ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 678 - (show annotations)
Sun Aug 28 15:23:03 2011 UTC (9 years, 6 months ago) by ph10
File MIME type: text/plain
File size: 377710 byte(s)
Error occurred while calculating annotation data.
Documentation for JIT support.
1 -----------------------------------------------------------------------------
2 This file contains a concatenation of the PCRE man pages, converted to plain
3 text format for ease of searching with a text editor, or for use on systems
4 that do not have a man page processor. The small individual files that give
5 synopses of each function in the library have not been included. Neither has
6 the pcredemo program. There are separate text files for the pcregrep and
7 pcretest commands.
8 -----------------------------------------------------------------------------
11 PCRE(3) PCRE(3)
15 PCRE - Perl-compatible regular expressions
20 The PCRE library is a set of functions that implement regular expres-
21 sion pattern matching using the same syntax and semantics as Perl, with
22 just a few differences. Some features that appeared in Python and PCRE
23 before they appeared in Perl are also available using the Python syn-
24 tax, there is some support for one or two .NET and Oniguruma syntax
25 items, and there is an option for requesting some minor changes that
26 give better JavaScript compatibility.
28 The current implementation of PCRE corresponds approximately with Perl
29 5.12, including support for UTF-8 encoded strings and Unicode general
30 category properties. However, UTF-8 and Unicode support has to be
31 explicitly enabled; it is not the default. The Unicode tables corre-
32 spond to Unicode release 6.0.0.
34 In addition to the Perl-compatible matching function, PCRE contains an
35 alternative function that matches the same compiled patterns in a dif-
36 ferent way. In certain circumstances, the alternative function has some
37 advantages. For a discussion of the two matching algorithms, see the
38 pcrematching page.
40 PCRE is written in C and released as a C library. A number of people
41 have written wrappers and interfaces of various kinds. In particular,
42 Google Inc. have provided a comprehensive C++ wrapper. This is now
43 included as part of the PCRE distribution. The pcrecpp page has details
44 of this interface. Other people's contributions can be found in the
45 Contrib directory at the primary FTP site, which is:
47 ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre
49 Details of exactly which Perl regular expression features are and are
50 not supported by PCRE are given in separate documents. See the pcrepat-
51 tern and pcrecompat pages. There is a syntax summary in the pcresyntax
52 page.
54 Some features of PCRE can be included, excluded, or changed when the
55 library is built. The pcre_config() function makes it possible for a
56 client to discover which features are available. The features them-
57 selves are described in the pcrebuild page. Documentation about build-
58 ing PCRE for various operating systems can be found in the README and
59 NON-UNIX-USE files in the source distribution.
61 The library contains a number of undocumented internal functions and
62 data tables that are used by more than one of the exported external
63 functions, but which are not intended for use by external callers.
64 Their names all begin with "_pcre_", which hopefully will not provoke
65 any name clashes. In some environments, it is possible to control which
66 external symbols are exported when a shared library is built, and in
67 these cases the undocumented symbols are not exported.
72 The user documentation for PCRE comprises a number of different sec-
73 tions. In the "man" format, each of these is a separate "man page". In
74 the HTML format, each is a separate page, linked from the index page.
75 In the plain text format, all the sections, except the pcredemo sec-
76 tion, are concatenated, for ease of searching. The sections are as fol-
77 lows:
79 pcre this document
80 pcre-config show PCRE installation configuration information
81 pcreapi details of PCRE's native C API
82 pcrebuild options for building PCRE
83 pcrecallout details of the callout feature
84 pcrecompat discussion of Perl compatibility
85 pcrecpp details of the C++ wrapper
86 pcredemo a demonstration C program that uses PCRE
87 pcregrep description of the pcregrep command
88 pcrejit discussion of the just-in-time optimization support
89 pcrelimits details of size and other limits
90 pcrematching discussion of the two matching algorithms
91 pcrepartial details of the partial matching facility
92 pcrepattern syntax and semantics of supported
93 regular expressions
94 pcreperform discussion of performance issues
95 pcreposix the POSIX-compatible C API
96 pcreprecompile details of saving and re-using precompiled patterns
97 pcresample discussion of the pcredemo program
98 pcrestack discussion of stack usage
99 pcresyntax quick syntax reference
100 pcretest description of the pcretest testing command
101 pcreunicode discussion of Unicode and UTF-8 support
103 In addition, in the "man" and HTML formats, there is a short page for
104 each C library function, listing its arguments and results.
109 Philip Hazel
110 University Computing Service
111 Cambridge CB2 3QH, England.
113 Putting an actual email address here seems to have been a spam magnet,
114 so I've taken it away. If you want to email me, use my two initials,
115 followed by the two digits 10, at the domain cam.ac.uk.
120 Last updated: 24 August 2011
121 Copyright (c) 1997-2011 University of Cambridge.
122 ------------------------------------------------------------------------------
128 NAME
129 PCRE - Perl-compatible regular expressions
134 This document describes the optional features of PCRE that can be
135 selected when the library is compiled. It assumes use of the configure
136 script, where the optional features are selected or deselected by pro-
137 viding options to configure before running the make command. However,
138 the same options can be selected in both Unix-like and non-Unix-like
139 environments using the GUI facility of cmake-gui if you are using CMake
140 instead of configure to build PCRE.
142 There is a lot more information about building PCRE in non-Unix-like
143 environments in the file called NON_UNIX_USE, which is part of the PCRE
144 distribution. You should consult this file as well as the README file
145 if you are building in a non-Unix-like environment.
147 The complete list of options for configure (which includes the standard
148 ones such as the selection of the installation directory) can be
149 obtained by running
151 ./configure --help
153 The following sections include descriptions of options whose names
154 begin with --enable or --disable. These settings specify changes to the
155 defaults for the configure command. Because of the way that configure
156 works, --enable and --disable always come in pairs, so the complemen-
157 tary option always exists as well, but as it specifies the default, it
158 is not described.
163 The PCRE building process uses libtool to build both shared and static
164 Unix libraries by default. You can suppress one of these by adding one
165 of
167 --disable-shared
168 --disable-static
170 to the configure command, as required.
175 By default, the configure script will search for a C++ compiler and C++
176 header files. If it finds them, it automatically builds the C++ wrapper
177 library for PCRE. You can disable this by adding
179 --disable-cpp
181 to the configure command.
186 To build PCRE with support for UTF-8 Unicode character strings, add
188 --enable-utf8
190 to the configure command. Of itself, this does not make PCRE treat
191 strings as UTF-8. As well as compiling PCRE with this option, you also
192 have have to set the PCRE_UTF8 option when you call the pcre_compile()
193 or pcre_compile2() functions.
195 If you set --enable-utf8 when compiling in an EBCDIC environment, PCRE
196 expects its input to be either ASCII or UTF-8 (depending on the runtime
197 option). It is not possible to support both EBCDIC and UTF-8 codes in
198 the same version of the library. Consequently, --enable-utf8 and
199 --enable-ebcdic are mutually exclusive.
204 UTF-8 support allows PCRE to process character values greater than 255
205 in the strings that it handles. On its own, however, it does not pro-
206 vide any facilities for accessing the properties of such characters. If
207 you want to be able to use the pattern escapes \P, \p, and \X, which
208 refer to Unicode character properties, you must add
210 --enable-unicode-properties
212 to the configure command. This implies UTF-8 support, even if you have
213 not explicitly requested it.
215 Including Unicode property support adds around 30K of tables to the
216 PCRE library. Only the general category properties such as Lu and Nd
217 are supported. Details are given in the pcrepattern documentation.
222 By default, PCRE interprets the linefeed (LF) character as indicating
223 the end of a line. This is the normal newline character on Unix-like
224 systems. You can compile PCRE to use carriage return (CR) instead, by
225 adding
227 --enable-newline-is-cr
229 to the configure command. There is also a --enable-newline-is-lf
230 option, which explicitly specifies linefeed as the newline character.
232 Alternatively, you can specify that line endings are to be indicated by
233 the two character sequence CRLF. If you want this, add
235 --enable-newline-is-crlf
237 to the configure command. There is a fourth option, specified by
239 --enable-newline-is-anycrlf
241 which causes PCRE to recognize any of the three sequences CR, LF, or
242 CRLF as indicating a line ending. Finally, a fifth option, specified by
244 --enable-newline-is-any
246 causes PCRE to recognize any Unicode newline sequence.
248 Whatever line ending convention is selected when PCRE is built can be
249 overridden when the library functions are called. At build time it is
250 conventional to use the standard for your operating system.
255 By default, the sequence \R in a pattern matches any Unicode newline
256 sequence, whatever has been selected as the line ending sequence. If
257 you specify
259 --enable-bsr-anycrlf
261 the default is changed so that \R matches only CR, LF, or CRLF. What-
262 ever is selected when PCRE is built can be overridden when the library
263 functions are called.
268 When PCRE is called through the POSIX interface (see the pcreposix doc-
269 umentation), additional working storage is required for holding the
270 pointers to capturing substrings, because PCRE requires three integers
271 per substring, whereas the POSIX interface provides only two. If the
272 number of expected substrings is small, the wrapper function uses space
273 on the stack, because this is faster than using malloc() for each call.
274 The default threshold above which the stack is no longer used is 10; it
275 can be changed by adding a setting such as
277 --with-posix-malloc-threshold=20
279 to the configure command.
284 Within a compiled pattern, offset values are used to point from one
285 part to another (for example, from an opening parenthesis to an alter-
286 nation metacharacter). By default, two-byte values are used for these
287 offsets, leading to a maximum size for a compiled pattern of around
288 64K. This is sufficient to handle all but the most gigantic patterns.
289 Nevertheless, some people do want to process truyl enormous patterns,
290 so it is possible to compile PCRE to use three-byte or four-byte off-
291 sets by adding a setting such as
293 --with-link-size=3
295 to the configure command. The value given must be 2, 3, or 4. Using
296 longer offsets slows down the operation of PCRE because it has to load
297 additional bytes when handling them.
302 When matching with the pcre_exec() function, PCRE implements backtrack-
303 ing by making recursive calls to an internal function called match().
304 In environments where the size of the stack is limited, this can se-
305 verely limit PCRE's operation. (The Unix environment does not usually
306 suffer from this problem, but it may sometimes be necessary to increase
307 the maximum stack size. There is a discussion in the pcrestack docu-
308 mentation.) An alternative approach to recursion that uses memory from
309 the heap to remember data, instead of using recursive function calls,
310 has been implemented to work round the problem of limited stack size.
311 If you want to build a version of PCRE that works this way, add
313 --disable-stack-for-recursion
315 to the configure command. With this configuration, PCRE will use the
316 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
317 ment functions. By default these point to malloc() and free(), but you
318 can replace the pointers so that your own functions are used instead.
320 Separate functions are provided rather than using pcre_malloc and
321 pcre_free because the usage is very predictable: the block sizes
322 requested are always the same, and the blocks are always freed in
323 reverse order. A calling program might be able to implement optimized
324 functions that perform better than malloc() and free(). PCRE runs
325 noticeably more slowly when built in this way. This option affects only
326 the pcre_exec() function; it is not relevant for pcre_dfa_exec().
331 Internally, PCRE has a function called match(), which it calls repeat-
332 edly (sometimes recursively) when matching a pattern with the
333 pcre_exec() function. By controlling the maximum number of times this
334 function may be called during a single matching operation, a limit can
335 be placed on the resources used by a single call to pcre_exec(). The
336 limit can be changed at run time, as described in the pcreapi documen-
337 tation. The default is 10 million, but this can be changed by adding a
338 setting such as
340 --with-match-limit=500000
342 to the configure command. This setting has no effect on the
343 pcre_dfa_exec() matching function.
345 In some environments it is desirable to limit the depth of recursive
346 calls of match() more strictly than the total number of calls, in order
347 to restrict the maximum amount of stack (or heap, if --disable-stack-
348 for-recursion is specified) that is used. A second limit controls this;
349 it defaults to the value that is set for --with-match-limit, which
350 imposes no additional constraints. However, you can set a lower limit
351 by adding, for example,
353 --with-match-limit-recursion=10000
355 to the configure command. This value can also be overridden at run
356 time.
