where it cannot be interpreted as indicating a range, typically as the
3393 first or last character in the class.
3394
3395 It is not possible to have the literal character "]" as the end charac-
3396 ter of a range. A pattern such as [W-]46] is interpreted as a class of
3397 two characters ("W" and "-") followed by a literal string "46]", so it
3398 would match "W46]" or "-46]". However, if the "]" is escaped with a
3399 backslash it is interpreted as the end of range, so [W-\]46] is inter-
3400 preted as a class containing a range followed by two other characters.
3401 The octal or hexadecimal representation of "]" can also be used to end
3402 a range.
3403
3404 Ranges operate in the collating sequence of character values. They can
3405 also be used for characters specified numerically, for example
3406 [\000-\037]. In UTF-8 mode, ranges can include characters whose values
3407 are greater than 255, for example [\x{100}-\x{2ff}].
3408
3409 If a range that includes letters is used when caseless matching is set,
3410 it matches the letters in either case. For example, [W-c] is equivalent
3411 to [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if
3412 character tables for a French locale are in use, [\xc8-\xcb] matches
3413 accented E characters in both cases. In UTF-8 mode, PCRE supports the
3414 concept of case for characters with values greater than 128 only when
3415 it is compiled with Unicode property support.
3416
3417 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
3418 in a character class, and add the characters that they match to the
3419 class. For example, [\dABCDEF] matches any hexadecimal digit. A circum-
3420 flex can conveniently be used with the upper case character types to
3421 specify a more restricted set of characters than the matching lower
3422 case type. For example, the class [^\W_] matches any letter or digit,
3423 but not underscore.
3424
3425 The only metacharacters that are recognized in character classes are
3426 backslash, hyphen (only where it can be interpreted as specifying a
3427 range), circumflex (only at the start), opening square bracket (only
3428 when it can be interpreted as introducing a POSIX class name - see the
3429 next section), and the terminating closing square bracket. However,
3430 escaping other non-alphanumeric characters does no harm.
3431
3432
3433 POSIX CHARACTER CLASSES
3434
3435 Perl supports the POSIX notation for character classes. This uses names
3436 enclosed by [: and :] within the enclosing square brackets. PCRE also
3437 supports this notation. For example,
3438
3439 [01[:alpha:]%]
3440
3441 matches "0", "1", any alphabetic character, or "%". The supported class
3442 names are
3443
3444 alnum letters and digits
3445 alpha letters
3446 ascii character codes 0 - 127
3447 blank space or tab only
3448 cntrl control characters
3449 digit decimal digits (same as \d)
3450 graph printing characters, excluding space
3451 lower lower case letters
3452 print printing characters, including space
3453 punct printing characters, excluding letters and digits
3454 space white space (not quite the same as \s)
3455 upper upper case letters
3456 word "word" characters (same as \w)
3457 xdigit hexadecimal digits
3458
3459 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
3460 and space (32). Notice that this list includes the VT character (code
3461 11). This makes "space" different to \s, which does not include VT (for
3462 Perl compatibility).
3463
3464 The name "word" is a Perl extension, and "blank" is a GNU extension
3465 from Perl 5.8. Another Perl extension is negation, which is indicated
3466 by a ^ character after the colon. For example,
3467
3468 [12[:^digit:]]
3469
3470 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
3471 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
3472 these are not supported, and an error is given if they are encountered.
3473
3474 In UTF-8 mode, characters with values greater than 128 do not match any
3475 of the POSIX character classes.
3476
3477
3478 VERTICAL BAR
3479
3480 Vertical bar characters are used to separate alternative patterns. For
3481 example, the pattern
3482
3483 gilbert|sullivan
3484
3485 matches either "gilbert" or "sullivan". Any number of alternatives may
3486 appear, and an empty alternative is permitted (matching the empty
3487 string). The matching process tries each alternative in turn, from left
3488 to right, and the first one that succeeds is used. If the alternatives
3489 are within a subpattern (defined below), "succeeds" means matching the
3490 rest of the main pattern as well as the alternative in the subpattern.
3491
3492
3493 INTERNAL OPTION SETTING
3494
3495 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
3496 PCRE_EXTENDED options can be changed from within the pattern by a
3497 sequence of Perl option letters enclosed between "(?" and ")". The
3498 option letters are
3499
3500 i for PCRE_CASELESS
3501 m for PCRE_MULTILINE
3502 s for PCRE_DOTALL
3503 x for PCRE_EXTENDED
3504
3505 For example, (?im) sets caseless, multiline matching. It is also possi-
3506 ble to unset these options by preceding the letter with a hyphen, and a
3507 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
3508 LESS and PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED,
3509 is also permitted. If a letter appears both before and after the
3510 hyphen, the option is unset.
3511
3512 When an option change occurs at top level (that is, not inside subpat-
3513 tern parentheses), the change applies to the remainder of the pattern
3514 that follows. If the change is placed right at the start of a pattern,
3515 PCRE extracts it into the global options (and it will therefore show up
3516 in data extracted by the pcre_fullinfo() function).
3517
3518 An option change within a subpattern (see below for a description of
3519 subpatterns) affects only that part of the current pattern that follows
3520 it, so
3521
3522 (a(?i)b)c
3523
3524 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
3525 used). By this means, options can be made to have different settings
3526 in different parts of the pattern. Any changes made in one alternative
3527 do carry on into subsequent branches within the same subpattern. For
3528 example,
3529
3530 (a(?i)b|c)
3531
3532 matches "ab", "aB", "c", and "C", even though when matching "C" the
3533 first branch is abandoned before the option setting. This is because
3534 the effects of option settings happen at compile time. There would be
3535 some very weird behaviour otherwise.
3536
3537 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA
3538 can be changed in the same way as the Perl-compatible options by using
3539 the characters J, U and X respectively.
3540
3541
3542 SUBPATTERNS
3543
3544 Subpatterns are delimited by parentheses (round brackets), which can be
3545 nested. Turning part of a pattern into a subpattern does two things:
3546
3547 1. It localizes a set of alternatives. For example, the pattern
3548
3549 cat(aract|erpillar|)
3550
3551 matches one of the words "cat", "cataract", or "caterpillar". Without
3552 the parentheses, it would match "cataract", "erpillar" or an empty
3553 string.
3554
3555 2. It sets up the subpattern as a capturing subpattern. This means
3556 that, when the whole pattern matches, that portion of the subject
3557 string that matched the subpattern is passed back to the caller via the
3558 ovector argument of pcre_exec(). Opening parentheses are counted from
3559 left to right (starting from 1) to obtain numbers for the capturing
3560 subpatterns.
3561
3562 For example, if the string "the red king" is matched against the pat-
3563 tern
3564
3565 the ((red|white) (king|queen))
3566
3567 the captured substrings are "red king", "red", and "king", and are num-
3568 bered 1, 2, and 3, respectively.
3569
3570 The fact that plain parentheses fulfil two functions is not always
3571 helpful. There are often times when a grouping subpattern is required
3572 without a capturing requirement. If an opening parenthesis is followed
3573 by a question mark and a colon, the subpattern does not do any captur-
3574 ing, and is not counted when computing the number of any subsequent
3575 capturing subpatterns. For example, if the string "the white queen" is
3576 matched against the pattern
3577
3578 the ((?:red|white) (king|queen))
3579
3580 the captured substrings are "white queen" and "queen", and are numbered
3581 1 and 2. The maximum number of capturing subpatterns is 65535.
3582
3583 As a convenient shorthand, if any option settings are required at the
3584 start of a non-capturing subpattern, the option letters may appear
3585 between the "?" and the ":". Thus the two patterns
3586
3587 (?i:saturday|sunday)
3588 (?:(?i)saturday|sunday)
3589
3590 match exactly the same set of strings. Because alternative branches are
3591 tried from left to right, and options are not reset until the end of
3592 the subpattern is reached, an option setting in one branch does affect
3593 subsequent branches, so the above patterns match "SUNDAY" as well as
3594 "Saturday".
3595
3596
3597 DUPLICATE SUBPATTERN NUMBERS
3598
3599 Perl 5.10 introduced a feature whereby each alternative in a subpattern
3600 uses the same numbers for its capturing parentheses. Such a subpattern
3601 starts with (?| and is itself a non-capturing subpattern. For example,
3602 consider this pattern:
3603
3604 (?|(Sat)ur|(Sun))day
3605
3606 Because the two alternatives are inside a (?| group, both sets of cap-
3607 turing parentheses are numbered one. Thus, when the pattern matches,
3608 you can look at captured substring number one, whichever alternative
3609 matched. This construct is useful when you want to capture part, but
3610 not all, of one of a number of alternatives. Inside a (?| group, paren-
3611 theses are numbered as usual, but the number is reset at the start of
3612 each branch. The numbers of any capturing buffers that follow the sub-
3613 pattern start after the highest number used in any branch. The follow-
3614 ing example is taken from the Perl documentation. The numbers under-
3615 neath show in which buffer the captured content will be stored.
3616
3617 # before ---------------branch-reset----------- after
3618 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
3619 # 1 2 2 3 2 3 4
3620
3621 A backreference or a recursive call to a numbered subpattern always
3622 refers to the first one in the pattern with the given number.
3623
3624 An alternative approach to using this "branch reset" feature is to use
3625 duplicate named subpatterns, as described in the next section.
3626
3627
3628 NAMED SUBPATTERNS
3629
3630 Identifying capturing parentheses by number is simple, but it can be
3631 very hard to keep track of the numbers in complicated regular expres-
3632 sions. Furthermore, if an expression is modified, the numbers may
3633 change. To help with this difficulty, PCRE supports the naming of sub-
3634 patterns. This feature was not added to Perl until release 5.10. Python
3635 had the feature earlier, and PCRE introduced it at release 4.0, using
3636 the Python syntax. PCRE now supports both the Perl and the Python syn-
3637 tax.
3638
3639 In PCRE, a subpattern can be named in one of three ways: (?<name>...)
3640 or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
3641 to capturing parentheses from other parts of the pattern, such as back-
3642 references, recursion, and conditions, can be made by name as well as
3643 by number.
3644
3645 Names consist of up to 32 alphanumeric characters and underscores.
3646 Named capturing parentheses are still allocated numbers as well as
3647 names, exactly as if the names were not present. The PCRE API provides
3648 function calls for extracting the name-to-number translation table from
3649 a compiled pattern. There is also a convenience function for extracting
3650 a captured substring by name.
3651
3652 By default, a name must be unique within a pattern, but it is possible
3653 to relax this constraint by setting the PCRE_DUPNAMES option at compile
3654 time. This can be useful for patterns where only one instance of the
3655 named parentheses can match. Suppose you want to match the name of a
3656 weekday, either as a 3-letter abbreviation or as the full name, and in
3657 both cases you want to extract the abbreviation. This pattern (ignoring
3658 the line breaks) does the job:
3659
3660 (?<DN>Mon|Fri|Sun)(?:day)?|
3661 (?<DN>Tue)(?:sday)?|
3662 (?<DN>Wed)(?:nesday)?|
3663 (?<DN>Thu)(?:rsday)?|
3664 (?<DN>Sat)(?:urday)?
3665
3666 There are five capturing substrings, but only one is ever set after a
3667 match. (An alternative way of solving this problem is to use a "branch
3668 reset" subpattern, as described in the previous section.)
3669
3670 The convenience function for extracting the data by name returns the
3671 substring for the first (and in this example, the only) subpattern of
3672 that name that matched. This saves searching to find which numbered
3673 subpattern it was. If you make a reference to a non-unique named sub-
3674 pattern from elsewhere in the pattern, the one that corresponds to the
3675 lowest number is used. For further details of the interfaces for han-
3676 dling named subpatterns, see the pcreapi documentation.