361 PCRE uses fixed tables for processing characters whose code values are
362 less than 256. By default, PCRE is built with a set of tables that are
363 distributed in the file pcre_chartables.c.dist. These tables are for
364 ASCII codes only. If you add
366 --enable-rebuild-chartables
368 to the configure command, the distributed tables are no longer used.
369 Instead, a program called dftables is compiled and run. This outputs
370 the source for new set of tables, created in the default locale of your
371 C runtime system. (This method of replacing the tables does not work if
372 you are cross compiling, because dftables is run on the local host. If
373 you need to create alternative tables when cross compiling, you will
374 have to do so "by hand".)
379 PCRE assumes by default that it will run in an environment where the
380 character code is ASCII (or Unicode, which is a superset of ASCII).
381 This is the case for most computer operating systems. PCRE can, how-
382 ever, be compiled to run in an EBCDIC environment by adding
384 --enable-ebcdic
386 to the configure command. This setting implies --enable-rebuild-charta-
387 bles. You should only use it if you know that you are in an EBCDIC
388 environment (for example, an IBM mainframe operating system). The
389 --enable-ebcdic option is incompatible with --enable-utf8.
394 By default, pcregrep reads all files as plain text. You can build it so
395 that it recognizes files whose names end in .gz or .bz2, and reads them
396 with libz or libbz2, respectively, by adding one or both of
398 --enable-pcregrep-libz
399 --enable-pcregrep-libbz2
401 to the configure command. These options naturally require that the rel-
402 evant libraries are installed on your system. Configuration will fail
403 if they are not.
408 pcregrep uses an internal buffer to hold a "window" on the file it is
409 scanning, in order to be able to output "before" and "after" lines when
410 it finds a match. The size of the buffer is controlled by a parameter
411 whose default value is 20K. The buffer itself is three times this size,
412 but because of the way it is used for holding "before" lines, the long-
413 est line that is guaranteed to be processable is the parameter size.
414 You can change the default parameter value by adding, for example,
416 --with-pcregrep-bufsize=50K
418 to the configure command. The caller of pcregrep can, however, override
419 this value by specifying a run-time option.
424 If you add
426 --enable-pcretest-libreadline
428 to the configure command, pcretest is linked with the libreadline
429 library, and when its input is from a terminal, it reads it using the
430 readline() function. This provides line-editing and history facilities.
431 Note that libreadline is GPL-licensed, so if you distribute a binary of
432 pcretest linked in this way, there may be licensing issues.
434 Setting this option causes the -lreadline option to be added to the
435 pcretest build. In many operating environments with a sytem-installed
436 libreadline this is sufficient. However, in some environments (e.g. if
437 an unmodified distribution version of readline is in use), some extra
438 configuration may be necessary. The INSTALL file for libreadline says
439 this:
441 "Readline uses the termcap functions, but does not link with the
442 termcap or curses library itself, allowing applications which link
443 with readline the to choose an appropriate library."
445 If your environment has not been set up so that an appropriate library
446 is automatically included, you may need to add something like
448 LIBS="-ncurses"
450 immediately before the configure command.
455 pcreapi(3), pcre_config(3).
460 Philip Hazel
461 University Computing Service
462 Cambridge CB2 3QH, England.
467 Last updated: 02 August 2011
468 Copyright (c) 1997-2011 University of Cambridge.
469 ------------------------------------------------------------------------------
475 NAME
476 PCRE - Perl-compatible regular expressions
481 This document describes the two different algorithms that are available
482 in PCRE for matching a compiled regular expression against a given sub-
483 ject string. The "standard" algorithm is the one provided by the
484 pcre_exec() function. This works in the same was as Perl's matching
485 function, and provides a Perl-compatible matching operation.
487 An alternative algorithm is provided by the pcre_dfa_exec() function;
488 this operates in a different way, and is not Perl-compatible. It has
489 advantages and disadvantages compared with the standard algorithm, and
490 these are described below.
492 When there is only one possible way in which a given subject string can
493 match a pattern, the two algorithms give the same answer. A difference
494 arises, however, when there are multiple possibilities. For example, if
495 the pattern
497 ^<.*>
499 is matched against the string
501 <something> <something else> <something further>
503 there are three possible answers. The standard algorithm finds only one
504 of them, whereas the alternative algorithm finds all three.
509 The set of strings that are matched by a regular expression can be rep-
510 resented as a tree structure. An unlimited repetition in the pattern
511 makes the tree of infinite size, but it is still a tree. Matching the
512 pattern to a given subject string (from a given starting point) can be
513 thought of as a search of the tree. There are two ways to search a
514 tree: depth-first and breadth-first, and these correspond to the two
515 matching algorithms provided by PCRE.
520 In the terminology of Jeffrey Friedl's book "Mastering Regular Expres-
521 sions", the standard algorithm is an "NFA algorithm". It conducts a
522 depth-first search of the pattern tree. That is, it proceeds along a
523 single path through the tree, checking that the subject matches what is
524 required. When there is a mismatch, the algorithm tries any alterna-
525 tives at the current point, and if they all fail, it backs up to the
526 previous branch point in the tree, and tries the next alternative
527 branch at that level. This often involves backing up (moving to the
528 left) in the subject string as well. The order in which repetition
529 branches are tried is controlled by the greedy or ungreedy nature of
530 the quantifier.
532 If a leaf node is reached, a matching string has been found, and at
533 that point the algorithm stops. Thus, if there is more than one possi-
534 ble match, this algorithm returns the first one that it finds. Whether
535 this is the shortest, the longest, or some intermediate length depends
536 on the way the greedy and ungreedy repetition quantifiers are specified
537 in the pattern.
539 Because it ends up with a single path through the tree, it is rela-
540 tively straightforward for this algorithm to keep track of the sub-
541 strings that are matched by portions of the pattern in parentheses.
542 This provides support for capturing parentheses and back references.
547 This algorithm conducts a breadth-first search of the tree. Starting
548 from the first matching point in the subject, it scans the subject
549 string from left to right, once, character by character, and as it does
550 this, it remembers all the paths through the tree that represent valid
551 matches. In Friedl's terminology, this is a kind of "DFA algorithm",
552 though it is not implemented as a traditional finite state machine (it
553 keeps multiple states active simultaneously).
555 Although the general principle of this matching algorithm is that it
556 scans the subject string only once, without backtracking, there is one
557 exception: when a lookaround assertion is encountered, the characters
558 following or preceding the current point have to be independently
559 inspected.
561 The scan continues until either the end of the subject is reached, or
562 there are no more unterminated paths. At this point, terminated paths
563 represent the different matching possibilities (if there are none, the
564 match has failed). Thus, if there is more than one possible match,
565 this algorithm finds all of them, and in particular, it finds the long-
566 est. The matches are returned in decreasing order of length. There is
567 an option to stop the algorithm after the first match (which is neces-
568 sarily the shortest) is found.
570 Note that all the matches that are found start at the same point in the
571 subject. If the pattern
573 cat(er(pillar)?)?
575 is matched against the string "the caterpillar catchment", the result
576 will be the three strings "caterpillar", "cater", and "cat" that start
577 at the fifth character of the subject. The algorithm does not automati-
578 cally move on to find matches that start at later positions.
580 There are a number of features of PCRE regular expressions that are not
581 supported by the alternative matching algorithm. They are as follows:
583 1. Because the algorithm finds all possible matches, the greedy or
584 ungreedy nature of repetition quantifiers is not relevant. Greedy and
585 ungreedy quantifiers are treated in exactly the same way. However, pos-
586 sessive quantifiers can make a difference when what follows could also
587 match what is quantified, for example in a pattern like this:
589 ^a++\w!
591 This pattern matches "aaab!" but not "aaa!", which would be matched by
592 a non-possessive quantifier. Similarly, if an atomic group is present,
593 it is matched as if it were a standalone pattern at the current point,
594 and the longest match is then "locked in" for the rest of the overall
595 pattern.
597 2. When dealing with multiple paths through the tree simultaneously, it
598 is not straightforward to keep track of captured substrings for the
599 different matching possibilities, and PCRE's implementation of this
600 algorithm does not attempt to do this. This means that no captured sub-
601 strings are available.
603 3. Because no substrings are captured, back references within the pat-
604 tern are not supported, and cause errors if encountered.
606 4. For the same reason, conditional expressions that use a backrefer-
607 ence as the condition or test for a specific group recursion are not
608 supported.
610 5. Because many paths through the tree may be active, the \K escape
611 sequence, which resets the start of the match when encountered (but may
612 be on some paths and not on others), is not supported. It causes an
613 error if encountered.
615 6. Callouts are supported, but the value of the capture_top field is
616 always 1, and the value of the capture_last field is always -1.
618 7. The \C escape sequence, which (in the standard algorithm) matches a
619 single byte, even in UTF-8 mode, is not supported because the alterna-
620 tive algorithm moves through the subject string one character at a
621 time, for all active paths through the tree.
623 8. Except for (*FAIL), the backtracking control verbs such as (*PRUNE)
624 are not supported. (*FAIL) is supported, and behaves like a failing
625 negative assertion.
630 Using the alternative matching algorithm provides the following advan-
631 tages:
633 1. All possible matches (at a single point in the subject) are automat-
634 ically found, and in particular, the longest match is found. To find
635 more than one match using the standard algorithm, you have to do kludgy
636 things with callouts.
638 2. Because the alternative algorithm scans the subject string just
639 once, and never needs to backtrack, it is possible to pass very long
640 subject strings to the matching function in several pieces, checking
641 for partial matching each time. Although it is possible to do multi-
642 segment matching using the standard algorithm (pcre_exec()), by retain-
643 ing partially matched substrings, it is more complicated. The pcrepar-
644 tial documentation gives details of partial matching and discusses
645 multi-segment matching.
650 The alternative algorithm suffers from a number of disadvantages:
652 1. It is substantially slower than the standard algorithm. This is
653 partly because it has to search for all possible matches, but is also
654 because it is less susceptible to optimization.
656 2. Capturing parentheses and back references are not supported.
658 3. Although atomic groups are supported, their use does not provide the
659 performance advantage that it does for the standard algorithm.
664 Philip Hazel
665 University Computing Service
666 Cambridge CB2 3QH, England.
671 Last updated: 17 November 2010
672 Copyright (c) 1997-2010 University of Cambridge.
673 ------------------------------------------------------------------------------
679 NAME
680 PCRE - Perl-compatible regular expressions
685 #include <pcre.h>
687 pcre *pcre_compile(const char *pattern, int options,
688 const char **errptr, int *erroffset,
689 const unsigned char *tableptr);
691 pcre *pcre_compile2(const char *pattern, int options,
692 int *errorcodeptr,
693 const char **errptr, int *erroffset,
694 const unsigned char *tableptr);
696 pcre_extra *pcre_study(const pcre *code, int options,
697 const char **errptr);
699 int pcre_exec(const pcre *code, const pcre_extra *extra,
700 const char *subject, int length, int startoffset,
701 int options, int *ovector, int ovecsize);
703 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
704 const char *subject, int length, int startoffset,
705 int options, int *ovector, int ovecsize,
706 int *workspace, int wscount);
708 int pcre_copy_named_substring(const pcre *code,
709 const char *subject, int *ovector,
710 int stringcount, const char *stringname,
711 char *buffer, int buffersize);
713 int pcre_copy_substring(const char *subject, int *ovector,
714 int stringcount, int stringnumber, char *buffer,
715 int buffersize);
717 int pcre_get_named_substring(const pcre *code,
718 const char *subject, int *ovector,
719 int stringcount, const char *stringname,
720 const char **stringptr);
722 int pcre_get_stringnumber(const pcre *code,
723 const char *name);
725 int pcre_get_stringtable_entries(const pcre *code,
726 const char *name, char **first, char **last);
728 int pcre_get_substring(const char *subject, int *ovector,
729 int stringcount, int stringnumber,
730 const char **stringptr);
732 int pcre_get_substring_list(const char *subject,
733 int *ovector, int stringcount, const char ***listptr);
735 void pcre_free_substring(const char *stringptr);
737 void pcre_free_substring_list(const char **stringptr);
739 const unsigned char *pcre_maketables(void);
741 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
742 int what, void *where);
744 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
746 int pcre_refcount(pcre *code, int adjust);
748 int pcre_config(int what, void *where);
750 char *pcre_version(void);
752 void *(*pcre_malloc)(size_t);
754 void (*pcre_free)(void *);
756 void *(*pcre_stack_malloc)(size_t);
758 void (*pcre_stack_free)(void *);
760 int (*pcre_callout)(pcre_callout_block *);
765 PCRE has its own native API, which is described in this document. There
766 are also some wrapper functions that correspond to the POSIX regular
767 expression API. These are described in the pcreposix documentation.
768 Both of these APIs define a set of C function calls. A C++ wrapper is
769 distributed with PCRE. It is documented in the pcrecpp page.
771 The native API C function prototypes are defined in the header file
772 pcre.h, and on Unix systems the library itself is called libpcre. It
773 can normally be accessed by adding -lpcre to the command for linking an
774 application that uses PCRE. The header file defines the macros
775 PCRE_MAJOR and PCRE_MINOR to contain the major and minor release num-
776 bers for the library. Applications can use these to include support
777 for different releases of PCRE.
779 In a Windows environment, if you want to statically link an application
780 program against a non-dll pcre.a file, you must define PCRE_STATIC
781 before including pcre.h or pcrecpp.h, because otherwise the pcre_mal-
782 loc() and pcre_free() exported functions will be declared
783 __declspec(dllimport), with unwanted results.
785 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
786 pcre_exec() are used for compiling and matching regular expressions in
787 a Perl-compatible manner. A sample program that demonstrates the sim-
788 plest way of using them is provided in the file called pcredemo.c in
789 the PCRE source distribution. A listing of this program is given in the
790 pcredemo documentation, and the pcresample documentation describes how
791 to compile and run it.
793 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
794 ble, is also provided. This uses a different algorithm for the match-
795 ing. The alternative algorithm finds all possible matches (at a given
796 point in the subject), and scans the subject just once (unless there
797 are lookbehind assertions). However, this algorithm does not return
798 captured substrings. A description of the two matching algorithms and
799 their advantages and disadvantages is given in the pcrematching docu-
800 mentation.
802 In addition to the main compiling and matching functions, there are
803 convenience functions for extracting captured substrings from a subject
804 string that is matched by pcre_exec(). They are:
806 pcre_copy_substring()
807 pcre_copy_named_substring()
808 pcre_get_substring()
809 pcre_get_named_substring()
810 pcre_get_substring_list()
811 pcre_get_stringnumber()
812 pcre_get_stringtable_entries()
814 pcre_free_substring() and pcre_free_substring_list() are also provided,
815 to free the memory used for extracted strings.
817 The function pcre_maketables() is used to build a set of character
818 tables in the current locale for passing to pcre_compile(),
819 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
820 provided for specialist use. Most commonly, no special tables are
821 passed, in which case internal tables that are generated when PCRE is
822 built are used.
824 The function pcre_fullinfo() is used to find out information about a
825 compiled pattern; pcre_info() is an obsolete version that returns only
826 some of the available information, but is retained for backwards com-
827 patibility. The function pcre_version() returns a pointer to a string
828 containing the version of PCRE and its date of release.
830 The function pcre_refcount() maintains a reference count in a data
831 block containing a compiled pattern. This is provided for the benefit
832 of object-oriented applications.
834 The global variables pcre_malloc and pcre_free initially contain the
835 entry points of the standard malloc() and free() functions, respec-
836 tively. PCRE calls the memory management functions via these variables,
837 so a calling program can replace them if it wishes to intercept the
838 calls. This should be done before calling any PCRE functions.
840 The global variables pcre_stack_malloc and pcre_stack_free are also
841 indirections to memory management functions. These special functions
842 are used only when PCRE is compiled to use the heap for remembering
843 data, instead of recursive function calls, when running the pcre_exec()
844 function. See the pcrebuild documentation for details of how to do
845 this. It is a non-standard way of building PCRE, for use in environ-
846 ments that have limited stacks. Because of the greater use of memory
847 management, it runs more slowly. Separate functions are provided so
848 that special-purpose external code can be used for this case. When
849 used, these functions are always called in a stack-like manner (last
850 obtained, first freed), and always for memory blocks of the same size.
851 There is a discussion about PCRE's stack usage in the pcrestack docu-
852 mentation.
854 The global variable pcre_callout initially contains NULL. It can be set
855 by the caller to a "callout" function, which PCRE will then call at
856 specified points during a matching operation. Details are given in the
857 pcrecallout documentation.
862 PCRE supports five different conventions for indicating line breaks in
863 strings: a single CR (carriage return) character, a single LF (line-
864 feed) character, the two-character sequence CRLF, any of the three pre-
865 ceding, or any Unicode newline sequence. The Unicode newline sequences
866 are the three just mentioned, plus the single characters VT (vertical
867 tab, U+000B), FF (formfeed, U+000C), NEL (next line, U+0085), LS (line
868 separator, U+2028), and PS (paragraph separator, U+2029).
870 Each of the first three conventions is used by at least one operating
871 system as its standard newline sequence. When PCRE is built, a default
872 can be specified. The default default is LF, which is the Unix stan-
873 dard. When PCRE is run, the default can be overridden, either when a
874 pattern is compiled, or when it is matched.
876 At compile time, the newline convention can be specified by the options
877 argument of pcre_compile(), or it can be specified by special text at
878 the start of the pattern itself; this overrides any other settings. See
879 the pcrepattern page for details of the special character sequences.
881 In the PCRE documentation the word "newline" is used to mean "the char-
882 acter or pair of characters that indicate a line break". The choice of
883 newline convention affects the handling of the dot, circumflex, and
884 dollar metacharacters, the handling of #-comments in /x mode, and, when
885 CRLF is a recognized line ending sequence, the match position advance-
886 ment for a non-anchored pattern. There is more detail about this in the
887 section on pcre_exec() options below.
889 The choice of newline convention does not affect the interpretation of
890 the \n or \r escape sequences, nor does it affect what \R matches,
891 which is controlled in a similar way, but by separate options.
896 The PCRE functions can be used in multi-threading applications, with
897 the proviso that the memory management functions pointed to by
898 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
899 callout function pointed to by pcre_callout, are shared by all threads.
901 The compiled form of a regular expression is not altered during match-
902 ing, so the same compiled pattern can safely be used by several threads
903 at once.
908 The compiled form of a regular expression can be saved and re-used at a
909 later time, possibly by a different program, and even on a host other
910 than the one on which it was compiled. Details are given in the
911 pcreprecompile documentation. However, compiling a regular expression
912 with one version of PCRE for use with a different version is not guar-
913 anteed to work and may cause crashes.
918 int pcre_config(int what, void *where);
920 The function pcre_config() makes it possible for a PCRE client to dis-
921 cover which optional features have been compiled into the PCRE library.
922 The pcrebuild documentation has more details about these optional fea-
923 tures.
925 The first argument for pcre_config() is an integer, specifying which
926 information is required; the second argument is a pointer to a variable
927 into which the information is placed. The following information is
928 available:
932 The output is an integer that is set to one if UTF-8 support is avail-
933 able; otherwise it is set to zero.
937 The output is an integer that is set to one if support for Unicode
938 character properties is available; otherwise it is set to zero.
942 The output is an integer whose value specifies the default character
943 sequence that is recognized as meaning "newline". The four values that
944 are supported are: 10 for LF, 13 for CR, 3338 for CRLF, -2 for ANYCRLF,
945 and -1 for ANY. Though they are derived from ASCII, the same values
946 are returned in EBCDIC environments. The default should normally corre-
947 spond to the standard sequence for your operating system.
951 The output is an integer whose value indicates what character sequences
952 the \R escape sequence matches by default. A value of 0 means that \R
953 matches any Unicode line ending sequence; a value of 1 means that \R
954 matches only CR, LF, or CRLF. The default can be overridden when a pat-
955 tern is compiled or matched.
959 The output is an integer that contains the number of bytes used for
960 internal linkage in compiled regular expressions. The value is 2, 3, or
961 4. Larger values allow larger regular expressions to be compiled, at
962 the expense of slower matching. The default value of 2 is sufficient
963 for all but the most massive patterns, since it allows the compiled
964 pattern to be up to 64K in size.
968 The output is an integer that contains the threshold above which the
969 POSIX interface uses malloc() for output vectors. Further details are
970 given in the pcreposix documentation.
974 The output is a long integer that gives the default limit for the num-
975 ber of internal matching function calls in a pcre_exec() execution.
976 Further details are given with pcre_exec() below.
980 The output is a long integer that gives the default limit for the depth
981 of recursion when calling the internal matching function in a
982 pcre_exec() execution. Further details are given with pcre_exec()
983 below.
987 The output is an integer that is set to one if internal recursion when
988 running pcre_exec() is implemented by recursive function calls that use
989 the stack to remember their state. This is the usual way that PCRE is
990 compiled. The output is zero if PCRE was compiled to use blocks of data
991 on the heap instead of recursive function calls. In this case,
992 pcre_stack_malloc and pcre_stack_free are called to manage memory
993 blocks on the heap, thus avoiding the use of the stack.
998 pcre *pcre_compile(const char *pattern, int options,
999 const char **errptr, int *erroffset,
1000 const unsigned char *tableptr);
1002 pcre *pcre_compile2(const char *pattern, int options,
1003 int *errorcodeptr,
1004 const char **errptr, int *erroffset,
1005 const unsigned char *tableptr);
1007 Either of the functions pcre_compile() or pcre_compile2() can be called
1008 to compile a pattern into an internal form. The only difference between
1009 the two interfaces is that pcre_compile2() has an additional argument,
1010 errorcodeptr, via which a numerical error code can be returned. To
1011 avoid too much repetition, we refer just to pcre_compile() below, but
1012 the information applies equally to pcre_compile2().
1014 The pattern is a C string terminated by a binary zero, and is passed in
1015 the pattern argument. A pointer to a single block of memory that is
1016 obtained via pcre_malloc is returned. This contains the compiled code
1017 and related data. The pcre type is defined for the returned block; this
1018 is a typedef for a structure whose contents are not externally defined.
1019 It is up to the caller to free the memory (via pcre_free) when it is no
1020 longer required.
1022 Although the compiled code of a PCRE regex is relocatable, that is, it
1023 does not depend on memory location, the complete pcre data block is not
1024 fully relocatable, because it may contain a copy of the tableptr argu-
1025 ment, which is an address (see below).
1027 The options argument contains various bit settings that affect the com-
1028 pilation. It should be zero if no options are required. The available
1029 options are described below. Some of them (in particular, those that
1030 are compatible with Perl, but some others as well) can also be set and
1031 unset from within the pattern (see the detailed description in the
1032 pcrepattern documentation). For those options that can be different in
1033 different parts of the pattern, the contents of the options argument
1034 specifies their settings at the start of compilation and execution. The
1036 PCRE_NO_START_OPT options can be set at the time of matching as well as
1037 at compile time.
1039 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
1040 if compilation of a pattern fails, pcre_compile() returns NULL, and
1041 sets the variable pointed to by errptr to point to a textual error mes-
1042 sage. This is a static string that is part of the library. You must not
1043 try to free it. Normally, the offset from the start of the pattern to
1044 the byte that was being processed when the error was discovered is
1045 placed in the variable pointed to by erroffset, which must not be NULL
1046 (if it is, an immediate error is given). However, for an invalid UTF-8
1047 string, the offset is that of the first byte of the failing character.
1048 Also, some errors are not detected until checks are carried out when
1049 the whole pattern has been scanned; in these cases the offset passed
1050 back is the length of the pattern.
1052 Note that the offset is in bytes, not characters, even in UTF-8 mode.
1053 It may sometimes point into the middle of a UTF-8 character.
1055 If pcre_compile2() is used instead of pcre_compile(), and the error-
1056 codeptr argument is not NULL, a non-zero error code number is returned
1057 via this argument in the event of an error. This is in addition to the
1058 textual error message. Error codes and messages are listed below.