3677
3678
3679 REPETITION
3680
3681 Repetition is specified by quantifiers, which can follow any of the
3682 following items:
3683
3684 a literal data character
3685 the dot metacharacter
3686 the \C escape sequence
3687 the \X escape sequence (in UTF-8 mode with Unicode properties)
3688 the \R escape sequence
3689 an escape such as \d that matches a single character
3690 a character class
3691 a back reference (see next section)
3692 a parenthesized subpattern (unless it is an assertion)
3693
3694 The general repetition quantifier specifies a minimum and maximum num-
3695 ber of permitted matches, by giving the two numbers in curly brackets
3696 (braces), separated by a comma. The numbers must be less than 65536,
3697 and the first must be less than or equal to the second. For example:
3698
3699 z{2,4}
3700
3701 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
3702 special character. If the second number is omitted, but the comma is
3703 present, there is no upper limit; if the second number and the comma
3704 are both omitted, the quantifier specifies an exact number of required
3705 matches. Thus
3706
3707 [aeiou]{3,}
3708
3709 matches at least 3 successive vowels, but may match many more, while
3710
3711 \d{8}
3712
3713 matches exactly 8 digits. An opening curly bracket that appears in a
3714 position where a quantifier is not allowed, or one that does not match
3715 the syntax of a quantifier, is taken as a literal character. For exam-
3716 ple, {,6} is not a quantifier, but a literal string of four characters.
3717
3718 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to
3719 individual bytes. Thus, for example, \x{100}{2} matches two UTF-8 char-
3720 acters, each of which is represented by a two-byte sequence. Similarly,
3721 when Unicode property support is available, \X{3} matches three Unicode
3722 extended sequences, each of which may be several bytes long (and they
3723 may be of different lengths).
3724
3725 The quantifier {0} is permitted, causing the expression to behave as if
3726 the previous item and the quantifier were not present.
3727
3728 For convenience, the three most common quantifiers have single-charac-
3729 ter abbreviations:
3730
3731 * is equivalent to {0,}
3732 + is equivalent to {1,}
3733 ? is equivalent to {0,1}
3734
3735 It is possible to construct infinite loops by following a subpattern
3736 that can match no characters with a quantifier that has no upper limit,
3737 for example:
3738
3739 (a?)*
3740
3741 Earlier versions of Perl and PCRE used to give an error at compile time
3742 for such patterns. However, because there are cases where this can be
3743 useful, such patterns are now accepted, but if any repetition of the
3744 subpattern does in fact match no characters, the loop is forcibly bro-
3745 ken.
3746
3747 By default, the quantifiers are "greedy", that is, they match as much
3748 as possible (up to the maximum number of permitted times), without
3749 causing the rest of the pattern to fail. The classic example of where
3750 this gives problems is in trying to match comments in C programs. These
3751 appear between /* and */ and within the comment, individual * and /
3752 characters may appear. An attempt to match C comments by applying the
3753 pattern
3754
3755 /\*.*\*/
3756
3757 to the string
3758
3759 /* first comment */ not comment /* second comment */
3760
3761 fails, because it matches the entire string owing to the greediness of
3762 the .* item.
3763
3764 However, if a quantifier is followed by a question mark, it ceases to
3765 be greedy, and instead matches the minimum number of times possible, so
3766 the pattern
3767
3768 /\*.*?\*/
3769
3770 does the right thing with the C comments. The meaning of the various
3771 quantifiers is not otherwise changed, just the preferred number of
3772 matches. Do not confuse this use of question mark with its use as a
3773 quantifier in its own right. Because it has two uses, it can sometimes
3774 appear doubled, as in
3775
3776 \d??\d
3777
3778 which matches one digit by preference, but can match two if that is the
3779 only way the rest of the pattern matches.
3780
3781 If the PCRE_UNGREEDY option is set (an option that is not available in
3782 Perl), the quantifiers are not greedy by default, but individual ones
3783 can be made greedy by following them with a question mark. In other
3784 words, it inverts the default behaviour.
3785
3786 When a parenthesized subpattern is quantified with a minimum repeat
3787 count that is greater than 1 or with a limited maximum, more memory is
3788 required for the compiled pattern, in proportion to the size of the
3789 minimum or maximum.
3790
3791 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
3792 alent to Perl's /s) is set, thus allowing the dot to match newlines,
3793 the pattern is implicitly anchored, because whatever follows will be
3794 tried against every character position in the subject string, so there
3795 is no point in retrying the overall match at any position after the
3796 first. PCRE normally treats such a pattern as though it were preceded
3797 by \A.
3798
3799 In cases where it is known that the subject string contains no new-
3800 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
3801 mization, or alternatively using ^ to indicate anchoring explicitly.
3802
3803 However, there is one situation where the optimization cannot be used.
3804 When .* is inside capturing parentheses that are the subject of a
3805 backreference elsewhere in the pattern, a match at the start may fail
3806 where a later one succeeds. Consider, for example:
3807
3808 (.*)abc\1
3809
3810 If the subject is "xyz123abc123" the match point is the fourth charac-
3811 ter. For this reason, such a pattern is not implicitly anchored.
3812
3813 When a capturing subpattern is repeated, the value captured is the sub-
3814 string that matched the final iteration. For example, after
3815
3816 (tweedle[dume]{3}\s*)+
3817
3818 has matched "tweedledum tweedledee" the value of the captured substring
3819 is "tweedledee". However, if there are nested capturing subpatterns,
3820 the corresponding captured values may have been set in previous itera-
3821 tions. For example, after
3822
3823 /(a|(b))+/
3824
3825 matches "aba" the value of the second captured substring is "b".
3826
3827
3828 ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS
3829
3830 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
3831 repetition, failure of what follows normally causes the repeated item
3832 to be re-evaluated to see if a different number of repeats allows the
3833 rest of the pattern to match. Sometimes it is useful to prevent this,
3834 either to change the nature of the match, or to cause it fail earlier
3835 than it otherwise might, when the author of the pattern knows there is
3836 no point in carrying on.
3837
3838 Consider, for example, the pattern \d+foo when applied to the subject
3839 line
3840
3841 123456bar
3842
3843 After matching all 6 digits and then failing to match "foo", the normal
3844 action of the matcher is to try again with only 5 digits matching the
3845 \d+ item, and then with 4, and so on, before ultimately failing.
3846 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
3847 the means for specifying that once a subpattern has matched, it is not
3848 to be re-evaluated in this way.
3849
3850 If we use atomic grouping for the previous example, the matcher gives
3851 up immediately on failing to match "foo" the first time. The notation
3852 is a kind of special parenthesis, starting with (?> as in this example:
3853
3854 (?>\d+)foo
3855
3856 This kind of parenthesis "locks up" the part of the pattern it con-
3857 tains once it has matched, and a failure further into the pattern is
3858 prevented from backtracking into it. Backtracking past it to previous
3859 items, however, works as normal.
3860
3861 An alternative description is that a subpattern of this type matches
3862 the string of characters that an identical standalone pattern would
3863 match, if anchored at the current point in the subject string.
3864
3865 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
3866 such as the above example can be thought of as a maximizing repeat that
3867 must swallow everything it can. So, while both \d+ and \d+? are pre-
3868 pared to adjust the number of digits they match in order to make the
3869 rest of the pattern match, (?>\d+) can only match an entire sequence of
3870 digits.
3871
3872 Atomic groups in general can of course contain arbitrarily complicated
3873 subpatterns, and can be nested. However, when the subpattern for an
3874 atomic group is just a single repeated item, as in the example above, a
3875 simpler notation, called a "possessive quantifier" can be used. This
3876 consists of an additional + character following a quantifier. Using
3877 this notation, the previous example can be rewritten as
3878
3879 \d++foo
3880
3881 Possessive quantifiers are always greedy; the setting of the
3882 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
3883 simpler forms of atomic group. However, there is no difference in the
3884 meaning of a possessive quantifier and the equivalent atomic group,
3885 though there may be a performance difference; possessive quantifiers
3886 should be slightly faster.
3887
3888 The possessive quantifier syntax is an extension to the Perl 5.8 syn-
3889 tax. Jeffrey Friedl originated the idea (and the name) in the first
3890 edition of his book. Mike McCloskey liked it, so implemented it when he
3891 built Sun's Java package, and PCRE copied it from there. It ultimately
3892 found its way into Perl at release 5.10.
3893
3894 PCRE has an optimization that automatically "possessifies" certain sim-
3895 ple pattern constructs. For example, the sequence A+B is treated as
3896 A++B because there is no point in backtracking into a sequence of A's
3897 when B must follow.
3898
3899 When a pattern contains an unlimited repeat inside a subpattern that
3900 can itself be repeated an unlimited number of times, the use of an
3901 atomic group is the only way to avoid some failing matches taking a
3902 very long time indeed. The pattern
3903
3904 (\D+|<\d+>)*[!?]
3905
3906 matches an unlimited number of substrings that either consist of non-
3907 digits, or digits enclosed in <>, followed by either ! or ?. When it
3908 matches, it runs quickly. However, if it is applied to
3909
3910 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
3911
3912 it takes a long time before reporting failure. This is because the
3913 string can be divided between the internal \D+ repeat and the external
3914 * repeat in a large number of ways, and all have to be tried. (The
3915 example uses [!?] rather than a single character at the end, because
3916 both PCRE and Perl have an optimization that allows for fast failure
3917 when a single character is used. They remember the last single charac-
3918 ter that is required for a match, and fail early if it is not present
3919 in the string.) If the pattern is changed so that it uses an atomic
3920 group, like this:
3921
3922 ((?>\D+)|<\d+>)*[!?]
3923
3924 sequences of non-digits cannot be broken, and failure happens quickly.
3925
3926
3927 BACK REFERENCES
3928
3929 Outside a character class, a backslash followed by a digit greater than
3930 0 (and possibly further digits) is a back reference to a capturing sub-
3931 pattern earlier (that is, to its left) in the pattern, provided there
3932 have been that many previous capturing left parentheses.
3933
3934 However, if the decimal number following the backslash is less than 10,
3935 it is always taken as a back reference, and causes an error only if
3936 there are not that many capturing left parentheses in the entire pat-
3937 tern. In other words, the parentheses that are referenced need not be
3938 to the left of the reference for numbers less than 10. A "forward back
3939 reference" of this type can make sense when a repetition is involved
3940 and the subpattern to the right has participated in an earlier itera-
3941 tion.
3942
3943 It is not possible to have a numerical "forward back reference" to a
3944 subpattern whose number is 10 or more using this syntax because a
3945 sequence such as \50 is interpreted as a character defined in octal.
3946 See the subsection entitled "Non-printing characters" above for further
3947 details of the handling of digits following a backslash. There is no
3948 such problem when named parentheses are used. A back reference to any
3949 subpattern is possible using named parentheses (see below).
3950
3951 Another way of avoiding the ambiguity inherent in the use of digits
3952 following a backslash is to use the \g escape sequence, which is a fea-
3953 ture introduced in Perl 5.10. This escape must be followed by a posi-
3954 tive or a negative number, optionally enclosed in braces. These exam-
3955 ples are all identical:
3956
3957 (ring), \1
3958 (ring), \g1
3959 (ring), \g{1}
3960
3961 A positive number specifies an absolute reference without the ambiguity
3962 that is present in the older syntax. It is also useful when literal
3963 digits follow the reference. A negative number is a relative reference.
3964 Consider this example:
3965
3966 (abc(def)ghi)\g{-1}
3967
3968 The sequence \g{-1} is a reference to the most recently started captur-
3969 ing subpattern before \g, that is, is it equivalent to \2. Similarly,
3970 \g{-2} would be equivalent to \1. The use of relative references can be
3971 helpful in long patterns, and also in patterns that are created by
3972 joining together fragments that contain references within themselves.
3973
3974 A back reference matches whatever actually matched the capturing sub-
3975 pattern in the current subject string, rather than anything matching
3976 the subpattern itself (see "Subpatterns as subroutines" below for a way
3977 of doing that). So the pattern
3978
3979 (sens|respons)e and \1ibility
3980
3981 matches "sense and sensibility" and "response and responsibility", but
3982 not "sense and responsibility". If caseful matching is in force at the
3983 time of the back reference, the case of letters is relevant. For exam-
3984 ple,
3985
3986 ((?i)rah)\s+\1
3987
3988 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
3989 original capturing subpattern is matched caselessly.