1060 If the final argument, tableptr, is NULL, PCRE uses a default set of
1061 character tables that are built when PCRE is compiled, using the
1062 default C locale. Otherwise, tableptr must be an address that is the
1063 result of a call to pcre_maketables(). This value is stored with the
1064 compiled pattern, and used again by pcre_exec(), unless another table
1065 pointer is passed to it. For more discussion, see the section on locale
1066 support below.
1068 This code fragment shows a typical straightforward call to pcre_com-
1069 pile():
1071 pcre *re;
1072 const char *error;
1073 int erroffset;
1074 re = pcre_compile(
1075 "^A.*Z", /* the pattern */
1076 0, /* default options */
1077 &error, /* for error message */
1078 &erroffset, /* for error offset */
1079 NULL); /* use default character tables */
1081 The following names for option bits are defined in the pcre.h header
1082 file:
1086 If this bit is set, the pattern is forced to be "anchored", that is, it
1087 is constrained to match only at the first matching point in the string
1088 that is being searched (the "subject string"). This effect can also be
1089 achieved by appropriate constructs in the pattern itself, which is the
1090 only way to do it in Perl.
1094 If this bit is set, pcre_compile() automatically inserts callout items,
1095 all with number 255, before each pattern item. For discussion of the
1096 callout facility, see the pcrecallout documentation.
1101 These options (which are mutually exclusive) control what the \R escape
1102 sequence matches. The choice is either to match only CR, LF, or CRLF,
1103 or to match any Unicode newline sequence. The default is specified when
1104 PCRE is built. It can be overridden from within the pattern, or by set-
1105 ting an option when a compiled pattern is matched.
1109 If this bit is set, letters in the pattern match both upper and lower
1110 case letters. It is equivalent to Perl's /i option, and it can be
1111 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
1112 always understands the concept of case for characters whose values are
1113 less than 128, so caseless matching is always possible. For characters
1114 with higher values, the concept of case is supported if PCRE is com-
1115 piled with Unicode property support, but not otherwise. If you want to
1116 use caseless matching for characters 128 and above, you must ensure
1117 that PCRE is compiled with Unicode property support as well as with
1118 UTF-8 support.
1122 If this bit is set, a dollar metacharacter in the pattern matches only
1123 at the end of the subject string. Without this option, a dollar also
1124 matches immediately before a newline at the end of the string (but not
1125 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
1126 if PCRE_MULTILINE is set. There is no equivalent to this option in
1127 Perl, and no way to set it within a pattern.
1131 If this bit is set, a dot metacharacter in the pattern matches a char-
1132 acter of any value, including one that indicates a newline. However, it
1133 only ever matches one character, even if newlines are coded as CRLF.
1134 Without this option, a dot does not match when the current position is
1135 at a newline. This option is equivalent to Perl's /s option, and it can
1136 be changed within a pattern by a (?s) option setting. A negative class
1137 such as [^a] always matches newline characters, independent of the set-
1138 ting of this option.
1142 If this bit is set, names used to identify capturing subpatterns need
1143 not be unique. This can be helpful for certain types of pattern when it
1144 is known that only one instance of the named subpattern can ever be
1145 matched. There are more details of named subpatterns below; see also
1146 the pcrepattern documentation.
1150 If this bit is set, whitespace data characters in the pattern are
1151 totally ignored except when escaped or inside a character class. White-
1152 space does not include the VT character (code 11). In addition, charac-
1153 ters between an unescaped # outside a character class and the next new-
1154 line, inclusive, are also ignored. This is equivalent to Perl's /x
1155 option, and it can be changed within a pattern by a (?x) option set-
1156 ting.
1158 Which characters are interpreted as newlines is controlled by the
1159 options passed to pcre_compile() or by a special sequence at the start
1160 of the pattern, as described in the section entitled "Newline conven-
1161 tions" in the pcrepattern documentation. Note that the end of this type
1162 of comment is a literal newline sequence in the pattern; escape
1163 sequences that happen to represent a newline do not count.
1165 This option makes it possible to include comments inside complicated
1166 patterns. Note, however, that this applies only to data characters.
1167 Whitespace characters may never appear within special character
1168 sequences in a pattern, for example within the sequence (?( that intro-
1169 duces a conditional subpattern.
1173 This option was invented in order to turn on additional functionality
1174 of PCRE that is incompatible with Perl, but it is currently of very
1175 little use. When set, any backslash in a pattern that is followed by a
1176 letter that has no special meaning causes an error, thus reserving
1177 these combinations for future expansion. By default, as in Perl, a
1178 backslash followed by a letter with no special meaning is treated as a
1179 literal. (Perl can, however, be persuaded to give an error for this, by
1180 running it with the -w option.) There are at present no other features
1181 controlled by this option. It can also be set by a (?X) option setting
1182 within a pattern.
1186 If this option is set, an unanchored pattern is required to match
1187 before or at the first newline in the subject string, though the
1188 matched text may continue over the newline.
1192 If this option is set, PCRE's behaviour is changed in some ways so that
1193 it is compatible with JavaScript rather than Perl. The changes are as
1194 follows:
1196 (1) A lone closing square bracket in a pattern causes a compile-time
1197 error, because this is illegal in JavaScript (by default it is treated
1198 as a data character). Thus, the pattern AB]CD becomes illegal when this
1199 option is set.
1201 (2) At run time, a back reference to an unset subpattern group matches
1202 an empty string (by default this causes the current matching alterna-
1203 tive to fail). A pattern such as (\1)(a) succeeds when this option is
1204 set (assuming it can find an "a" in the subject), whereas it fails by
1205 default, for Perl compatibility.
1209 By default, PCRE treats the subject string as consisting of a single
1210 line of characters (even if it actually contains newlines). The "start
1211 of line" metacharacter (^) matches only at the start of the string,
1212 while the "end of line" metacharacter ($) matches only at the end of
1213 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1214 is set). This is the same as Perl.
1216 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1217 constructs match immediately following or immediately before internal
1218 newlines in the subject string, respectively, as well as at the very
1219 start and end. This is equivalent to Perl's /m option, and it can be
1220 changed within a pattern by a (?m) option setting. If there are no new-
1221 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1222 setting PCRE_MULTILINE has no effect.
1230 These options override the default newline definition that was chosen
1231 when PCRE was built. Setting the first or the second specifies that a
1232 newline is indicated by a single character (CR or LF, respectively).
1233 Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
1234 two-character CRLF sequence. Setting PCRE_NEWLINE_ANYCRLF specifies
1235 that any of the three preceding sequences should be recognized. Setting
1236 PCRE_NEWLINE_ANY specifies that any Unicode newline sequence should be
1237 recognized. The Unicode newline sequences are the three just mentioned,
1238 plus the single characters VT (vertical tab, U+000B), FF (formfeed,
1239 U+000C), NEL (next line, U+0085), LS (line separator, U+2028), and PS
1240 (paragraph separator, U+2029). The last two are recognized only in
1241 UTF-8 mode.
1243 The newline setting in the options word uses three bits that are
1244 treated as a number, giving eight possibilities. Currently only six are
1245 used (default plus the five values above). This means that if you set
1246 more than one newline option, the combination may or may not be sensi-
1247 ble. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equivalent to
1248 PCRE_NEWLINE_CRLF, but other combinations may yield unused numbers and
1249 cause an error.
1251 The only time that a line break in a pattern is specially recognized
1252 when compiling is when PCRE_EXTENDED is set. CR and LF are whitespace
1253 characters, and so are ignored in this mode. Also, an unescaped # out-
1254 side a character class indicates a comment that lasts until after the
1255 next line break sequence. In other circumstances, line break sequences
1256 in patterns are treated as literal data.
1258 The newline option that is set at compile time becomes the default that
1259 is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1263 If this option is set, it disables the use of numbered capturing paren-
1264 theses in the pattern. Any opening parenthesis that is not followed by
1265 ? behaves as if it were followed by ?: but named parentheses can still
1266 be used for capturing (and they acquire numbers in the usual way).
1267 There is no equivalent of this option in Perl.
1271 This is an option that acts at matching time; that is, it is really an
1272 option for pcre_exec() or pcre_dfa_exec(). If it is set at compile
1273 time, it is remembered with the compiled pattern and assumed at match-
1274 ing time. For details see the discussion of PCRE_NO_START_OPTIMIZE
1275 below.
1279 This option changes the way PCRE processes \B, \b, \D, \d, \S, \s, \W,
1280 \w, and some of the POSIX character classes. By default, only ASCII
1281 characters are recognized, but if PCRE_UCP is set, Unicode properties
1282 are used instead to classify characters. More details are given in the
1283 section on generic character types in the pcrepattern page. If you set
1284 PCRE_UCP, matching one of the items it affects takes much longer. The
1285 option is available only if PCRE has been compiled with Unicode prop-
1286 erty support.
1290 This option inverts the "greediness" of the quantifiers so that they
1291 are not greedy by default, but become greedy if followed by "?". It is
1292 not compatible with Perl. It can also be set by a (?U) option setting
1293 within the pattern.
1295 PCRE_UTF8
1297 This option causes PCRE to regard both the pattern and the subject as
1298 strings of UTF-8 characters instead of single-byte character strings.
1299 However, it is available only when PCRE is built to include UTF-8 sup-
1300 port. If not, the use of this option provokes an error. Details of how
1301 this option changes the behaviour of PCRE are given in the pcreunicode
1302 page.
1306 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1307 automatically checked. There is a discussion about the validity of
1308 UTF-8 strings in the main pcre page. If an invalid UTF-8 sequence of
1309 bytes is found, pcre_compile() returns an error. If you already know
1310 that your pattern is valid, and you want to skip this check for perfor-
1311 mance reasons, you can set the PCRE_NO_UTF8_CHECK option. When it is
1312 set, the effect of passing an invalid UTF-8 string as a pattern is
1313 undefined. It may cause your program to crash. Note that this option
1314 can also be passed to pcre_exec() and pcre_dfa_exec(), to suppress the
1315 UTF-8 validity checking of subject strings.
1320 The following table lists the error codes than may be returned by
1321 pcre_compile2(), along with the error messages that may be returned by
1322 both compiling functions. As PCRE has developed, some error codes have
1323 fallen out of use. To avoid confusion, they have not been re-used.
1325 0 no error
1326 1 \ at end of pattern
1327 2 \c at end of pattern
1328 3 unrecognized character follows \
1329 4 numbers out of order in {} quantifier
1330 5 number too big in {} quantifier
1331 6 missing terminating ] for character class
1332 7 invalid escape sequence in character class
1333 8 range out of order in character class
1334 9 nothing to repeat
1335 10 [this code is not in use]
1336 11 internal error: unexpected repeat
1337 12 unrecognized character after (? or (?-
1338 13 POSIX named classes are supported only within a class
1339 14 missing )
1340 15 reference to non-existent subpattern
1341 16 erroffset passed as NULL
1342 17 unknown option bit(s) set
1343 18 missing ) after comment
1344 19 [this code is not in use]
1345 20 regular expression is too large
1346 21 failed to get memory
1347 22 unmatched parentheses
1348 23 internal error: code overflow
1349 24 unrecognized character after (?<
1350 25 lookbehind assertion is not fixed length
1351 26 malformed number or name after (?(
1352 27 conditional group contains more than two branches
1353 28 assertion expected after (?(
1354 29 (?R or (?[+-]digits must be followed by )
1355 30 unknown POSIX class name
1356 31 POSIX collating elements are not supported
1357 32 this version of PCRE is not compiled with PCRE_UTF8 support
1358 33 [this code is not in use]
1359 34 character value in \x{...} sequence is too large
1360 35 invalid condition (?(0)
1361 36 \C not allowed in lookbehind assertion
1362 37 PCRE does not support \L, \l, \N{name}, \U, or \u
1363 38 number after (?C is > 255
1364 39 closing ) for (?C expected
1365 40 recursive call could loop indefinitely
1366 41 unrecognized character after (?P
1367 42 syntax error in subpattern name (missing terminator)
1368 43 two named subpatterns have the same name
1369 44 invalid UTF-8 string
1370 45 support for \P, \p, and \X has not been compiled
1371 46 malformed \P or \p sequence
1372 47 unknown property name after \P or \p
1373 48 subpattern name is too long (maximum 32 characters)
1374 49 too many named subpatterns (maximum 10000)
1375 50 [this code is not in use]
1376 51 octal value is greater than \377 (not in UTF-8 mode)
1377 52 internal error: overran compiling workspace
1378 53 internal error: previously-checked referenced subpattern
1379 not found
1380 54 DEFINE group contains more than one branch
1381 55 repeating a DEFINE group is not allowed
1382 56 inconsistent NEWLINE options
1383 57 \g is not followed by a braced, angle-bracketed, or quoted
1384 name/number or by a plain number
1385 58 a numbered reference must not be zero
1386 59 an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)
1387 60 (*VERB) not recognized
1388 61 number is too big
1389 62 subpattern name expected
1390 63 digit expected after (?+
1391 64 ] is an invalid data character in JavaScript compatibility mode
1392 65 different names for subpatterns of the same number are
1393 not allowed
1394 66 (*MARK) must have an argument
1395 67 this version of PCRE is not compiled with PCRE_UCP support
1396 68 \c must be followed by an ASCII character
1397 69 \k is not followed by a braced, angle-bracketed, or quoted name
1399 The numbers 32 and 10000 in errors 48 and 49 are defaults; different
1400 values may be used if the limits were changed when PCRE was built.