3990
3991 There are several different ways of writing back references to named
3992 subpatterns. The .NET syntax \k{name} and the Perl syntax \k<name> or
3993 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's
3994 unified back reference syntax, in which \g can be used for both numeric
3995 and named references, is also supported. We could rewrite the above
3996 example in any of the following ways:
3997
3998 (?<p1>(?i)rah)\s+\k<p1>
3999 (?'p1'(?i)rah)\s+\k{p1}
4000 (?P<p1>(?i)rah)\s+(?P=p1)
4001 (?<p1>(?i)rah)\s+\g{p1}
4002
4003 A subpattern that is referenced by name may appear in the pattern
4004 before or after the reference.
4005
4006 There may be more than one back reference to the same subpattern. If a
4007 subpattern has not actually been used in a particular match, any back
4008 references to it always fail. For example, the pattern
4009
4010 (a|(bc))\2
4011
4012 always fails if it starts to match "a" rather than "bc". Because there
4013 may be many capturing parentheses in a pattern, all digits following
4014 the backslash are taken as part of a potential back reference number.
4015 If the pattern continues with a digit character, some delimiter must be
4016 used to terminate the back reference. If the PCRE_EXTENDED option is
4017 set, this can be whitespace. Otherwise an empty comment (see "Com-
4018 ments" below) can be used.
4019
4020 A back reference that occurs inside the parentheses to which it refers
4021 fails when the subpattern is first used, so, for example, (a\1) never
4022 matches. However, such references can be useful inside repeated sub-
4023 patterns. For example, the pattern
4024
4025 (a|b\1)+
4026
4027 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
4028 ation of the subpattern, the back reference matches the character
4029 string corresponding to the previous iteration. In order for this to
4030 work, the pattern must be such that the first iteration does not need
4031 to match the back reference. This can be done using alternation, as in
4032 the example above, or by a quantifier with a minimum of zero.
4033
4034
4035 ASSERTIONS
4036
4037 An assertion is a test on the characters following or preceding the
4038 current matching point that does not actually consume any characters.
4039 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
4040 described above.
4041
4042 More complicated assertions are coded as subpatterns. There are two
4043 kinds: those that look ahead of the current position in the subject
4044 string, and those that look behind it. An assertion subpattern is
4045 matched in the normal way, except that it does not cause the current
4046 matching position to be changed.
4047
4048 Assertion subpatterns are not capturing subpatterns, and may not be
4049 repeated, because it makes no sense to assert the same thing several
4050 times. If any kind of assertion contains capturing subpatterns within
4051 it, these are counted for the purposes of numbering the capturing sub-
4052 patterns in the whole pattern. However, substring capturing is carried
4053 out only for positive assertions, because it does not make sense for
4054 negative assertions.
4055
4056 Lookahead assertions
4057
4058 Lookahead assertions start with (?= for positive assertions and (?! for
4059 negative assertions. For example,
4060
4061 \w+(?=;)
4062
4063 matches a word followed by a semicolon, but does not include the semi-
4064 colon in the match, and
4065
4066 foo(?!bar)
4067
4068 matches any occurrence of "foo" that is not followed by "bar". Note
4069 that the apparently similar pattern
4070
4071 (?!foo)bar
4072
4073 does not find an occurrence of "bar" that is preceded by something
4074 other than "foo"; it finds any occurrence of "bar" whatsoever, because
4075 the assertion (?!foo) is always true when the next three characters are
4076 "bar". A lookbehind assertion is needed to achieve the other effect.
4077
4078 If you want to force a matching failure at some point in a pattern, the
4079 most convenient way to do it is with (?!) because an empty string
4080 always matches, so an assertion that requires there not to be an empty
4081 string must always fail.
4082
4083 Lookbehind assertions
4084
4085 Lookbehind assertions start with (?<= for positive assertions and (?<!
4086 for negative assertions. For example,
4087
4088 (?<!foo)bar
4089
4090 does find an occurrence of "bar" that is not preceded by "foo". The
4091 contents of a lookbehind assertion are restricted such that all the
4092 strings it matches must have a fixed length. However, if there are sev-
4093 eral top-level alternatives, they do not all have to have the same
4094 fixed length. Thus
4095
4096 (?<=bullock|donkey)
4097
4098 is permitted, but
4099
4100 (?<!dogs?|cats?)
4101
4102 causes an error at compile time. Branches that match different length
4103 strings are permitted only at the top level of a lookbehind assertion.
4104 This is an extension compared with Perl (at least for 5.8), which
4105 requires all branches to match the same length of string. An assertion
4106 such as
4107
4108 (?<=ab(c|de))
4109
4110 is not permitted, because its single top-level branch can match two
4111 different lengths, but it is acceptable if rewritten to use two top-
4112 level branches:
4113
4114 (?<=abc|abde)
4115
4116 In some cases, the Perl 5.10 escape sequence \K (see above) can be used
4117 instead of a lookbehind assertion; this is not restricted to a fixed-
4118 length.
4119
4120 The implementation of lookbehind assertions is, for each alternative,
4121 to temporarily move the current position back by the fixed length and
4122 then try to match. If there are insufficient characters before the cur-
4123 rent position, the assertion fails.
4124
4125 PCRE does not allow the \C escape (which matches a single byte in UTF-8
4126 mode) to appear in lookbehind assertions, because it makes it impossi-
4127 ble to calculate the length of the lookbehind. The \X and \R escapes,
4128 which can match different numbers of bytes, are also not permitted.
4129
4130 Possessive quantifiers can be used in conjunction with lookbehind
4131 assertions to specify efficient matching at the end of the subject
4132 string. Consider a simple pattern such as
4133
4134 abcd$
4135
4136 when applied to a long string that does not match. Because matching
4137 proceeds from left to right, PCRE will look for each "a" in the subject
4138 and then see if what follows matches the rest of the pattern. If the
4139 pattern is specified as
4140
4141 ^.*abcd$
4142
4143 the initial .* matches the entire string at first, but when this fails
4144 (because there is no following "a"), it backtracks to match all but the
4145 last character, then all but the last two characters, and so on. Once
4146 again the search for "a" covers the entire string, from right to left,
4147 so we are no better off. However, if the pattern is written as
4148
4149 ^.*+(?<=abcd)
4150
4151 there can be no backtracking for the .*+ item; it can match only the
4152 entire string. The subsequent lookbehind assertion does a single test
4153 on the last four characters. If it fails, the match fails immediately.
4154 For long strings, this approach makes a significant difference to the
4155 processing time.
4156
4157 Using multiple assertions
4158
4159 Several assertions (of any sort) may occur in succession. For example,
4160
4161 (?<=\d{3})(?<!999)foo
4162
4163 matches "foo" preceded by three digits that are not "999". Notice that
4164 each of the assertions is applied independently at the same point in
4165 the subject string. First there is a check that the previous three
4166 characters are all digits, and then there is a check that the same
4167 three characters are not "999". This pattern does not match "foo" pre-
4168 ceded by six characters, the first of which are digits and the last
4169 three of which are not "999". For example, it doesn't match "123abc-
4170 foo". A pattern to do that is
4171
4172 (?<=\d{3}...)(?<!999)foo
4173
4174 This time the first assertion looks at the preceding six characters,
4175 checking that the first three are digits, and then the second assertion
4176 checks that the preceding three characters are not "999".
4177
4178 Assertions can be nested in any combination. For example,
4179
4180 (?<=(?<!foo)bar)baz
4181
4182 matches an occurrence of "baz" that is preceded by "bar" which in turn
4183 is not preceded by "foo", while
4184
4185 (?<=\d{3}(?!999)...)foo
4186
4187 is another pattern that matches "foo" preceded by three digits and any
4188 three characters that are not "999".
4189
4190
4191 CONDITIONAL SUBPATTERNS
4192
4193 It is possible to cause the matching process to obey a subpattern con-
4194 ditionally or to choose between two alternative subpatterns, depending
4195 on the result of an assertion, or whether a previous capturing subpat-
4196 tern matched or not. The two possible forms of conditional subpattern
4197 are
4198
4199 (?(condition)yes-pattern)
4200 (?(condition)yes-pattern|no-pattern)
4201
4202 If the condition is satisfied, the yes-pattern is used; otherwise the
4203 no-pattern (if present) is used. If there are more than two alterna-
4204 tives in the subpattern, a compile-time error occurs.
4205
4206 There are four kinds of condition: references to subpatterns, refer-
4207 ences to recursion, a pseudo-condition called DEFINE, and assertions.
4208
4209 Checking for a used subpattern by number
4210
4211 If the text between the parentheses consists of a sequence of digits,
4212 the condition is true if the capturing subpattern of that number has
4213 previously matched. An alternative notation is to precede the digits
4214 with a plus or minus sign. In this case, the subpattern number is rela-
4215 tive rather than absolute. The most recently opened parentheses can be
4216 referenced by (?(-1), the next most recent by (?(-2), and so on. In
4217 looping constructs it can also make sense to refer to subsequent groups
4218 with constructs such as (?(+2).
4219
4220 Consider the following pattern, which contains non-significant white
4221 space to make it more readable (assume the PCRE_EXTENDED option) and to
4222 divide it into three parts for ease of discussion:
4223
4224 ( \( )? [^()]+ (?(1) \) )
4225
4226 The first part matches an optional opening parenthesis, and if that
4227 character is present, sets it as the first captured substring. The sec-
4228 ond part matches one or more characters that are not parentheses. The
4229 third part is a conditional subpattern that tests whether the first set
4230 of parentheses matched or not. If they did, that is, if subject started
4231 with an opening parenthesis, the condition is true, and so the yes-pat-
4232 tern is executed and a closing parenthesis is required. Otherwise,
4233 since no-pattern is not present, the subpattern matches nothing. In
4234 other words, this pattern matches a sequence of non-parentheses,
4235 optionally enclosed in parentheses.
4236
4237 If you were embedding this pattern in a larger one, you could use a
4238 relative reference:
4239
4240 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
4241
4242 This makes the fragment independent of the parentheses in the larger
4243 pattern.
4244
4245 Checking for a used subpattern by name
4246
4247 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
4248 used subpattern by name. For compatibility with earlier versions of
4249 PCRE, which had this facility before Perl, the syntax (?(name)...) is
4250 also recognized. However, there is a possible ambiguity with this syn-
4251 tax, because subpattern names may consist entirely of digits. PCRE
4252 looks first for a named subpattern; if it cannot find one and the name
4253 consists entirely of digits, PCRE looks for a subpattern of that num-
4254 ber, which must be greater than zero. Using subpattern names that con-
4255 sist entirely of digits is not recommended.
4256
4257 Rewriting the above example to use a named subpattern gives this:
4258
4259 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
4260
4261
4262 Checking for pattern recursion
4263
4264 If the condition is the string (R), and there is no subpattern with the
4265 name R, the condition is true if a recursive call to the whole pattern
4266 or any subpattern has been made. If digits or a name preceded by amper-
4267 sand follow the letter R, for example:
4268
4269 (?(R3)...) or (?(R&name)...)
4270
4271 the condition is true if the most recent recursion is into the subpat-
4272 tern whose number or name is given. This condition does not check the
4273 entire recursion stack.
4274
4275 At "top level", all these recursion test conditions are false. Recur-
4276 sive patterns are described below.
4277
4278 Defining subpatterns for use by reference only
4279
4280 If the condition is the string (DEFINE), and there is no subpattern
4281 with the name DEFINE, the condition is always false. In this case,
4282 there may be only one alternative in the subpattern. It is always
4283 skipped if control reaches this point in the pattern; the idea of
4284 DEFINE is that it can be used to define "subroutines" that can be ref-
4285 erenced from elsewhere. (The use of "subroutines" is described below.)
4286 For example, a pattern to match an IPv4 address could be written like
4287 this (ignore whitespace and line breaks):
4288
4289 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
4290 \b (?&byte) (\.(?&byte)){3} \b
4291
4292 The first part of the pattern is a DEFINE group inside which a another
4293 group named "byte" is defined. This matches an individual component of
4294 an IPv4 address (a number less than 256). When matching takes place,
4295 this part of the pattern is skipped because DEFINE acts like a false
4296 condition.
4297
4298 The rest of the pattern uses references to the named group to match the
4299 four dot-separated components of an IPv4 address, insisting on a word
4300 boundary at each end.
4301
4302 Assertion conditions
4303
4304 If the condition is not in any of the above formats, it must be an
4305 assertion. This may be a positive or negative lookahead or lookbehind
4306 assertion. Consider this pattern, again containing non-significant
4307 white space, and with the two alternatives on the second line:
4308
4309 (?(?=[^a-z]*[a-z])
4310 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
4311
4312 The condition is a positive lookahead assertion that matches an
4313 optional sequence of non-letters followed by a letter. In other words,
4314 it tests for the presence of at least one letter in the subject. If a
4315 letter is found, the subject is matched against the first alternative;
4316 otherwise it is matched against the second. This pattern matches
4317 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
4318 letters and dd are digits.
4319
4320
4321 COMMENTS
4322
4323 The sequence (?# marks the start of a comment that continues up to the
4324 next closing parenthesis. Nested parentheses are not permitted. The
4325 characters that make up a comment play no part in the pattern matching
4326 at all.
4327
4328 If the PCRE_EXTENDED option is set, an unescaped # character outside a
4329 character class introduces a comment that continues to immediately
4330 after the next newline in the pattern.
4331
4332
4333 RECURSIVE PATTERNS
4334
4335 Consider the problem of matching a string in parentheses, allowing for
4336 unlimited nested parentheses. Without the use of recursion, the best
4337 that can be done is to use a pattern that matches up to some fixed
4338 depth of nesting. It is not possible to handle an arbitrary nesting
4339 depth.
4340
4341 For some time, Perl has provided a facility that allows regular expres-
4342 sions to recurse (amongst other things). It does this by interpolating
4343 Perl code in the expression at run time, and the code can refer to the
4344 expression itself. A Perl pattern using code interpolation to solve the
4345 parentheses problem can be created like this:
4346
4347 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
4348
4349 The (?p{...}) item interpolates Perl code at run time, and in this case
4350 refers recursively to the pattern in which it appears.
4351
4352 Obviously, PCRE cannot support the interpolation of Perl code. Instead,
4353 it supports special syntax for recursion of the entire pattern, and
4354 also for individual subpattern recursion. After its introduction in
4355 PCRE and Python, this kind of recursion was introduced into Perl at
4356 release 5.10.
4357
4358 A special item that consists of (? followed by a number greater than
4359 zero and a closing parenthesis is a recursive call of the subpattern of
4360 the given number, provided that it occurs inside that subpattern. (If
4361 not, it is a "subroutine" call, which is described in the next sec-
4362 tion.) The special item (?R) or (?0) is a recursive call of the entire
4363 regular expression.
4364
4365 In PCRE (like Python, but unlike Perl), a recursive subpattern call is
4366 always treated as an atomic group. That is, once it has matched some of
4367 the subject string, it is never re-entered, even if it contains untried
4368 alternatives and there is a subsequent matching failure.
4369
4370 This PCRE pattern solves the nested parentheses problem (assume the
4371 PCRE_EXTENDED option is set so that white space is ignored):
4372
4373 \( ( (?>[^()]+) | (?R) )* \)
4374
4375 First it matches an opening parenthesis. Then it matches any number of
4376 substrings which can either be a sequence of non-parentheses, or a
4377 recursive match of the pattern itself (that is, a correctly parenthe-
4378 sized substring). Finally there is a closing parenthesis.
4379
4380 If this were part of a larger pattern, you would not want to recurse
4381 the entire pattern, so instead you could use this:
4382
4383 ( \( ( (?>[^()]+) | (?1) )* \) )
4384
4385 We have put the pattern into parentheses, and caused the recursion to
4386 refer to them instead of the whole pattern.
4387
4388 In a larger pattern, keeping track of parenthesis numbers can be
4389 tricky. This is made easier by the use of relative references. (A Perl
4390 5.10 feature.) Instead of (?1) in the pattern above you can write
4391 (?-2) to refer to the second most recently opened parentheses preceding
4392 the recursion. In other words, a negative number counts capturing
4393 parentheses leftwards from the point at which it is encountered.
4394
4395 It is also possible to refer to subsequently opened parentheses, by
4396 writing references such as (?+2). However, these cannot be recursive
4397 because the reference is not inside the parentheses that are refer-
4398 enced. They are always "subroutine" calls, as described in the next
4399 section.
4400
4401 An alternative approach is to use named parentheses instead. The Perl
4402 syntax for this is (?&name); PCRE's earlier syntax (?P>name) is also
4403 supported. We could rewrite the above example as follows:
4404
4405 (?<pn> \( ( (?>[^()]+) | (?&pn) )* \) )
4406
4407 If there is more than one subpattern with the same name, the earliest
4408 one is used.
4409
4410 This particular example pattern that we have been looking at contains
4411 nested unlimited repeats, and so the use of atomic grouping for match-
4412 ing strings of non-parentheses is important when applying the pattern
4413 to strings that do not match. For example, when this pattern is applied
4414 to
4415
4416 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
4417
4418 it yields "no match" quickly. However, if atomic grouping is not used,
4419 the match runs for a very long time indeed because there are so many
4420 different ways the + and * repeats can carve up the subject, and all
4421 have to be tested before failure can be reported.
4422
4423 At the end of a match, the values set for any capturing subpatterns are
4424 those from the outermost level of the recursion at which the subpattern
4425 value is set. If you want to obtain intermediate values, a callout
4426 function can be used (see below and the pcrecallout documentation). If
4427 the pattern above is matched against
4428
4429 (ab(cd)ef)
4430
4431 the value for the capturing parentheses is "ef", which is the last
4432 value taken on at the top level. If additional parentheses are added,
4433 giving
4434
4435 \( ( ( (?>[^()]+) | (?R) )* ) \)
4436 ^ ^
4437 ^ ^
4438
4439 the string they capture is "ab(cd)ef", the contents of the top level
4440 parentheses. If there are more than 15 capturing parentheses in a pat-
4441 tern, PCRE has to obtain extra memory to store data during a recursion,
4442 which it does by using pcre_malloc, freeing it via pcre_free after-
4443 wards. If no memory can be obtained, the match fails with the
4444 PCRE_ERROR_NOMEMORY error.
4445
4446 Do not confuse the (?R) item with the condition (R), which tests for
4447 recursion. Consider this pattern, which matches text in angle brack-
4448 ets, allowing for arbitrary nesting. Only digits are allowed in nested
4449 brackets (that is, when recursing), whereas any characters are permit-
4450 ted at the outer level.
4451
4452 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
4453
4454 In this pattern, (?(R) is the start of a conditional subpattern, with
4455 two different alternatives for the recursive and non-recursive cases.
4456 The (?R) item is the actual recursive call.
4457
4458
4459 SUBPATTERNS AS SUBROUTINES
4460
4461 If the syntax for a recursive subpattern reference (either by number or
4462 by name) is used outside the parentheses to which it refers, it oper-
4463 ates like a subroutine in a programming language. The "called" subpat-
4464 tern may be defined before or after the reference. A numbered reference
4465 can be absolute or relative, as in these examples:
4466
4467 (...(absolute)...)...(?2)...
4468 (...(relative)...)...(?-1)...
4469 (...(?+1)...(relative)...
4470
4471 An earlier example pointed out that the pattern
4472
4473 (sens|respons)e and \1ibility
4474
4475 matches "sense and sensibility" and "response and responsibility", but
4476 not "sense and responsibility". If instead the pattern
4477
4478 (sens|respons)e and (?1)ibility
4479
4480 is used, it does match "sense and responsibility" as well as the other
4481 two strings. Another example is given in the discussion of DEFINE
4482 above.
4483
4484 Like recursive subpatterns, a "subroutine" call is always treated as an
4485 atomic group. That is, once it has matched some of the subject string,
4486 it is never re-entered, even if it contains untried alternatives and
4487 there is a subsequent matching failure.
4488
4489 When a subpattern is used as a subroutine, processing options such as
4490 case-independence are fixed when the subpattern is defined. They cannot
4491 be changed for different calls. For example, consider this pattern:
4492
4493 (abc)(?i:(?-1))
4494
4495 It matches "abcabc". It does not match "abcABC" because the change of
4496 processing option does not affect the called subpattern.
4497
4498
4499 CALLOUTS
4500
4501 Perl has a feature whereby using the sequence (?{...}) causes arbitrary
4502 Perl code to be obeyed in the middle of matching a regular expression.
4503 This makes it possible, amongst other things, to extract different sub-
4504 strings that match the same pair of parentheses when there is a repeti-
4505 tion.
4506
4507 PCRE provides a similar feature, but of course it cannot obey arbitrary
4508 Perl code. The feature is called "callout". The caller of PCRE provides
4509 an external function by putting its entry point in the global variable
4510 pcre_callout. By default, this variable contains NULL, which disables
4511 all calling out.
4512
4513 Within a regular expression, (?C) indicates the points at which the
4514 external function is to be called. If you want to identify different
4515 callout points, you can put a number less than 256 after the letter C.
4516 The default value is zero. For example, this pattern has two callout
4517 points:
4518
4519 (?C1)abc(?C2)def
4520
4521 If the PCRE_AUTO_CALLOUT flag is passed to pcre_compile(), callouts are
4522 automatically installed before each item in the pattern. They are all
4523 numbered 255.
4524
4525 During matching, when PCRE reaches a callout point (and pcre_callout is
4526 set), the external function is called. It is provided with the number
4527 of the callout, the position in the pattern, and, optionally, one item
4528 of data originally supplied by the caller of pcre_exec(). The callout
4529 function may cause matching to proceed, to backtrack, or to fail alto-
4530 gether. A complete description of the interface to the callout function
4531 is given in the pcrecallout documentation.
4532
4533
4534 SEE ALSO
4535
4536 pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
4537
4538
4539 AUTHOR
4540
4541 Philip Hazel
4542 University Computing Service
4543 Cambridge CB2 3QH, England.
4544
4545
4546 REVISION
4547
4548 Last updated: 19 June 2007
4549 Copyright (c) 1997-2007 University of Cambridge.
4550 ------------------------------------------------------------------------------
4551
4552
4553 PCREPARTIAL(3) PCREPARTIAL(3)
4554
4555
4556 NAME
4557 PCRE - Perl-compatible regular expressions
4558
4559
4560 PARTIAL MATCHING IN PCRE
4561
4562 In normal use of PCRE, if the subject string that is passed to
4563 pcre_exec() or pcre_dfa_exec() matches as far as it goes, but is too
4564 short to match the entire pattern, PCRE_ERROR_NOMATCH is returned.
4565 There are circumstances where it might be helpful to distinguish this
4566 case from other cases in which there is no match.
4567
4568 Consider, for example, an application where a human is required to type
4569 in data for a field with specific formatting requirements. An example
4570 might be a date in the form ddmmmyy, defined by this pattern:
4571
4572 ^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$
4573
4574 If the application sees the user's keystrokes one by one, and can check
4575 that what has been typed so far is potentially valid, it is able to
4576 raise an error as soon as a mistake is made, possibly beeping and not
4577 reflecting the character that has been typed. This immediate feedback
4578 is likely to be a better user interface than a check that is delayed
4579 until the entire string has been entered.
4580
4581 PCRE supports the concept of partial matching by means of the PCRE_PAR-
4582 TIAL option, which can be set when calling pcre_exec() or
4583 pcre_dfa_exec(). When this flag is set for pcre_exec(), the return code
4584 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if at any time
4585 during the matching process the last part of the subject string matched
4586 part of the pattern. Unfortunately, for non-anchored matching, it is
4587 not possible to obtain the position of the start of the partial match.
4588 No captured data is set when PCRE_ERROR_PARTIAL is returned.
4589
4590 When PCRE_PARTIAL is set for pcre_dfa_exec(), the return code
4591 PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end of
4592 the subject is reached, there have been no complete matches, but there
4593 is still at least one matching possibility. The portion of the string
4594 that provided the partial match is set as the first matching string.
4595
4596 Using PCRE_PARTIAL disables one of PCRE's optimizations. PCRE remembers
4597 the last literal byte in a pattern, and abandons matching immediately
4598 if such a byte is not present in the subject string. This optimization
4599 cannot be used for a subject string that might match only partially.