1405 pcre_extra *pcre_study(const pcre *code, int options
1406 const char **errptr);
1408 If a compiled pattern is going to be used several times, it is worth
1409 spending more time analyzing it in order to speed up the time taken for
1410 matching. The function pcre_study() takes a pointer to a compiled pat-
1411 tern as its first argument. If studying the pattern produces additional
1412 information that will help speed up matching, pcre_study() returns a
1413 pointer to a pcre_extra block, in which the study_data field points to
1414 the results of the study.
1416 The returned value from pcre_study() can be passed directly to
1417 pcre_exec() or pcre_dfa_exec(). However, a pcre_extra block also con-
1418 tains other fields that can be set by the caller before the block is
1419 passed; these are described below in the section on matching a pattern.
1421 If studying the pattern does not produce any useful information,
1422 pcre_study() returns NULL. In that circumstance, if the calling program
1423 wants to pass any of the other fields to pcre_exec() or
1424 pcre_dfa_exec(), it must set up its own pcre_extra block.
1426 The second argument of pcre_study() contains option bits. At present,
1427 no options are defined, and this argument should always be zero.
1429 The third argument for pcre_study() is a pointer for an error message.
1430 If studying succeeds (even if no data is returned), the variable it
1431 points to is set to NULL. Otherwise it is set to point to a textual
1432 error message. This is a static string that is part of the library. You
1433 must not try to free it. You should test the error pointer for NULL
1434 after calling pcre_study(), to be sure that it has run successfully.
1436 This is a typical call to pcre_study():
1438 pcre_extra *pe;
1439 pe = pcre_study(
1440 re, /* result of pcre_compile() */
1441 0, /* no options exist */
1442 &error); /* set to NULL or points to a message */
1444 Studying a pattern does two things: first, a lower bound for the length
1445 of subject string that is needed to match the pattern is computed. This
1446 does not mean that there are any strings of that length that match, but
1447 it does guarantee that no shorter strings match. The value is used by
1448 pcre_exec() and pcre_dfa_exec() to avoid wasting time by trying to
1449 match strings that are shorter than the lower bound. You can find out
1450 the value in a calling program via the pcre_fullinfo() function.
1452 Studying a pattern is also useful for non-anchored patterns that do not
1453 have a single fixed starting character. A bitmap of possible starting
1454 bytes is created. This speeds up finding a position in the subject at
1455 which to start matching.
1457 The two optimizations just described can be disabled by setting the
1458 PCRE_NO_START_OPTIMIZE option when calling pcre_exec() or
1459 pcre_dfa_exec(). You might want to do this if your pattern contains
1460 callouts or (*MARK), and you want to make use of these facilities in
1461 cases where matching fails. See the discussion of PCRE_NO_START_OPTI-
1462 MIZE below.
1467 PCRE handles caseless matching, and determines whether characters are
1468 letters, digits, or whatever, by reference to a set of tables, indexed
1469 by character value. When running in UTF-8 mode, this applies only to
1470 characters with codes less than 128. By default, higher-valued codes
1471 never match escapes such as \w or \d, but they can be tested with \p if
1472 PCRE is built with Unicode character property support. Alternatively,
1473 the PCRE_UCP option can be set at compile time; this causes \w and
1474 friends to use Unicode property support instead of built-in tables. The
1475 use of locales with Unicode is discouraged. If you are handling charac-
1476 ters with codes greater than 128, you should either use UTF-8 and Uni-
1477 code, or use locales, but not try to mix the two.
1479 PCRE contains an internal set of tables that are used when the final
1480 argument of pcre_compile() is NULL. These are sufficient for many
1481 applications. Normally, the internal tables recognize only ASCII char-
1482 acters. However, when PCRE is built, it is possible to cause the inter-
1483 nal tables to be rebuilt in the default "C" locale of the local system,
1484 which may cause them to be different.
1486 The internal tables can always be overridden by tables supplied by the
1487 application that calls PCRE. These may be created in a different locale
1488 from the default. As more and more applications change to using Uni-
1489 code, the need for this locale support is expected to die away.
1491 External tables are built by calling the pcre_maketables() function,
1492 which has no arguments, in the relevant locale. The result can then be
1493 passed to pcre_compile() or pcre_exec() as often as necessary. For
1494 example, to build and use tables that are appropriate for the French
1495 locale (where accented characters with values greater than 128 are
1496 treated as letters), the following code could be used:
1498 setlocale(LC_CTYPE, "fr_FR");
1499 tables = pcre_maketables();
1500 re = pcre_compile(..., tables);
1502 The locale name "fr_FR" is used on Linux and other Unix-like systems;
1503 if you are using Windows, the name for the French locale is "french".
1505 When pcre_maketables() runs, the tables are built in memory that is
1506 obtained via pcre_malloc. It is the caller's responsibility to ensure
1507 that the memory containing the tables remains available for as long as
1508 it is needed.
1510 The pointer that is passed to pcre_compile() is saved with the compiled
1511 pattern, and the same tables are used via this pointer by pcre_study()
1512 and normally also by pcre_exec(). Thus, by default, for any single pat-
1513 tern, compilation, studying and matching all happen in the same locale,
1514 but different patterns can be compiled in different locales.
1516 It is possible to pass a table pointer or NULL (indicating the use of
1517 the internal tables) to pcre_exec(). Although not intended for this
1518 purpose, this facility could be used to match a pattern in a different
1519 locale from the one in which it was compiled. Passing table pointers at
1520 run time is discussed below in the section on matching a pattern.
1525 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1526 int what, void *where);
1528 The pcre_fullinfo() function returns information about a compiled pat-
1529 tern. It replaces the obsolete pcre_info() function, which is neverthe-
1530 less retained for backwards compability (and is documented below).
1532 The first argument for pcre_fullinfo() is a pointer to the compiled
1533 pattern. The second argument is the result of pcre_study(), or NULL if
1534 the pattern was not studied. The third argument specifies which piece
1535 of information is required, and the fourth argument is a pointer to a
1536 variable to receive the data. The yield of the function is zero for
1537 success, or one of the following negative numbers:
1539 PCRE_ERROR_NULL the argument code was NULL
1540 the argument where was NULL
1541 PCRE_ERROR_BADMAGIC the "magic number" was not found
1542 PCRE_ERROR_BADOPTION the value of what was invalid
1544 The "magic number" is placed at the start of each compiled pattern as
1545 an simple check against passing an arbitrary memory pointer. Here is a
1546 typical call of pcre_fullinfo(), to obtain the length of the compiled
1547 pattern:
1549 int rc;
1550 size_t length;
1551 rc = pcre_fullinfo(
1552 re, /* result of pcre_compile() */
1553 pe, /* result of pcre_study(), or NULL */
1554 PCRE_INFO_SIZE, /* what is required */
1555 &length); /* where to put the data */
1557 The possible values for the third argument are defined in pcre.h, and
1558 are as follows:
1562 Return the number of the highest back reference in the pattern. The
1563 fourth argument should point to an int variable. Zero is returned if
1564 there are no back references.
1568 Return the number of capturing subpatterns in the pattern. The fourth
1569 argument should point to an int variable.
1573 Return a pointer to the internal default character tables within PCRE.
1574 The fourth argument should point to an unsigned char * variable. This
1575 information call is provided for internal use by the pcre_study() func-
1576 tion. External callers can cause PCRE to use its internal tables by
1577 passing a NULL table pointer.
1581 Return information about the first byte of any matched string, for a
1582 non-anchored pattern. The fourth argument should point to an int vari-
1583 able. (This option used to be called PCRE_INFO_FIRSTCHAR; the old name
1584 is still recognized for backwards compatibility.)
1586 If there is a fixed first byte, for example, from a pattern such as
1587 (cat|cow|coyote), its value is returned. Otherwise, if either
1589 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
1590 branch starts with "^", or
1592 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
1593 set (if it were set, the pattern would be anchored),
1595 -1 is returned, indicating that the pattern matches only at the start
1596 of a subject string or after any newline within the string. Otherwise
1597 -2 is returned. For anchored patterns, -2 is returned.
1601 If the pattern was studied, and this resulted in the construction of a
1602 256-bit table indicating a fixed set of bytes for the first byte in any
1603 matching string, a pointer to the table is returned. Otherwise NULL is
1604 returned. The fourth argument should point to an unsigned char * vari-
1605 able.
1609 Return 1 if the pattern contains any explicit matches for CR or LF
1610 characters, otherwise 0. The fourth argument should point to an int
1611 variable. An explicit match is either a literal CR or LF character, or
1612 \r or \n.
1616 Return 1 if the (?J) or (?-J) option setting is used in the pattern,
1617 otherwise 0. The fourth argument should point to an int variable. (?J)
1618 and (?-J) set and unset the local PCRE_DUPNAMES option, respectively.
1622 Return the value of the rightmost literal byte that must exist in any
1623 matched string, other than at its start, if such a byte has been
1624 recorded. The fourth argument should point to an int variable. If there
1625 is no such byte, -1 is returned. For anchored patterns, a last literal
1626 byte is recorded only if it follows something of variable length. For
1627 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
1628 /^a\dz\d/ the returned value is -1.
1632 If the pattern was studied and a minimum length for matching subject
1633 strings was computed, its value is returned. Otherwise the returned
1634 value is -1. The value is a number of characters, not bytes (this may
1635 be relevant in UTF-8 mode). The fourth argument should point to an int
1636 variable. A non-negative value is a lower bound to the length of any
1637 matching string. There may not be any strings of that length that do
1638 actually match, but every string that does match is at least that long.
1644 PCRE supports the use of named as well as numbered capturing parenthe-
1645 ses. The names are just an additional way of identifying the parenthe-
1646 ses, which still acquire numbers. Several convenience functions such as
1647 pcre_get_named_substring() are provided for extracting captured sub-
1648 strings by name. It is also possible to extract the data directly, by
1649 first converting the name to a number in order to access the correct
1650 pointers in the output vector (described with pcre_exec() below). To do
1651 the conversion, you need to use the name-to-number map, which is
1652 described by these three values.
1654 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
1655 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
1656 of each entry; both of these return an int value. The entry size
1657 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
1658 a pointer to the first entry of the table (a pointer to char). The
1659 first two bytes of each entry are the number of the capturing parenthe-
1660 sis, most significant byte first. The rest of the entry is the corre-
1661 sponding name, zero terminated.
1663 The names are in alphabetical order. Duplicate names may appear if (?|
1664 is used to create multiple groups with the same number, as described in
1665 the section on duplicate subpattern numbers in the pcrepattern page.
1666 Duplicate names for subpatterns with different numbers are permitted
1667 only if PCRE_DUPNAMES is set. In all cases of duplicate names, they
1668 appear in the table in the order in which they were found in the pat-
1669 tern. In the absence of (?| this is the order of increasing number;
1670 when (?| is used this is not necessarily the case because later subpat-
1671 terns may have lower numbers.