4600
4601
4602 RESTRICTED PATTERNS FOR PCRE_PARTIAL
4603
4604 Because of the way certain internal optimizations are implemented in
4605 the pcre_exec() function, the PCRE_PARTIAL option cannot be used with
4606 all patterns. These restrictions do not apply when pcre_dfa_exec() is
4607 used. For pcre_exec(), repeated single characters such as
4608
4609 a{2,4}
4610
4611 and repeated single metasequences such as
4612
4613 \d+
4614
4615 are not permitted if the maximum number of occurrences is greater than
4616 one. Optional items such as \d? (where the maximum is one) are permit-
4617 ted. Quantifiers with any values are permitted after parentheses, so
4618 the invalid examples above can be coded thus:
4619
4620 (a){2,4}
4621 (\d)+
4622
4623 These constructions run more slowly, but for the kinds of application
4624 that are envisaged for this facility, this is not felt to be a major
4625 restriction.
4626
4627 If PCRE_PARTIAL is set for a pattern that does not conform to the
4628 restrictions, pcre_exec() returns the error code PCRE_ERROR_BADPARTIAL
4629 (-13). You can use the PCRE_INFO_OKPARTIAL call to pcre_fullinfo() to
4630 find out if a compiled pattern can be used for partial matching.
4631
4632
4633 EXAMPLE OF PARTIAL MATCHING USING PCRETEST
4634
4635 If the escape sequence \P is present in a pcretest data line, the
4636 PCRE_PARTIAL flag is used for the match. Here is a run of pcretest that
4637 uses the date example quoted above:
4638
4639 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
4640 data> 25jun04\P
4641 0: 25jun04
4642 1: jun
4643 data> 25dec3\P
4644 Partial match
4645 data> 3ju\P
4646 Partial match
4647 data> 3juj\P
4648 No match
4649 data> j\P
4650 No match
4651
4652 The first data string is matched completely, so pcretest shows the
4653 matched substrings. The remaining four strings do not match the com-
4654 plete pattern, but the first two are partial matches. The same test,
4655 using pcre_dfa_exec() matching (by means of the \D escape sequence),
4656 produces the following output:
4657
4658 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
4659 data> 25jun04\P\D
4660 0: 25jun04
4661 data> 23dec3\P\D
4662 Partial match: 23dec3
4663 data> 3ju\P\D
4664 Partial match: 3ju
4665 data> 3juj\P\D
4666 No match
4667 data> j\P\D
4668 No match
4669
4670 Notice that in this case the portion of the string that was matched is
4671 made available.
4672
4673
4674 MULTI-SEGMENT MATCHING WITH pcre_dfa_exec()
4675
4676 When a partial match has been found using pcre_dfa_exec(), it is possi-
4677 ble to continue the match by providing additional subject data and
4678 calling pcre_dfa_exec() again with the same compiled regular expres-
4679 sion, this time setting the PCRE_DFA_RESTART option. You must also pass
4680 the same working space as before, because this is where details of the
4681 previous partial match are stored. Here is an example using pcretest,
4682 using the \R escape sequence to set the PCRE_DFA_RESTART option (\P and
4683 \D are as above):
4684
4685 re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
4686 data> 23ja\P\D
4687 Partial match: 23ja
4688 data> n05\R\D
4689 0: n05
4690
4691 The first call has "23ja" as the subject, and requests partial match-
4692 ing; the second call has "n05" as the subject for the continued
4693 (restarted) match. Notice that when the match is complete, only the
4694 last part is shown; PCRE does not retain the previously partially-
4695 matched string. It is up to the calling program to do that if it needs
4696 to.
4697
4698 You can set PCRE_PARTIAL with PCRE_DFA_RESTART to continue partial
4699 matching over multiple segments. This facility can be used to pass very
4700 long subject strings to pcre_dfa_exec(). However, some care is needed
4701 for certain types of pattern.
4702
4703 1. If the pattern contains tests for the beginning or end of a line,
4704 you need to pass the PCRE_NOTBOL or PCRE_NOTEOL options, as appropri-
4705 ate, when the subject string for any call does not contain the begin-
4706 ning or end of a line.
4707
4708 2. If the pattern contains backward assertions (including \b or \B),
4709 you need to arrange for some overlap in the subject strings to allow
4710 for this. For example, you could pass the subject in chunks that are
4711 500 bytes long, but in a buffer of 700 bytes, with the starting offset
4712 set to 200 and the previous 200 bytes at the start of the buffer.
4713
4714 3. Matching a subject string that is split into multiple segments does
4715 not always produce exactly the same result as matching over one single
4716 long string. The difference arises when there are multiple matching
4717 possibilities, because a partial match result is given only when there
4718 are no completed matches in a call to pcre_dfa_exec(). This means that
4719 as soon as the shortest match has been found, continuation to a new
4720 subject segment is no longer possible. Consider this pcretest example:
4721
4722 re> /dog(sbody)?/
4723 data> do\P\D
4724 Partial match: do
4725 data> gsb\R\P\D
4726 0: g
4727 data> dogsbody\D
4728 0: dogsbody
4729 1: dog
4730
4731 The pattern matches the words "dog" or "dogsbody". When the subject is
4732 presented in several parts ("do" and "gsb" being the first two) the
4733 match stops when "dog" has been found, and it is not possible to con-
4734 tinue. On the other hand, if "dogsbody" is presented as a single
4735 string, both matches are found.
4736
4737 Because of this phenomenon, it does not usually make sense to end a
4738 pattern that is going to be matched in this way with a variable repeat.
4739
4740 4. Patterns that contain alternatives at the top level which do not all
4741 start with the same pattern item may not work as expected. For example,
4742 consider this pattern:
4743
4744 1234|3789
4745
4746 If the first part of the subject is "ABC123", a partial match of the
4747 first alternative is found at offset 3. There is no partial match for
4748 the second alternative, because such a match does not start at the same
4749 point in the subject string. Attempting to continue with the string
4750 "789" does not yield a match because only those alternatives that match
4751 at one point in the subject are remembered. The problem arises because
4752 the start of the second alternative matches within the first alterna-
4753 tive. There is no problem with anchored patterns or patterns such as:
4754
4755 1234|ABCD
4756
4757 where no string can be a partial match for both alternatives.
4758
4759
4760 AUTHOR
4761
4762 Philip Hazel
4763 University Computing Service
4764 Cambridge CB2 3QH, England.
4765
4766
4767 REVISION
4768
4769 Last updated: 04 June 2007
4770 Copyright (c) 1997-2007 University of Cambridge.
4771 ------------------------------------------------------------------------------
4772
4773
4774 PCREPRECOMPILE(3) PCREPRECOMPILE(3)
4775
4776
4777 NAME
4778 PCRE - Perl-compatible regular expressions
4779
4780
4781 SAVING AND RE-USING PRECOMPILED PCRE PATTERNS
4782
4783 If you are running an application that uses a large number of regular
4784 expression patterns, it may be useful to store them in a precompiled
4785 form instead of having to compile them every time the application is
4786 run. If you are not using any private character tables (see the
4787 pcre_maketables() documentation), this is relatively straightforward.
4788 If you are using private tables, it is a little bit more complicated.
4789
4790 If you save compiled patterns to a file, you can copy them to a differ-
4791 ent host and run them there. This works even if the new host has the
4792 opposite endianness to the one on which the patterns were compiled.
4793 There may be a small performance penalty, but it should be insignifi-
4794 cant. However, compiling regular expressions with one version of PCRE
4795 for use with a different version is not guaranteed to work and may
4796 cause crashes.
4797
4798
4799 SAVING A COMPILED PATTERN
4800 The value returned by pcre_compile() points to a single block of memory
4801 that holds the compiled pattern and associated data. You can find the
4802 length of this block in bytes by calling pcre_fullinfo() with an argu-
4803 ment of PCRE_INFO_SIZE. You can then save the data in any appropriate
4804 manner. Here is sample code that compiles a pattern and writes it to a
4805 file. It assumes that the variable fd refers to a file that is open for
4806 output:
4807
4808 int erroroffset, rc, size;
4809 char *error;
4810 pcre *re;
4811
4812 re = pcre_compile("my pattern", 0, &error, &erroroffset, NULL);
4813 if (re == NULL) { ... handle errors ... }
4814 rc = pcre_fullinfo(re, NULL, PCRE_INFO_SIZE, &size);
4815 if (rc < 0) { ... handle errors ... }
4816 rc = fwrite(re, 1, size, fd);
4817 if (rc != size) { ... handle errors ... }
4818
4819 In this example, the bytes that comprise the compiled pattern are
4820 copied exactly. Note that this is binary data that may contain any of
4821 the 256 possible byte values. On systems that make a distinction
4822 between binary and non-binary data, be sure that the file is opened for
4823 binary output.
4824
4825 If you want to write more than one pattern to a file, you will have to
4826 devise a way of separating them. For binary data, preceding each pat-
4827 tern with its length is probably the most straightforward approach.
4828 Another possibility is to write out the data in hexadecimal instead of
4829 binary, one pattern to a line.
4830
4831 Saving compiled patterns in a file is only one possible way of storing
4832 them for later use. They could equally well be saved in a database, or
4833 in the memory of some daemon process that passes them via sockets to
4834 the processes that want them.
4835
4836 If the pattern has been studied, it is also possible to save the study
4837 data in a similar way to the compiled pattern itself. When studying
4838 generates additional information, pcre_study() returns a pointer to a
4839 pcre_extra data block. Its format is defined in the section on matching
4840 a pattern in the pcreapi documentation. The study_data field points to
4841 the binary study data, and this is what you must save (not the
4842 pcre_extra block itself). The length of the study data can be obtained
4843 by calling pcre_fullinfo() with an argument of PCRE_INFO_STUDYSIZE.
4844 Remember to check that pcre_study() did return a non-NULL value before
4845 trying to save the study data.
4846
4847
4848 RE-USING A PRECOMPILED PATTERN
4849
4850 Re-using a precompiled pattern is straightforward. Having reloaded it
4851 into main memory, you pass its pointer to pcre_exec() or
4852 pcre_dfa_exec() in the usual way. This should work even on another
4853 host, and even if that host has the opposite endianness to the one
4854 where the pattern was compiled.
4855
4856 However, if you passed a pointer to custom character tables when the
4857 pattern was compiled (the tableptr argument of pcre_compile()), you
4858 must now pass a similar pointer to pcre_exec() or pcre_dfa_exec(),
4859 because the value saved with the compiled pattern will obviously be
4860 nonsense. A field in a pcre_extra() block is used to pass this data, as
4861 described in the section on matching a pattern in the pcreapi documen-
4862 tation.
4863
4864 If you did not provide custom character tables when the pattern was
4865 compiled, the pointer in the compiled pattern is NULL, which causes
4866 pcre_exec() to use PCRE's internal tables. Thus, you do not need to
4867 take any special action at run time in this case.
4868
4869 If you saved study data with the compiled pattern, you need to create
4870 your own pcre_extra data block and set the study_data field to point to
4871 the reloaded study data. You must also set the PCRE_EXTRA_STUDY_DATA
4872 bit in the flags field to indicate that study data is present. Then
4873 pass the pcre_extra block to pcre_exec() or pcre_dfa_exec() in the
4874 usual way.
4875
4876
4877 COMPATIBILITY WITH DIFFERENT PCRE RELEASES
4878
4879 In general, it is safest to recompile all saved patterns when you
4880 update to a new PCRE release, though not all updates actually require
4881 this. Recompiling is definitely needed for release 7.2.
4882
4883
4884 AUTHOR
4885
4886 Philip Hazel
4887 University Computing Service
4888 Cambridge CB2 3QH, England.
4889
4890
4891 REVISION
4892
4893 Last updated: 13 June 2007
4894 Copyright (c) 1997-2007 University of Cambridge.
4895 ------------------------------------------------------------------------------
4896
4897
4898 PCREPERFORM(3) PCREPERFORM(3)
4899
4900
4901 NAME
4902 PCRE - Perl-compatible regular expressions
4903
4904
4905 PCRE PERFORMANCE
4906
4907 Two aspects of performance are discussed below: memory usage and pro-
4908 cessing time. The way you express your pattern as a regular expression
4909 can affect both of them.