1673 As a simple example of the name/number table, consider the following
1674 pattern (assume PCRE_EXTENDED is set, so white space - including new-
1675 lines - is ignored):
1677 (?<date> (?<year>(\d\d)?\d\d) -
1678 (?<month>\d\d) - (?<day>\d\d) )
1680 There are four named subpatterns, so the table has four entries, and
1681 each entry in the table is eight bytes long. The table is as follows,
1682 with non-printing bytes shows in hexadecimal, and undefined bytes shown
1683 as ??:
1685 00 01 d a t e 00 ??
1686 00 05 d a y 00 ?? ??
1687 00 04 m o n t h 00
1688 00 02 y e a r 00 ??
1690 When writing code to extract data from named subpatterns using the
1691 name-to-number map, remember that the length of the entries is likely
1692 to be different for each compiled pattern.
1696 Return 1 if the pattern can be used for partial matching with
1697 pcre_exec(), otherwise 0. The fourth argument should point to an int
1698 variable. From release 8.00, this always returns 1, because the
1699 restrictions that previously applied to partial matching have been
1700 lifted. The pcrepartial documentation gives details of partial match-
1701 ing.
1705 Return a copy of the options with which the pattern was compiled. The
1706 fourth argument should point to an unsigned long int variable. These
1707 option bits are those specified in the call to pcre_compile(), modified
1708 by any top-level option settings at the start of the pattern itself. In
1709 other words, they are the options that will be in force when matching
1710 starts. For example, if the pattern /(?im)abc(?-i)d/ is compiled with
1711 the PCRE_EXTENDED option, the result is PCRE_CASELESS, PCRE_MULTILINE,
1714 A pattern is automatically anchored by PCRE if all of its top-level
1715 alternatives begin with one of the following:
1717 ^ unless PCRE_MULTILINE is set
1718 \A always
1719 \G always
1720 .* if PCRE_DOTALL is set and there are no back
1721 references to the subpattern in which .* appears
1723 For such patterns, the PCRE_ANCHORED bit is set in the options returned
1724 by pcre_fullinfo().
1728 Return the size of the compiled pattern, that is, the value that was
1729 passed as the argument to pcre_malloc() when PCRE was getting memory in
1730 which to place the compiled data. The fourth argument should point to a
1731 size_t variable.
1735 Return the size of the data block pointed to by the study_data field in
1736 a pcre_extra block. If pcre_extra is NULL, or there is no study data,
1737 zero is returned. The fourth argument should point to a size_t vari-
1738 able. The study_data field is set by pcre_study() to record informa-
1739 tion that will speed up matching (see the section entitled "Studying a
1740 pattern" above). The format of the study_data block is private, but its
1741 length is made available via this option so that it can be saved and
1742 restored (see the pcreprecompile documentation for details).
1747 int pcre_info(const pcre *code, int *optptr, int *firstcharptr);
1749 The pcre_info() function is now obsolete because its interface is too
1750 restrictive to return all the available data about a compiled pattern.
1751 New programs should use pcre_fullinfo() instead. The yield of
1752 pcre_info() is the number of capturing subpatterns, or one of the fol-
1753 lowing negative numbers:
1755 PCRE_ERROR_NULL the argument code was NULL
1756 PCRE_ERROR_BADMAGIC the "magic number" was not found
1758 If the optptr argument is not NULL, a copy of the options with which
1759 the pattern was compiled is placed in the integer it points to (see
1760 PCRE_INFO_OPTIONS above).
1762 If the pattern is not anchored and the firstcharptr argument is not
1763 NULL, it is used to pass back information about the first character of
1764 any matched string (see PCRE_INFO_FIRSTBYTE above).
1769 int pcre_refcount(pcre *code, int adjust);
1771 The pcre_refcount() function is used to maintain a reference count in
1772 the data block that contains a compiled pattern. It is provided for the
1773 benefit of applications that operate in an object-oriented manner,
1774 where different parts of the application may be using the same compiled
1775 pattern, but you want to free the block when they are all done.
1777 When a pattern is compiled, the reference count field is initialized to
1778 zero. It is changed only by calling this function, whose action is to
1779 add the adjust value (which may be positive or negative) to it. The
1780 yield of the function is the new value. However, the value of the count
1781 is constrained to lie between 0 and 65535, inclusive. If the new value
1782 is outside these limits, it is forced to the appropriate limit value.
1784 Except when it is zero, the reference count is not correctly preserved
1785 if a pattern is compiled on one host and then transferred to a host
1786 whose byte-order is different. (This seems a highly unlikely scenario.)
1791 int pcre_exec(const pcre *code, const pcre_extra *extra,
1792 const char *subject, int length, int startoffset,
1793 int options, int *ovector, int ovecsize);
1795 The function pcre_exec() is called to match a subject string against a
1796 compiled pattern, which is passed in the code argument. If the pattern
1797 was studied, the result of the study should be passed in the extra
1798 argument. This function is the main matching facility of the library,
1799 and it operates in a Perl-like manner. For specialist use there is also
1800 an alternative matching function, which is described below in the sec-
1801 tion about the pcre_dfa_exec() function.
1803 In most applications, the pattern will have been compiled (and option-
1804 ally studied) in the same process that calls pcre_exec(). However, it
1805 is possible to save compiled patterns and study data, and then use them
1806 later in different processes, possibly even on different hosts. For a
1807 discussion about this, see the pcreprecompile documentation.
1809 Here is an example of a simple call to pcre_exec():
1811 int rc;
1812 int ovector[30];
1813 rc = pcre_exec(
1814 re, /* result of pcre_compile() */
1815 NULL, /* we didn't study the pattern */
1816 "some string", /* the subject string */
1817 11, /* the length of the subject string */
1818 0, /* start at offset 0 in the subject */
1819 0, /* default options */
1820 ovector, /* vector of integers for substring information */
1821 30); /* number of elements (NOT size in bytes) */
1823 Extra data for pcre_exec()
1825 If the extra argument is not NULL, it must point to a pcre_extra data
1826 block. The pcre_study() function returns such a block (when it doesn't
1827 return NULL), but you can also create one for yourself, and pass addi-
1828 tional information in it. The pcre_extra block contains the following
1829 fields (not necessarily in this order):
1831 unsigned long int flags;
1832 void *study_data;
1833 unsigned long int match_limit;
1834 unsigned long int match_limit_recursion;
1835 void *callout_data;
1836 const unsigned char *tables;
1837 unsigned char **mark;
1839 The flags field is a bitmap that specifies which of the other fields
1840 are set. The flag bits are:
1849 Other flag bits should be set to zero. The study_data field is set in
1850 the pcre_extra block that is returned by pcre_study(), together with
1851 the appropriate flag bit. You should not set this yourself, but you may
1852 add to the block by setting the other fields and their corresponding
1853 flag bits.
1855 The match_limit field provides a means of preventing PCRE from using up
1856 a vast amount of resources when running patterns that are not going to
1857 match, but which have a very large number of possibilities in their
1858 search trees. The classic example is a pattern that uses nested unlim-
1859 ited repeats.
1861 Internally, PCRE uses a function called match() which it calls repeat-
1862 edly (sometimes recursively). The limit set by match_limit is imposed
1863 on the number of times this function is called during a match, which
1864 has the effect of limiting the amount of backtracking that can take
1865 place. For patterns that are not anchored, the count restarts from zero
1866 for each position in the subject string.
1868 The default value for the limit can be set when PCRE is built; the
1869 default default is 10 million, which handles all but the most extreme
1870 cases. You can override the default by suppling pcre_exec() with a
1871 pcre_extra block in which match_limit is set, and
1872 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
1873 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
1875 The match_limit_recursion field is similar to match_limit, but instead
1876 of limiting the total number of times that match() is called, it limits
1877 the depth of recursion. The recursion depth is a smaller number than
1878 the total number of calls, because not all calls to match() are recur-
1879 sive. This limit is of use only if it is set smaller than match_limit.
1881 Limiting the recursion depth limits the amount of stack that can be
1882 used, or, when PCRE has been compiled to use memory on the heap instead
1883 of the stack, the amount of heap memory that can be used.
1885 The default value for match_limit_recursion can be set when PCRE is
1886 built; the default default is the same value as the default for
1887 match_limit. You can override the default by suppling pcre_exec() with
1888 a pcre_extra block in which match_limit_recursion is set, and
1889 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
1890 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
1892 The callout_data field is used in conjunction with the "callout" fea-
1893 ture, and is described in the pcrecallout documentation.
1895 The tables field is used to pass a character tables pointer to
1896 pcre_exec(); this overrides the value that is stored with the compiled
1897 pattern. A non-NULL value is stored with the compiled pattern only if
1898 custom tables were supplied to pcre_compile() via its tableptr argu-
1899 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
1900 PCRE's internal tables to be used. This facility is helpful when re-
1901 using patterns that have been saved after compiling with an external
1902 set of tables, because the external tables might be at a different
1903 address when pcre_exec() is called. See the pcreprecompile documenta-
1904 tion for a discussion of saving compiled patterns for later use.
1906 If PCRE_EXTRA_MARK is set in the flags field, the mark field must be
1907 set to point to a char * variable. If the pattern contains any back-
1908 tracking control verbs such as (*MARK:NAME), and the execution ends up
1909 with a name to pass back, a pointer to the name string (zero termi-
1910 nated) is placed in the variable pointed to by the mark field. The
1911 names are within the compiled pattern; if you wish to retain such a
1912 name you must copy it before freeing the memory of a compiled pattern.
1913 If there is no name to pass back, the variable pointed to by the mark
1914 field set to NULL. For details of the backtracking control verbs, see
1915 the section entitled "Backtracking control" in the pcrepattern documen-
1916 tation.
1918 Option bits for pcre_exec()
1920 The unused bits of the options argument for pcre_exec() must be zero.
1921 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
1928 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
1929 matching position. If a pattern was compiled with PCRE_ANCHORED, or
1930 turned out to be anchored by virtue of its contents, it cannot be made
1931 unachored at matching time.
1936 These options (which are mutually exclusive) control what the \R escape
1937 sequence matches. The choice is either to match only CR, LF, or CRLF,
1938 or to match any Unicode newline sequence. These options override the
1939 choice that was made or defaulted when the pattern was compiled.
1947 These options override the newline definition that was chosen or
1948 defaulted when the pattern was compiled. For details, see the descrip-
1949 tion of pcre_compile() above. During matching, the newline choice
1950 affects the behaviour of the dot, circumflex, and dollar metacharac-
1951 ters. It may also alter the way the match position is advanced after a
1952 match failure for an unanchored pattern.
1955 set, and a match attempt for an unanchored pattern fails when the cur-
1956 rent position is at a CRLF sequence, and the pattern contains no
1957 explicit matches for CR or LF characters, the match position is
1958 advanced by two characters instead of one, in other words, to after the
1959 CRLF.
1961 The above rule is a compromise that makes the most common cases work as
1962 expected. For example, if the pattern is .+A (and the PCRE_DOTALL
1963 option is not set), it does not match the string "\r\nA" because, after
1964 failing at the start, it skips both the CR and the LF before retrying.
1965 However, the pattern [\r\n]A does match that string, because it con-
1966 tains an explicit CR or LF reference, and so advances only by one char-
1967 acter after the first failure.
1969 An explicit match for CR of LF is either a literal appearance of one of
1970 those characters, or one of the \r or \n escape sequences. Implicit
1971 matches such as [^X] do not count, nor does \s (which includes CR and
1972 LF in the characters that it matches).
1974 Notwithstanding the above, anomalous effects may still occur when CRLF
1975 is a valid newline sequence and explicit \r or \n escapes appear in the
1976 pattern.
1980 This option specifies that first character of the subject string is not
1981 the beginning of a line, so the circumflex metacharacter should not
1982 match before it. Setting this without PCRE_MULTILINE (at compile time)
1983 causes circumflex never to match. This option affects only the behav-
1984 iour of the circumflex metacharacter. It does not affect \A.
1988 This option specifies that the end of the subject string is not the end
1989 of a line, so the dollar metacharacter should not match it nor (except
1990 in multiline mode) a newline immediately before it. Setting this with-
1991 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
1992 option affects only the behaviour of the dollar metacharacter. It does
1993 not affect \Z or \z.
1997 An empty string is not considered to be a valid match if this option is
1998 set. If there are alternatives in the pattern, they are tried. If all
1999 the alternatives match the empty string, the entire match fails. For
2000 example, if the pattern
2002 a?b?
2004 is applied to a string not beginning with "a" or "b", it matches an
2005 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
2006 match is not valid, so PCRE searches further into the string for occur-
2007 rences of "a" or "b".
2011 This is like PCRE_NOTEMPTY, except that an empty string match that is
2012 not at the start of the subject is permitted. If the pattern is
2013 anchored, such a match can occur only if the pattern contains \K.
2015 Perl has no direct equivalent of PCRE_NOTEMPTY or
2016 PCRE_NOTEMPTY_ATSTART, but it does make a special case of a pattern
2017 match of the empty string within its split() function, and when using
2018 the /g modifier. It is possible to emulate Perl's behaviour after
2019 matching a null string by first trying the match again at the same off-
2020 set with PCRE_NOTEMPTY_ATSTART and PCRE_ANCHORED, and then if that
2021 fails, by advancing the starting offset (see below) and trying an ordi-
2022 nary match again. There is some code that demonstrates how to do this
2023 in the pcredemo sample program. In the most general case, you have to
2024 check to see if the newline convention recognizes CRLF as a newline,
2025 and if so, and the current character is CR followed by LF, advance the
2026 starting offset by two characters instead of one.
2030 There are a number of optimizations that pcre_exec() uses at the start
2031 of a match, in order to speed up the process. For example, if it is
2032 known that an unanchored match must start with a specific character, it
2033 searches the subject for that character, and fails immediately if it
2034 cannot find it, without actually running the main matching function.
2035 This means that a special item such as (*COMMIT) at the start of a pat-
2036 tern is not considered until after a suitable starting point for the
2037 match has been found. When callouts or (*MARK) items are in use, these
2038 "start-up" optimizations can cause them to be skipped if the pattern is
2039 never actually used. The start-up optimizations are in effect a pre-
2040 scan of the subject that takes place before the pattern is run.
2042 The PCRE_NO_START_OPTIMIZE option disables the start-up optimizations,
2043 possibly causing performance to suffer, but ensuring that in cases
2044 where the result is "no match", the callouts do occur, and that items
2045 such as (*COMMIT) and (*MARK) are considered at every possible starting
2046 position in the subject string. If PCRE_NO_START_OPTIMIZE is set at
2047 compile time, it cannot be unset at matching time.
2049 Setting PCRE_NO_START_OPTIMIZE can change the outcome of a matching
2050 operation. Consider the pattern
2054 When this is compiled, PCRE records the fact that a match must start
2055 with the character "A". Suppose the subject string is "DEFABC". The
2056 start-up optimization scans along the subject, finds "A" and runs the
2057 first match attempt from there. The (*COMMIT) item means that the pat-
2058 tern must match the current starting position, which in this case, it
2059 does. However, if the same match is run with PCRE_NO_START_OPTIMIZE
2060 set, the initial scan along the subject string does not happen. The
2061 first match attempt is run starting from "D" and when this fails,
2062 (*COMMIT) prevents any further matches being tried, so the overall
2063 result is "no match". If the pattern is studied, more start-up opti-
2064 mizations may be used. For example, a minimum length for the subject
2065 may be recorded. Consider the pattern
2067 (*MARK:A)(X|Y)
2069 The minimum length for a match is one character. If the subject is
2070 "ABC", there will be attempts to match "ABC", "BC", "C", and then
2071 finally an empty string. If the pattern is studied, the final attempt
2072 does not take place, because PCRE knows that the subject is too short,
2073 and so the (*MARK) is never encountered. In this case, studying the
2074 pattern does not affect the overall match result, which is still "no
2075 match", but it does affect the auxiliary information that is returned.
2079 When PCRE_UTF8 is set at compile time, the validity of the subject as a
2080 UTF-8 string is automatically checked when pcre_exec() is subsequently
2081 called. The value of startoffset is also checked to ensure that it
2082 points to the start of a UTF-8 character. There is a discussion about
2083 the validity of UTF-8 strings in the section on UTF-8 support in the
2084 main pcre page. If an invalid UTF-8 sequence of bytes is found,
2085 pcre_exec() returns the error PCRE_ERROR_BADUTF8 or, if PCRE_PAR-
2086 TIAL_HARD is set and the problem is a truncated UTF-8 character at the
2087 end of the subject, PCRE_ERROR_SHORTUTF8. In both cases, information
2088 about the precise nature of the error may also be returned (see the
2089 descriptions of these errors in the section entitled Error return val-
2090 ues from pcre_exec() below). If startoffset contains a value that does
2091 not point to the start of a UTF-8 character (or to the end of the sub-
2092 ject), PCRE_ERROR_BADUTF8_OFFSET is returned.
2094 If you already know that your subject is valid, and you want to skip
2095 these checks for performance reasons, you can set the
2096 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
2097 do this for the second and subsequent calls to pcre_exec() if you are
2098 making repeated calls to find all the matches in a single subject
2099 string. However, you should be sure that the value of startoffset
2100 points to the start of a UTF-8 character (or the end of the subject).
2101 When PCRE_NO_UTF8_CHECK is set, the effect of passing an invalid UTF-8
2102 string as a subject or an invalid value of startoffset is undefined.
2103 Your program may crash.
2108 These options turn on the partial matching feature. For backwards com-
2109 patibility, PCRE_PARTIAL is a synonym for PCRE_PARTIAL_SOFT. A partial
2110 match occurs if the end of the subject string is reached successfully,
2111 but there are not enough subject characters to complete the match. If
2112 this happens when PCRE_PARTIAL_SOFT (but not PCRE_PARTIAL_HARD) is set,
2113 matching continues by testing any remaining alternatives. Only if no
2114 complete match can be found is PCRE_ERROR_PARTIAL returned instead of
2115 PCRE_ERROR_NOMATCH. In other words, PCRE_PARTIAL_SOFT says that the
2116 caller is prepared to handle a partial match, but only if no complete
2117 match can be found.
2119 If PCRE_PARTIAL_HARD is set, it overrides PCRE_PARTIAL_SOFT. In this
2120 case, if a partial match is found, pcre_exec() immediately returns
2121 PCRE_ERROR_PARTIAL, without considering any other alternatives. In
2122 other words, when PCRE_PARTIAL_HARD is set, a partial match is consid-
2123 ered to be more important that an alternative complete match.
2125 In both cases, the portion of the string that was inspected when the
2126 partial match was found is set as the first matching string. There is a
2127 more detailed discussion of partial and multi-segment matching, with
2128 examples, in the pcrepartial documentation.
2130 The string to be matched by pcre_exec()
2132 The subject string is passed to pcre_exec() as a pointer in subject, a
2133 length (in bytes) in length, and a starting byte offset in startoffset.
2134 If this is negative or greater than the length of the subject,
2135 pcre_exec() returns PCRE_ERROR_BADOFFSET. When the starting offset is
2136 zero, the search for a match starts at the beginning of the subject,
2137 and this is by far the most common case. In UTF-8 mode, the byte offset
2138 must point to the start of a UTF-8 character (or the end of the sub-
2139 ject). Unlike the pattern string, the subject may contain binary zero
2140 bytes.
2142 A non-zero starting offset is useful when searching for another match
2143 in the same subject by calling pcre_exec() again after a previous suc-
2144 cess. Setting startoffset differs from just passing over a shortened
2145 string and setting PCRE_NOTBOL in the case of a pattern that begins
2146 with any kind of lookbehind. For example, consider the pattern
2148 \Biss\B
2150 which finds occurrences of "iss" in the middle of words. (\B matches
2151 only if the current position in the subject is not a word boundary.)
2152 When applied to the string "Mississipi" the first call to pcre_exec()
2153 finds the first occurrence. If pcre_exec() is called again with just
2154 the remainder of the subject, namely "issipi", it does not match,
2155 because \B is always false at the start of the subject, which is deemed
2156 to be a word boundary. However, if pcre_exec() is passed the entire
2157 string again, but with startoffset set to 4, it finds the second occur-
2158 rence of "iss" because it is able to look behind the starting point to
2159 discover that it is preceded by a letter.
2161 Finding all the matches in a subject is tricky when the pattern can
2162 match an empty string. It is possible to emulate Perl's /g behaviour by
2163 first trying the match again at the same offset, with the
2164 PCRE_NOTEMPTY_ATSTART and PCRE_ANCHORED options, and then if that
2165 fails, advancing the starting offset and trying an ordinary match
2166 again. There is some code that demonstrates how to do this in the pcre-
2167 demo sample program. In the most general case, you have to check to see
2168 if the newline convention recognizes CRLF as a newline, and if so, and
2169 the current character is CR followed by LF, advance the starting offset
2170 by two characters instead of one.
2172 If a non-zero starting offset is passed when the pattern is anchored,
2173 one attempt to match at the given offset is made. This can only succeed
2174 if the pattern does not require the match to be at the start of the
2175 subject.
2177 How pcre_exec() returns captured substrings
2179 In general, a pattern matches a certain portion of the subject, and in
2180 addition, further substrings from the subject may be picked out by
2181 parts of the pattern. Following the usage in Jeffrey Friedl's book,
2182 this is called "capturing" in what follows, and the phrase "capturing
2183 subpattern" is used for a fragment of a pattern that picks out a sub-
2184 string. PCRE supports several other kinds of parenthesized subpattern
2185 that do not cause substrings to be captured.
2187 Captured substrings are returned to the caller via a vector of integers
2188 whose address is passed in ovector. The number of elements in the vec-
2189 tor is passed in ovecsize, which must be a non-negative number. Note:
2190 this argument is NOT the size of ovector in bytes.
2192 The first two-thirds of the vector is used to pass back captured sub-
2193 strings, each substring using a pair of integers. The remaining third
2194 of the vector is used as workspace by pcre_exec() while matching cap-
2195 turing subpatterns, and is not available for passing back information.
2196 The number passed in ovecsize should always be a multiple of three. If
2197 it is not, it is rounded down.
2199 When a match is successful, information about captured substrings is
2200 returned in pairs of integers, starting at the beginning of ovector,
2201 and continuing up to two-thirds of its length at the most. The first
2202 element of each pair is set to the byte offset of the first character
2203 in a substring, and the second is set to the byte offset of the first
2204 character after the end of a substring. Note: these values are always
2205 byte offsets, even in UTF-8 mode. They are not character counts.
2207 The first pair of integers, ovector[0] and ovector[1], identify the
2208 portion of the subject string matched by the entire pattern. The next
2209 pair is used for the first capturing subpattern, and so on. The value
2210 returned by pcre_exec() is one more than the highest numbered pair that
2211 has been set. For example, if two substrings have been captured, the
2212 returned value is 3. If there are no capturing subpatterns, the return
2213 value from a successful match is 1, indicating that just the first pair
2214 of offsets has been set.
2216 If a capturing subpattern is matched repeatedly, it is the last portion
2217 of the string that it matched that is returned.
2219 If the vector is too small to hold all the captured substring offsets,
2220 it is used as far as possible (up to two-thirds of its length), and the
2221 function returns a value of zero. If the substring offsets are not of
2222 interest, pcre_exec() may be called with ovector passed as NULL and
2223 ovecsize as zero. However, if the pattern contains back references and
2224 the ovector is not big enough to remember the related substrings, PCRE
2225 has to get additional memory for use during matching. Thus it is usu-
2226 ally advisable to supply an ovector.
2228 The pcre_fullinfo() function can be used to find out how many capturing
2229 subpatterns there are in a compiled pattern. The smallest size for
2230 ovector that will allow for n captured substrings, in addition to the
2231 offsets of the substring matched by the whole pattern, is (n+1)*3.
2233 It is possible for capturing subpattern number n+1 to match some part
2234 of the subject when subpattern n has not been used at all. For example,
2235 if the string "abc" is matched against the pattern (a|(z))(bc) the
2236 return from the function is 4, and subpatterns 1 and 3 are matched, but
2237 2 is not. When this happens, both values in the offset pairs corre-
2238 sponding to unused subpatterns are set to -1.