4910
4911
4912 MEMORY USAGE
4913
4914 Patterns are compiled by PCRE into a reasonably efficient byte code, so
4915 that most simple patterns do not use much memory. However, there is one
4916 case where memory usage can be unexpectedly large. When a parenthesized
4917 subpattern has a quantifier with a minimum greater than 1 and/or a lim-
4918 ited maximum, the whole subpattern is repeated in the compiled code.
4919 For example, the pattern
4920
4921 (abc|def){2,4}
4922
4923 is compiled as if it were
4924
4925 (abc|def)(abc|def)((abc|def)(abc|def)?)?
4926
4927 (Technical aside: It is done this way so that backtrack points within
4928 each of the repetitions can be independently maintained.)
4929
4930 For regular expressions whose quantifiers use only small numbers, this
4931 is not usually a problem. However, if the numbers are large, and par-
4932 ticularly if such repetitions are nested, the memory usage can become
4933 an embarrassment. For example, the very simple pattern
4934
4935 ((ab){1,1000}c){1,3}
4936
4937 uses 51K bytes when compiled. When PCRE is compiled with its default
4938 internal pointer size of two bytes, the size limit on a compiled pat-
4939 tern is 64K, and this is reached with the above pattern if the outer
4940 repetition is increased from 3 to 4. PCRE can be compiled to use larger
4941 internal pointers and thus handle larger compiled patterns, but it is
4942 better to try to rewrite your pattern to use less memory if you can.
4943
4944 One way of reducing the memory usage for such patterns is to make use
4945 of PCRE's "subroutine" facility. Re-writing the above pattern as
4946
4947 ((ab)(?2){0,999}c)(?1){0,2}
4948
4949 reduces the memory requirements to 18K, and indeed it remains under 20K
4950 even with the outer repetition increased to 100. However, this pattern
4951 is not exactly equivalent, because the "subroutine" calls are treated
4952 as atomic groups into which there can be no backtracking if there is a
4953 subsequent matching failure. Therefore, PCRE cannot do this kind of
4954 rewriting automatically. Furthermore, there is a noticeable loss of
4955 speed when executing the modified pattern. Nevertheless, if the atomic
4956 grouping is not a problem and the loss of speed is acceptable, this
4957 kind of rewriting will allow you to process patterns that PCRE cannot
4958 otherwise handle.
4959
4960
4961 PROCESSING TIME
4962
4963 Certain items in regular expression patterns are processed more effi-
4964 ciently than others. It is more efficient to use a character class like
4965 [aeiou] than a set of single-character alternatives such as
4966 (a|e|i|o|u). In general, the simplest construction that provides the
4967 required behaviour is usually the most efficient. Jeffrey Friedl's book
4968 contains a lot of useful general discussion about optimizing regular
4969 expressions for efficient performance. This document contains a few
4970 observations about PCRE.
4971
4972 Using Unicode character properties (the \p, \P, and \X escapes) is
4973 slow, because PCRE has to scan a structure that contains data for over
4974 fifteen thousand characters whenever it needs a character's property.
4975 If you can find an alternative pattern that does not use character
4976 properties, it will probably be faster.
4977
4978 When a pattern begins with .* not in parentheses, or in parentheses
4979 that are not the subject of a backreference, and the PCRE_DOTALL option
4980 is set, the pattern is implicitly anchored by PCRE, since it can match
4981 only at the start of a subject string. However, if PCRE_DOTALL is not
4982 set, PCRE cannot make this optimization, because the . metacharacter
4983 does not then match a newline, and if the subject string contains new-
4984 lines, the pattern may match from the character immediately following
4985 one of them instead of from the very start. For example, the pattern
4986
4987 .*second
4988
4989 matches the subject "first\nand second" (where \n stands for a newline
4990 character), with the match starting at the seventh character. In order
4991 to do this, PCRE has to retry the match starting after every newline in
4992 the subject.
4993
4994 If you are using such a pattern with subject strings that do not con-
4995 tain newlines, the best performance is obtained by setting PCRE_DOTALL,
4996 or starting the pattern with ^.* or ^.*? to indicate explicit anchor-
4997 ing. That saves PCRE from having to scan along the subject looking for
4998 a newline to restart at.
4999
5000 Beware of patterns that contain nested indefinite repeats. These can
5001 take a long time to run when applied to a string that does not match.
5002 Consider the pattern fragment
5003
5004 ^(a+)*
5005
5006 This can match "aaaa" in 16 different ways, and this number increases
5007 very rapidly as the string gets longer. (The * repeat can match 0, 1,
5008 2, 3, or 4 times, and for each of those cases other than 0 or 4, the +
5009 repeats can match different numbers of times.) When the remainder of
5010 the pattern is such that the entire match is going to fail, PCRE has in
5011 principle to try every possible variation, and this can take an
5012 extremely long time, even for relatively short strings.
5013
5014 An optimization catches some of the more simple cases such as
5015
5016 (a+)*b
5017
5018 where a literal character follows. Before embarking on the standard
5019 matching procedure, PCRE checks that there is a "b" later in the sub-
5020 ject string, and if there is not, it fails the match immediately. How-
5021 ever, when there is no following literal this optimization cannot be
5022 used. You can see the difference by comparing the behaviour of
5023
5024 (a+)*\d
5025
5026 with the pattern above. The former gives a failure almost instantly
5027 when applied to a whole line of "a" characters, whereas the latter
5028 takes an appreciable time with strings longer than about 20 characters.
5029
5030 In many cases, the solution to this kind of performance issue is to use
5031 an atomic group or a possessive quantifier.
5032
5033
5034 AUTHOR
5035
5036 Philip Hazel
5037 University Computing Service
5038 Cambridge CB2 3QH, England.
5039
5040
5041 REVISION
5042
5043 Last updated: 06 March 2007
5044 Copyright (c) 1997-2007 University of Cambridge.
5045 ------------------------------------------------------------------------------
5046
5047
5048 PCREPOSIX(3) PCREPOSIX(3)
5049
5050
5051 NAME
5052 PCRE - Perl-compatible regular expressions.
5053
5054
5055 SYNOPSIS OF POSIX API
5056
5057 #include <pcreposix.h>
5058
5059 int regcomp(regex_t *preg, const char *pattern,
5060 int cflags);
5061
5062 int regexec(regex_t *preg, const char *string,
5063 size_t nmatch, regmatch_t pmatch[], int eflags);
5064
5065 size_t regerror(int errcode, const regex_t *preg,
5066 char *errbuf, size_t errbuf_size);
5067
5068 void regfree(regex_t *preg);
5069
5070
5071 DESCRIPTION
5072
5073 This set of functions provides a POSIX-style API to the PCRE regular
5074 expression package. See the pcreapi documentation for a description of
5075 PCRE's native API, which contains much additional functionality.
5076
5077 The functions described here are just wrapper functions that ultimately
5078 call the PCRE native API. Their prototypes are defined in the
5079 pcreposix.h header file, and on Unix systems the library itself is
5080 called pcreposix.a, so can be accessed by adding -lpcreposix to the
5081 command for linking an application that uses them. Because the POSIX
5082 functions call the native ones, it is also necessary to add -lpcre.
5083
5084 I have implemented only those option bits that can be reasonably mapped
5085 to PCRE native options. In addition, the option REG_EXTENDED is defined
5086 with the value zero. This has no effect, but since programs that are
5087 written to the POSIX interface often use it, this makes it easier to
5088 slot in PCRE as a replacement library. Other POSIX options are not even
5089 defined.
5090
5091 When PCRE is called via these functions, it is only the API that is
5092 POSIX-like in style. The syntax and semantics of the regular expres-
5093 sions themselves are still those of Perl, subject to the setting of
5094 various PCRE options, as described below. "POSIX-like in style" means
5095 that the API approximates to the POSIX definition; it is not fully
5096 POSIX-compatible, and in multi-byte encoding domains it is probably
5097 even less compatible.
5098
5099 The header for these functions is supplied as pcreposix.h to avoid any
5100 potential clash with other POSIX libraries. It can, of course, be
5101 renamed or aliased as regex.h, which is the "correct" name. It provides
5102 two structure types, regex_t for compiled internal forms, and reg-
5103 match_t for returning captured substrings. It also defines some con-
5104 stants whose names start with "REG_"; these are used for setting
5105 options and identifying error codes.
5106
5107
5108 COMPILING A PATTERN
5109
5110 The function regcomp() is called to compile a pattern into an internal
5111 form. The pattern is a C string terminated by a binary zero, and is
5112 passed in the argument pattern. The preg argument is a pointer to a
5113 regex_t structure that is used as a base for storing information about
5114 the compiled regular expression.
5115
5116 The argument cflags is either zero, or contains one or more of the bits
5117 defined by the following macros:
5118
5119 REG_DOTALL
5120
5121 The PCRE_DOTALL option is set when the regular expression is passed for
5122 compilation to the native function. Note that REG_DOTALL is not part of
5123 the POSIX standard.
5124
5125 REG_ICASE
5126
5127 The PCRE_CASELESS option is set when the regular expression is passed
5128 for compilation to the native function.
5129
5130 REG_NEWLINE
5131
5132 The PCRE_MULTILINE option is set when the regular expression is passed
5133 for compilation to the native function. Note that this does not mimic
5134 the defined POSIX behaviour for REG_NEWLINE (see the following sec-
5135 tion).
5136
5137 REG_NOSUB
5138
5139 The PCRE_NO_AUTO_CAPTURE option is set when the regular expression is
5140 passed for compilation to the native function. In addition, when a pat-
5141 tern that is compiled with this flag is passed to regexec() for match-
5142 ing, the nmatch and pmatch arguments are ignored, and no captured
5143 strings are returned.
5144
5145 REG_UTF8
5146
5147 The PCRE_UTF8 option is set when the regular expression is passed for
5148 compilation to the native function. This causes the pattern itself and
5149 all data strings used for matching it to be treated as UTF-8 strings.
5150 Note that REG_UTF8 is not part of the POSIX standard.
5151
5152 In the absence of these flags, no options are passed to the native
5153 function. This means the the regex is compiled with PCRE default
5154 semantics. In particular, the way it handles newline characters in the
5155 subject string is the Perl way, not the POSIX way. Note that setting
5156 PCRE_MULTILINE has only some of the effects specified for REG_NEWLINE.
5157 It does not affect the way newlines are matched by . (they aren't) or
5158 by a negative class such as [^a] (they are).
5159
5160 The yield of regcomp() is zero on success, and non-zero otherwise. The
5161 preg structure is filled in on success, and one member of the structure
5162 is public: re_nsub contains the number of capturing subpatterns in the
5163 regular expression. Various error codes are defined in the header file.
5164
5165
5166 MATCHING NEWLINE CHARACTERS
5167
5168 This area is not simple, because POSIX and Perl take different views of
5169 things. It is not possible to get PCRE to obey POSIX semantics, but
5170 then PCRE was never intended to be a POSIX engine. The following table
5171 lists the different possibilities for matching newline characters in
5172 PCRE:
5173
5174 Default Change with
5175
5176 . matches newline no PCRE_DOTALL
5177 newline matches [^a] yes not changeable
5178 $ matches \n at end yes PCRE_DOLLARENDONLY
5179 $ matches \n in middle no PCRE_MULTILINE
5180 ^ matches \n in middle no PCRE_MULTILINE
5181
5182 This is the equivalent table for POSIX:
5183
5184 Default Change with
5185
5186 . matches newline yes REG_NEWLINE
5187 newline matches [^a] yes REG_NEWLINE
5188 $ matches \n at end no REG_NEWLINE
5189 $ matches \n in middle no REG_NEWLINE
5190 ^ matches \n in middle no REG_NEWLINE
5191
5192 PCRE's behaviour is the same as Perl's, except that there is no equiva-
5193 lent for PCRE_DOLLAR_ENDONLY in Perl. In both PCRE and Perl, there is
5194 no way to stop newline from matching [^a].
5195
5196 The default POSIX newline handling can be obtained by setting
5197 PCRE_DOTALL and PCRE_DOLLAR_ENDONLY, but there is no way to make PCRE
5198 behave exactly as for the REG_NEWLINE action.
5199
5200
5201 MATCHING A PATTERN
5202
5203 The function regexec() is called to match a compiled pattern preg
5204 against a given string, which is terminated by a zero byte, subject to
5205 the options in eflags. These can be:
5206
5207 REG_NOTBOL
5208
5209 The PCRE_NOTBOL option is set when calling the underlying PCRE matching
5210 function.
5211
5212 REG_NOTEOL
5213
5214 The PCRE_NOTEOL option is set when calling the underlying PCRE matching
5215 function.
5216
5217 If the pattern was compiled with the REG_NOSUB flag, no data about any
5218 matched strings is returned. The nmatch and pmatch arguments of
5219 regexec() are ignored.
5220
5221 Otherwise,the portion of the string that was matched, and also any cap-
5222 tured substrings, are returned via the pmatch argument, which points to
5223 an array of nmatch structures of type regmatch_t, containing the mem-
5224 bers rm_so and rm_eo. These contain the offset to the first character
5225 of each substring and the offset to the first character after the end
5226 of each substring, respectively. The 0th element of the vector relates
5227 to the entire portion of string that was matched; subsequent elements
5228 relate to the capturing subpatterns of the regular expression. Unused
5229 entries in the array have both structure members set to -1.
5230
5231 A successful match yields a zero return; various error codes are
5232 defined in the header file, of which REG_NOMATCH is the "expected"
5233 failure code.
5234
5235
5236 ERROR MESSAGES
5237
5238 The regerror() function maps a non-zero errorcode from either regcomp()
5239 or regexec() to a printable message. If preg is not NULL, the error
5240 should have arisen from the use of that structure. A message terminated
5241 by a binary zero is placed in errbuf. The length of the message,
5242 including the zero, is limited to errbuf_size. The yield of the func-
5243 tion is the size of buffer needed to hold the whole message.
5244
5245
5246 MEMORY USAGE
5247
5248 Compiling a regular expression causes memory to be allocated and asso-
5249 ciated with the preg structure. The function regfree() frees all such
5250 memory, after which preg may no longer be used as a compiled expres-
5251 sion.
5252
5253
5254 AUTHOR
5255
5256 Philip Hazel
5257 University Computing Service
5258 Cambridge CB2 3QH, England.
5259
5260
5261 REVISION
5262
5263 Last updated: 06 March 2007
5264 Copyright (c) 1997-2007 University of Cambridge.
5265 ------------------------------------------------------------------------------
5266
5267
5268 PCRECPP(3) PCRECPP(3)
5269
5270
5271 NAME
5272 PCRE - Perl-compatible regular expressions.
5273
5274
5275 SYNOPSIS OF C++ WRAPPER
5276
5277 #include <pcrecpp.h>
5278
5279
5280 DESCRIPTION
5281
5282 The C++ wrapper for PCRE was provided by Google Inc. Some additional
5283 functionality was added by Giuseppe Maxia. This brief man page was con-
5284 structed from the notes in the pcrecpp.h file, which should be con-
5285 sulted for further details.
5286
5287
5288 MATCHING INTERFACE
5289
5290 The "FullMatch" operation checks that supplied text matches a supplied
5291 pattern exactly. If pointer arguments are supplied, it copies matched
5292 sub-strings that match sub-patterns into them.
5293
5294 Example: successful match
5295 pcrecpp::RE re("h.*o");
5296 re.FullMatch("hello");
5297
5298 Example: unsuccessful match (requires full match):
5299 pcrecpp::RE re("e");
5300 !re.FullMatch("hello");
5301
5302 Example: creating a temporary RE object:
5303 pcrecpp::RE("h.*o").FullMatch("hello");
5304
5305 You can pass in a "const char*" or a "string" for "text". The examples
5306 below tend to use a const char*. You can, as in the different examples
5307 above, store the RE object explicitly in a variable or use a temporary
5308 RE object. The examples below use one mode or the other arbitrarily.
5309 Either could correctly be used for any of these examples.
5310
5311 You must supply extra pointer arguments to extract matched subpieces.
5312
5313 Example: extracts "ruby" into "s" and 1234 into "i"
5314 int i;
5315 string s;
5316 pcrecpp::RE re("(\\w+):(\\d+)");
5317 re.FullMatch("ruby:1234", &s, &i);
5318
5319 Example: does not try to extract any extra sub-patterns
5320 re.FullMatch("ruby:1234", &s);
5321
5322 Example: does not try to extract into NULL
5323 re.FullMatch("ruby:1234", NULL, &i);
5324
5325 Example: integer overflow causes failure
5326 !re.FullMatch("ruby:1234567891234", NULL, &i);
5327
5328 Example: fails because there aren't enough sub-patterns:
5329 !pcrecpp::RE("\\w+:\\d+").FullMatch("ruby:1234", &s);
5330
5331 Example: fails because string cannot be stored in integer
5332 !pcrecpp::RE("(.*)").FullMatch("ruby", &i);
5333
5334 The provided pointer arguments can be pointers to any scalar numeric
5335 type, or one of:
5336
5337 string (matched piece is copied to string)
5338 StringPiece (StringPiece is mutated to point to matched piece)
5339 T (where "bool T::ParseFrom(const char*, int)" exists)
5340 NULL (the corresponding matched sub-pattern is not copied)
5341
5342 The function returns true iff all of the following conditions are sat-
5343 isfied:
5344
5345 a. "text" matches "pattern" exactly;
5346
5347 b. The number of matched sub-patterns is >= number of supplied
5348 pointers;
5349
5350 c. The "i"th argument has a suitable type for holding the
5351 string captured as the "i"th sub-pattern. If you pass in
5352 NULL for the "i"th argument, or pass fewer arguments than
5353 number of sub-patterns, "i"th captured sub-pattern is
5354 ignored.
5355
5356 CAVEAT: An optional sub-pattern that does not exist in the matched
5357 string is assigned the empty string. Therefore, the following will
5358 return false (because the empty string is not a valid number):
5359
5360 int number;
5361 pcrecpp::RE::FullMatch("abc", "[a-z]+(\\d+)?", &number);
5362
5363 The matching interface supports at most 16 arguments per call. If you
5364 need more, consider using the more general interface
5365 pcrecpp::RE::DoMatch. See pcrecpp.h for the signature for DoMatch.
5366
5367
5368 QUOTING METACHARACTERS
5369
5370 You can use the "QuoteMeta" operation to insert backslashes before all
5371 potentially meaningful characters in a string. The returned string,
5372 used as a regular expression, will exactly match the original string.
5373
5374 Example:
5375 string quoted = RE::QuoteMeta(unquoted);
5376
5377 Note that it's legal to escape a character even if it has no special
5378 meaning in a regular expression -- so this function does that. (This
5379 also makes it identical to the perl function of the same name; see
5380 "perldoc -f quotemeta".) For example, "1.5-2.0?" becomes
5381 "1\.5\-2\.0\?".
5382
5383
5384 PARTIAL MATCHES
5385
5386 You can use the "PartialMatch" operation when you want the pattern to
5387 match any substring of the text.
5388
5389 Example: simple search for a string:
5390 pcrecpp::RE("ell").PartialMatch("hello");
5391
5392 Example: find first number in a string:
5393 int number;
5394 pcrecpp::RE re("(\\d+)");
5395 re.PartialMatch("x*100 + 20", &number);
5396 assert(number == 100);
5397
5398
5399 UTF-8 AND THE MATCHING INTERFACE
5400
5401 By default, pattern and text are plain text, one byte per character.
5402 The UTF8 flag, passed to the constructor, causes both pattern and
5403 string to be treated as UTF-8 text, still a byte stream but potentially
5404 multiple bytes per character. In practice, the text is likelier to be
5405 UTF-8 than the pattern, but the match returned may depend on the UTF8
5406 flag, so always use it when matching UTF8 text. For example, "." will
5407 match one byte normally but with UTF8 set may match up to three bytes
5408 of a multi-byte character.
5409
5410 Example:
5411 pcrecpp::RE_Options options;
5412 options.set_utf8();
5413 pcrecpp::RE re(utf8_pattern, options);
5414 re.FullMatch(utf8_string);
5415
5416 Example: using the convenience function UTF8():
5417 pcrecpp::RE re(utf8_pattern, pcrecpp::UTF8());
5418 re.FullMatch(utf8_string);
5419
5420 NOTE: The UTF8 flag is ignored if pcre was not configured with the
5421 --enable-utf8 flag.
5422
5423
5424 PASSING MODIFIERS TO THE REGULAR EXPRESSION ENGINE
5425
5426 PCRE defines some modifiers to change the behavior of the regular
5427 expression engine. The C++ wrapper defines an auxiliary class,
5428 RE_Options, as a vehicle to pass such modifiers to a RE class. Cur-
5429 rently, the following modifiers are supported:
5430
5431 modifier description Perl corresponding
5432
5433 PCRE_CASELESS case insensitive match /i
5434 PCRE_MULTILINE multiple lines match /m
5435 PCRE_DOTALL dot matches newlines /s
5436 PCRE_DOLLAR_ENDONLY $ matches only at end N/A
5437 PCRE_EXTRA strict escape parsing N/A
5438 PCRE_EXTENDED ignore whitespaces /x
5439 PCRE_UTF8 handles UTF8 chars built-in
5440 PCRE_UNGREEDY reverses * and *? N/A
5441 PCRE_NO_AUTO_CAPTURE disables capturing parens N/A (*)
5442
5443 (*) Both Perl and PCRE allow non capturing parentheses by means of the
5444 "?:" modifier within the pattern itself. e.g. (?:ab|cd) does not cap-
5445 ture, while (ab|cd) does.
5446
5447 For a full account on how each modifier works, please check the PCRE
5448 API reference page.
5449
5450 For each modifier, there are two member functions whose name is made
5451 out of the modifier in lowercase, without the "PCRE_" prefix. For
5452 instance, PCRE_CASELESS is handled by
5453
5454 bool caseless()
5455
5456 which returns true if the modifier is set, and
5457
5458 RE_Options & set_caseless(bool)
5459
5460 which sets or unsets the modifier. Moreover, PCRE_EXTRA_MATCH_LIMIT can
5461 be accessed through the set_match_limit() and match_limit() member
5462 functions. Setting match_limit to a non-zero value will limit the exe-
5463 cution of pcre to keep it from doing bad things like blowing the stack
5464 or taking an eternity to return a result. A value of 5000 is good
5465 enough to stop stack blowup in a 2MB thread stack. Setting match_limit
5466 to zero disables match limiting. Alternatively, you can call
5467 match_limit_recursion() which uses PCRE_EXTRA_MATCH_LIMIT_RECURSION to
5468 limit how much PCRE recurses. match_limit() limits the number of
5469 matches PCRE does; match_limit_recursion() limits the depth of internal
5470 recursion, and therefore the amount of stack that is used.
5471
5472 Normally, to pass one or more modifiers to a RE class, you declare a
5473 RE_Options object, set the appropriate options, and pass this object to
5474 a RE constructor. Example:
5475
5476 RE_options opt;
5477 opt.set_caseless(true);
5478 if (RE("HELLO", opt).PartialMatch("hello world")) ...
5479
5480 RE_options has two constructors. The default constructor takes no argu-
5481 ments and creates a set of flags that are off by default. The optional
5482 parameter option_flags is to facilitate transfer of legacy code from C
5483 programs. This lets you do
5484
5485 RE(pattern,
5486 RE_Options(PCRE_CASELESS|PCRE_MULTILINE)).PartialMatch(str);
5487
5488 However, new code is better off doing
5489
5490 RE(pattern,
5491 RE_Options().set_caseless(true).set_multiline(true))
5492 .PartialMatch(str);
5493
5494 If you are going to pass one of the most used modifiers, there are some
5495 convenience functions that return a RE_Options class with the appropri-
5496 ate modifier already set: CASELESS(), UTF8(), MULTILINE(), DOTALL(),
5497 and EXTENDED().
5498
5499 If you need to set several options at once, and you don't want to go
5500 through the pains of declaring a RE_Options object and setting several
5501 options, there is a parallel method that give you such ability on the
5502 fly. You can concatenate several set_xxxxx() member functions, since
5503 each of them returns a reference to its class object. For example, to
5504 pass PCRE_CASELESS, PCRE_EXTENDED, and PCRE_MULTILINE to a RE with one
5505 statement, you may write:
5506
5507 RE(" ^ xyz \\s+ .* blah$",
5508 RE_Options()
5509 .set_caseless(true)
5510 .set_extended(true)
5511 .set_multiline(true)).PartialMatch(sometext);
5512
5513
5514 SCANNING TEXT INCREMENTALLY
5515
5516 The "Consume" operation may be useful if you want to repeatedly match
5517 regular expressions at the front of a string and skip over them as they
5518 match. This requires use of the "StringPiece" type, which represents a
5519 sub-range of a real string. Like RE, StringPiece is defined in the
5520 pcrecpp namespace.
5521
5522 Example: read lines of the form "var = value" from a string.
5523 string contents = ...; // Fill string somehow
5524 pcrecpp::StringPiece input(contents); // Wrap in a StringPiece
5525
5526 string var;
5527 int value;
5528 pcrecpp::RE re("(\\w+) = (\\d+)\n");
5529 while (re.Consume(&input, &var, &value)) {
5530 ...;
5531 }
5532
5533 Each successful call to "Consume" will set "var/value", and also
5534 advance "input" so it points past the matched text.
5535
5536 The "FindAndConsume" operation is similar to "Consume" but does not
5537 anchor your match at the beginning of the string. For example, you
5538 could extract all words from a string by repeatedly calling
5539
5540 pcrecpp::RE("(\\w+)").FindAndConsume(&input, &word)
5541
5542
5543 PARSING HEX/OCTAL/C-RADIX NUMBERS
5544
5545 By default, if you pass a pointer to a numeric value, the corresponding
5546 text is interpreted as a base-10 number. You can instead wrap the
5547 pointer with a call to one of the operators Hex(), Octal(), or CRadix()
5548 to interpret the text in another base. The CRadix operator interprets
5549 C-style "0" (base-8) and "0x" (base-16) prefixes, but defaults to
5550 base-10.
5551
5552 Example:
5553 int a, b, c, d;
5554 pcrecpp::RE re("(.*) (.*) (.*) (.*)");
5555 re.FullMatch("100 40 0100 0x40",
5556 pcrecpp::Octal(&a), pcrecpp::Hex(&b),
5557 pcrecpp::CRadix(&c), pcrecpp::CRadix(&d));
5558
5559 will leave 64 in a, b, c, and d.
5560
5561
5562 REPLACING PARTS OF STRINGS
5563
5564 You can replace the first match of "pattern" in "str" with "rewrite".
5565 Within "rewrite", backslash-escaped digits (\1 to \9) can be used to
5566 insert text matching corresponding parenthesized group from the pat-
5567 tern. \0 in "rewrite" refers to the entire matching text. For example:
5568
5569 string s = "yabba dabba doo";
5570 pcrecpp::RE("b+").Replace("d", &s);
5571
5572 will leave "s" containing "yada dabba doo". The result is true if the
5573 pattern matches and a replacement occurs, false otherwise.
5574
5575 GlobalReplace is like Replace except that it replaces all occurrences
5576 of the pattern in the string with the rewrite. Replacements are not
5577 subject to re-matching. For example:
5578
5579 string s = "yabba dabba doo";
5580 pcrecpp::RE("b+").GlobalReplace("d", &s);
5581
5582 will leave "s" containing "yada dada doo". It returns the number of
5583 replacements made.
5584
5585 Extract is like Replace, except that if the pattern matches, "rewrite"
5586 is copied into "out" (an additional argument) with substitutions. The
5587 non-matching portions of "text" are ignored. Returns true iff a match
5588 occurred and the extraction happened successfully; if no match occurs,
5589 the string is left unaffected.
5590
5591
5592 AUTHOR
5593
5594 The C++ wrapper was contributed by Google Inc.
5595 Copyright (c) 2007 Google Inc.
5596
5597
5598 REVISION
5599
5600 Last updated: 06 March 2007
5601 ------------------------------------------------------------------------------
5602
5603
5604 PCRESAMPLE(3) PCRESAMPLE(3)
5605
5606
5607 NAME
5608 PCRE - Perl-compatible regular expressions
5609
5610
5611 PCRE SAMPLE PROGRAM
5612
5613 A simple, complete demonstration program, to get you started with using
5614 PCRE, is supplied in the file pcredemo.c in the PCRE distribution.
5615
5616 The program compiles the regular expression that is its first argument,
5617 and matches it against the subject string in its second argument. No
5618 PCRE options are set, and default character tables are used. If match-
5619 ing succeeds, the program outputs the portion of the subject that
5620 matched, together with the contents of any captured substrings.
5621
5622 If the -g option is given on the command line, the program then goes on
5623 to check for further matches of the same regular expression in the same
5624 subject string. The logic is a little bit tricky because of the possi-
5625 bility of matching an empty string. Comments in the code explain what
5626 is going on.
5627
5628 The demonstration program is automatically built if you use "./config-
5629 ure;make" to build PCRE. Otherwise, if PCRE is installed in the stan-
5630 dard include and library directories for your system, you should be
5631 able to compile the demonstration program using this command:
5632
5633 gcc -o pcredemo pcredemo.c -lpcre
5634
5635 If PCRE is installed elsewhere, you may need to add additional options
5636 to the command line. For example, on a Unix-like system that has PCRE
5637 installed in /usr/local, you can compile the demonstration program
5638 using a command like this:
5639
5640 gcc -o pcredemo -I/usr/local/include pcredemo.c \
5641 -L/usr/local/lib -lpcre
5642
5643 Once you have compiled the demonstration program, you can run simple
5644 tests like this:
5645
5646 ./pcredemo 'cat|dog' 'the cat sat on the mat'
5647 ./pcredemo -g 'cat|dog' 'the dog sat on the cat'
5648
5649 Note that there is a much more comprehensive test program, called
5650 pcretest, which supports many more facilities for testing regular
5651 expressions and the PCRE library. The pcredemo program is provided as a
5652 simple coding example.
5653
5654 On some operating systems (e.g. Solaris), when PCRE is not installed in
5655 the standard library directory, you may get an error like this when you
5656 try to run pcredemo:
5657
5658 ld.so.1: a.out: fatal: libpcre.so.0: open failed: No such file or
5659 directory
5660
5661 This is caused by the way shared library support works on those sys-
5662 tems. You need to add
5663
5664 -R/usr/local/lib
5665
5666 (for example) to the compile command to get round this problem.
5667
5668
5669 AUTHOR
5670
5671 Philip Hazel
5672 University Computing Service
5673 Cambridge CB2 3QH, England.
5674
5675
5676 REVISION
5677
5678 Last updated: 13 June 2007
5679 Copyright (c) 1997-2007 University of Cambridge.
5680 ------------------------------------------------------------------------------
5681 PCRESTACK(3) PCRESTACK(3)
5682
5683
5684 NAME
5685 PCRE - Perl-compatible regular expressions
5686
5687
5688 PCRE DISCUSSION OF STACK USAGE
5689
5690 When you call pcre_exec(), it makes use of an internal function called
5691 match(). This calls itself recursively at branch points in the pattern,
5692 in order to remember the state of the match so that it can back up and
5693 try a different alternative if the first one fails. As matching pro-
5694 ceeds deeper and deeper into the tree of possibilities, the recursion
5695 depth increases.
5696
5697 Not all calls of match() increase the recursion depth; for an item such
5698 as a* it may be called several times at the same level, after matching
5699 different numbers of a's. Furthermore, in a number of cases where the
5700 result of the recursive call would immediately be passed back as the
5701 result of the current call (a "tail recursion"), the function is just
5702 restarted instead.
5703
5704 The pcre_dfa_exec() function operates in an entirely different way, and
5705 hardly uses recursion at all. The limit on its complexity is the amount
5706 of workspace it is given. The comments that follow do NOT apply to
5707 pcre_dfa_exec(); they are relevant only for pcre_exec().
5708
5709 You can set limits on the number of times that match() is called, both
5710 in total and recursively. If the limit is exceeded, an error occurs.
5711 For details, see the section on extra data for pcre_exec() in the
5712 pcreapi documentation.
5713
5714 Each time that match() is actually called recursively, it uses memory
5715 from the process stack. For certain kinds of pattern and data, very
5716 large amounts of stack may be needed, despite the recognition of "tail
5717 recursion". You can often reduce the amount of recursion, and there-
5718 fore the amount of stack used, by modifying the pattern that is being
5719 matched. Consider, for example, this pattern:
5720
5721 ([^<]|<(?!inet))+
5722
5723 It matches from wherever it starts until it encounters "<inet" or the
5724 end of the data, and is the kind of pattern that might be used when
5725 processing an XML file. Each iteration of the outer parentheses matches
5726 either one character that is not "<" or a "<" that is not followed by
5727 "inet". However, each time a parenthesis is processed, a recursion
5728 occurs, so this formulation uses a stack frame for each matched charac-
5729 ter. For a long string, a lot of stack is required. Consider now this
5730 rewritten pattern, which matches exactly the same strings:
5731
5732 ([^<]++|<(?!inet))+
5733
5734 This uses very much less stack, because runs of characters that do not
5735 contain "<" are "swallowed" in one item inside the parentheses. Recur-
5736 sion happens only when a "<" character that is not followed by "inet"
5737 is encountered (and we assume this is relatively rare). A possessive
5738 quantifier is used to stop any backtracking into the runs of non-"<"
5739 characters, but that is not related to stack usage.
5740
5741 This example shows that one way of avoiding stack problems when match-
5742 ing long subject strings is to write repeated parenthesized subpatterns
5743 to match more than one character whenever possible.
5744
5745 In environments where stack memory is constrained, you might want to
5746 compile PCRE to use heap memory instead of stack for remembering back-
5747 up points. This makes it run a lot more slowly, however. Details of how
5748 to do this are given in the pcrebuild documentation. When built in this
5749 way, instead of using the stack, PCRE obtains and frees memory by call-
5750 ing the functions that are pointed to by the pcre_stack_malloc and
5751 pcre_stack_free variables. By default, these point to malloc() and
5752 free(), but you can replace the pointers to cause PCRE to use your own
5753 functions. Since the block sizes are always the same, and are always
5754 freed in reverse order, it may be possible to implement customized mem-
5755 ory handlers that are more efficient than the standard functions.
5756
5757 In Unix-like environments, there is not often a problem with the stack
5758 unless very long strings are involved, though the default limit on
5759 stack size varies from system to system. Values from 8Mb to 64Mb are
5760 common. You can find your default limit by running the command:
5761
5762 ulimit -s
5763
5764 Unfortunately, the effect of running out of stack is often SIGSEGV,
5765 though sometimes a more explicit error message is given. You can nor-
5766 mally increase the limit on stack size by code such as this:
5767
5768 struct rlimit rlim;
5769 getrlimit(RLIMIT_STACK, &rlim);
5770 rlim.rlim_cur = 100*1024*1024;
5771 setrlimit(RLIMIT_STACK, &rlim);
5772
5773 This reads the current limits (soft and hard) using getrlimit(), then
5774 attempts to increase the soft limit to 100Mb using setrlimit(). You
5775 must do this before calling pcre_exec().
5776
5777 PCRE has an internal counter that can be used to limit the depth of
5778 recursion, and thus cause pcre_exec() to give an error code before it
5779 runs out of stack. By default, the limit is very large, and unlikely
5780 ever to operate. It can be changed when PCRE is built, and it can also
5781 be set when pcre_exec() is called. For details of these interfaces, see
5782 the pcrebuild and pcreapi documentation.
5783
5784 As a very rough rule of thumb, you should reckon on about 500 bytes per
5785 recursion. Thus, if you want to limit your stack usage to 8Mb, you
5786 should set the limit at 16000 recursions. A 64Mb stack, on the other
5787 hand, can support around 128000 recursions. The pcretest test program
5788 has a command line option (-S) that can be used to increase the size of
5789 its stack.
5790
5791
5792 AUTHOR
5793
5794 Philip Hazel
5795 University Computing Service
5796 Cambridge CB2 3QH, England.
5797
5798
5799 REVISION
5800
5801 Last updated: 05 June 2007
5802 Copyright (c) 1997-2007 University of Cambridge.
5803 ------------------------------------------------------------------------------
5804
5805

Properties

Name Value
svn:eol-style native
svn:keywords "Author Date Id Revision Url"

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12