2240 Offset values that correspond to unused subpatterns at the end of the
2241 expression are also set to -1. For example, if the string "abc" is
2242 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
2243 matched. The return from the function is 2, because the highest used
2244 capturing subpattern number is 1, and the offsets for for the second
2245 and third capturing subpatterns (assuming the vector is large enough,
2246 of course) are set to -1.
2248 Note: Elements of ovector that do not correspond to capturing parenthe-
2249 ses in the pattern are never changed. That is, if a pattern contains n
2250 capturing parentheses, no more than ovector[0] to ovector[2n+1] are set
2251 by pcre_exec(). The other elements retain whatever values they previ-
2252 ously had.
2254 Some convenience functions are provided for extracting the captured
2255 substrings as separate strings. These are described below.
2257 Error return values from pcre_exec()
2259 If pcre_exec() fails, it returns a negative number. The following are
2260 defined in the header file:
2264 The subject string did not match the pattern.
2268 Either code or subject was passed as NULL, or ovector was NULL and
2269 ovecsize was not zero.
2273 An unrecognized bit was set in the options argument.
2277 PCRE stores a 4-byte "magic number" at the start of the compiled code,
2278 to catch the case when it is passed a junk pointer and to detect when a
2279 pattern that was compiled in an environment of one endianness is run in
2280 an environment with the other endianness. This is the error that PCRE
2281 gives when the magic number is not present.
2285 While running the pattern match, an unknown item was encountered in the
2286 compiled pattern. This error could be caused by a bug in PCRE or by
2287 overwriting of the compiled pattern.
2291 If a pattern contains back references, but the ovector that is passed
2292 to pcre_exec() is not big enough to remember the referenced substrings,
2293 PCRE gets a block of memory at the start of matching to use for this
2294 purpose. If the call via pcre_malloc() fails, this error is given. The
2295 memory is automatically freed at the end of matching.
2297 This error is also given if pcre_stack_malloc() fails in pcre_exec().
2298 This can happen only when PCRE has been compiled with --disable-stack-
2299 for-recursion.
2303 This error is used by the pcre_copy_substring(), pcre_get_substring(),
2304 and pcre_get_substring_list() functions (see below). It is never
2305 returned by pcre_exec().
2309 The backtracking limit, as specified by the match_limit field in a
2310 pcre_extra structure (or defaulted) was reached. See the description
2311 above.
2315 This error is never generated by pcre_exec() itself. It is provided for
2316 use by callout functions that want to yield a distinctive error code.
2317 See the pcrecallout documentation for details.
2321 A string that contains an invalid UTF-8 byte sequence was passed as a
2322 subject, and the PCRE_NO_UTF8_CHECK option was not set. If the size of
2323 the output vector (ovecsize) is at least 2, the byte offset to the
2324 start of the the invalid UTF-8 character is placed in the first ele-
2325 ment, and a reason code is placed in the second element. The reason
2326 codes are listed in the following section. For backward compatibility,
2327 if PCRE_PARTIAL_HARD is set and the problem is a truncated UTF-8 char-
2328 acter at the end of the subject (reason codes 1 to 5),
2329 PCRE_ERROR_SHORTUTF8 is returned instead of PCRE_ERROR_BADUTF8.
2333 The UTF-8 byte sequence that was passed as a subject was checked and
2334 found to be valid (the PCRE_NO_UTF8_CHECK option was not set), but the
2335 value of startoffset did not point to the beginning of a UTF-8 charac-
2336 ter or the end of the subject.
2340 The subject string did not match, but it did match partially. See the
2341 pcrepartial documentation for details of partial matching.
2345 This code is no longer in use. It was formerly returned when the
2346 PCRE_PARTIAL option was used with a compiled pattern containing items
2347 that were not supported for partial matching. From release 8.00
2348 onwards, there are no restrictions on partial matching.
2352 An unexpected internal error has occurred. This error could be caused
2353 by a bug in PCRE or by overwriting of the compiled pattern.
2357 This error is given if the value of the ovecsize argument is negative.
2361 The internal recursion limit, as specified by the match_limit_recursion
2362 field in a pcre_extra structure (or defaulted) was reached. See the
2363 description above.
2367 An invalid combination of PCRE_NEWLINE_xxx options was given.
2371 The value of startoffset was negative or greater than the length of the
2372 subject, that is, the value in length.
2376 This error is returned instead of PCRE_ERROR_BADUTF8 when the subject
2377 string ends with a truncated UTF-8 character and the PCRE_PARTIAL_HARD
2378 option is set. Information about the failure is returned as for
2379 PCRE_ERROR_BADUTF8. It is in fact sufficient to detect this case, but
2380 this special error code for PCRE_PARTIAL_HARD precedes the implementa-
2381 tion of returned information; it is retained for backwards compatibil-
2382 ity.
2386 This error is returned when pcre_exec() detects a recursion loop within
2387 the pattern. Specifically, it means that either the whole pattern or a
2388 subpattern has been called recursively for the second time at the same
2389 position in the subject string. Some simple patterns that might do this
2390 are detected and faulted at compile time, but more complicated cases,
2391 in particular mutual recursions between two different subpatterns, can-
2392 not be detected until run time.
2394 Error numbers -16 to -20 and -22 are not used by pcre_exec().
2396 Reason codes for invalid UTF-8 strings
2398 When pcre_exec() returns either PCRE_ERROR_BADUTF8 or PCRE_ERROR_SHORT-
2399 UTF8, and the size of the output vector (ovecsize) is at least 2, the
2400 offset of the start of the invalid UTF-8 character is placed in the
2401 first output vector element (ovector[0]) and a reason code is placed in
2402 the second element (ovector[1]). The reason codes are given names in
2403 the pcre.h header file:
2411 The string ends with a truncated UTF-8 character; the code specifies
2412 how many bytes are missing (1 to 5). Although RFC 3629 restricts UTF-8
2413 characters to be no longer than 4 bytes, the encoding scheme (origi-
2414 nally defined by RFC 2279) allows for up to 6 bytes, and this is
2415 checked first; hence the possibility of 4 or 5 missing bytes.
2421 PCRE_UTF8_ERR10
2423 The two most significant bits of the 2nd, 3rd, 4th, 5th, or 6th byte of
2424 the character do not have the binary value 0b10 (that is, either the
2425 most significant bit is 0, or the next bit is 1).
2427 PCRE_UTF8_ERR11
2428 PCRE_UTF8_ERR12
2430 A character that is valid by the RFC 2279 rules is either 5 or 6 bytes
2431 long; these code points are excluded by RFC 3629.
2433 PCRE_UTF8_ERR13
2435 A 4-byte character has a value greater than 0x10fff; these code points
2436 are excluded by RFC 3629.
2438 PCRE_UTF8_ERR14
2440 A 3-byte character has a value in the range 0xd800 to 0xdfff; this
2441 range of code points are reserved by RFC 3629 for use with UTF-16, and
2442 so are excluded from UTF-8.
2444 PCRE_UTF8_ERR15
2445 PCRE_UTF8_ERR16
2446 PCRE_UTF8_ERR17
2447 PCRE_UTF8_ERR18
2448 PCRE_UTF8_ERR19
2450 A 2-, 3-, 4-, 5-, or 6-byte character is "overlong", that is, it codes
2451 for a value that can be represented by fewer bytes, which is invalid.
2452 For example, the two bytes 0xc0, 0xae give the value 0x2e, whose cor-
2453 rect coding uses just one byte.
2455 PCRE_UTF8_ERR20
2457 The two most significant bits of the first byte of a character have the
2458 binary value 0b10 (that is, the most significant bit is 1 and the sec-
2459 ond is 0). Such a byte can only validly occur as the second or subse-
2460 quent byte of a multi-byte character.
2462 PCRE_UTF8_ERR21
2464 The first byte of a character has the value 0xfe or 0xff. These values
2465 can never occur in a valid UTF-8 string.
2470 int pcre_copy_substring(const char *subject, int *ovector,
2471 int stringcount, int stringnumber, char *buffer,
2472 int buffersize);
2474 int pcre_get_substring(const char *subject, int *ovector,
2475 int stringcount, int stringnumber,
2476 const char **stringptr);
2478 int pcre_get_substring_list(const char *subject,
2479 int *ovector, int stringcount, const char ***listptr);
2481 Captured substrings can be accessed directly by using the offsets
2482 returned by pcre_exec() in ovector. For convenience, the functions
2483 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
2484 string_list() are provided for extracting captured substrings as new,
2485 separate, zero-terminated strings. These functions identify substrings
2486 by number. The next section describes functions for extracting named
2487 substrings.
2489 A substring that contains a binary zero is correctly extracted and has
2490 a further zero added on the end, but the result is not, of course, a C
2491 string. However, you can process such a string by referring to the
2492 length that is returned by pcre_copy_substring() and pcre_get_sub-
2493 string(). Unfortunately, the interface to pcre_get_substring_list() is
2494 not adequate for handling strings containing binary zeros, because the
2495 end of the final string is not independently indicated.
2497 The first three arguments are the same for all three of these func-
2498 tions: subject is the subject string that has just been successfully
2499 matched, ovector is a pointer to the vector of integer offsets that was
2500 passed to pcre_exec(), and stringcount is the number of substrings that
2501 were captured by the match, including the substring that matched the
2502 entire regular expression. This is the value returned by pcre_exec() if
2503 it is greater than zero. If pcre_exec() returned zero, indicating that
2504 it ran out of space in ovector, the value passed as stringcount should
2505 be the number of elements in the vector divided by three.
2507 The functions pcre_copy_substring() and pcre_get_substring() extract a
2508 single substring, whose number is given as stringnumber. A value of
2509 zero extracts the substring that matched the entire pattern, whereas
2510 higher values extract the captured substrings. For pcre_copy_sub-
2511 string(), the string is placed in buffer, whose length is given by
2512 buffersize, while for pcre_get_substring() a new block of memory is
2513 obtained via pcre_malloc, and its address is returned via stringptr.
2514 The yield of the function is the length of the string, not including
2515 the terminating zero, or one of these error codes:
2519 The buffer was too small for pcre_copy_substring(), or the attempt to
2520 get memory failed for pcre_get_substring().
2524 There is no substring whose number is stringnumber.
2526 The pcre_get_substring_list() function extracts all available sub-
2527 strings and builds a list of pointers to them. All this is done in a
2528 single block of memory that is obtained via pcre_malloc. The address of
2529 the memory block is returned via listptr, which is also the start of
2530 the list of string pointers. The end of the list is marked by a NULL
2531 pointer. The yield of the function is zero if all went well, or the
2532 error code
2536 if the attempt to get the memory block failed.
2538 When any of these functions encounter a substring that is unset, which
2539 can happen when capturing subpattern number n+1 matches some part of
2540 the subject, but subpattern n has not been used at all, they return an
2541 empty string. This can be distinguished from a genuine zero-length sub-
2542 string by inspecting the appropriate offset in ovector, which is nega-
2543 tive for unset substrings.
2545 The two convenience functions pcre_free_substring() and pcre_free_sub-
2546 string_list() can be used to free the memory returned by a previous
2547 call of pcre_get_substring() or pcre_get_substring_list(), respec-
2548 tively. They do nothing more than call the function pointed to by
2549 pcre_free, which of course could be called directly from a C program.
2550 However, PCRE is used in some situations where it is linked via a spe-
2551 cial interface to another programming language that cannot use
2552 pcre_free directly; it is for these cases that the functions are pro-
2553 vided.
2558 int pcre_get_stringnumber(const pcre *code,
2559 const char *name);
2561 int pcre_copy_named_substring(const pcre *code,
2562 const char *subject, int *ovector,
2563 int stringcount, const char *stringname,
2564 char *buffer, int buffersize);
2566 int pcre_get_named_substring(const pcre *code,
2567 const char *subject, int *ovector,
2568 int stringcount, const char *stringname,
2569 const char **stringptr);
2571 To extract a substring by name, you first have to find associated num-
2572 ber. For example, for this pattern
2574 (a+)b(?<xxx>\d+)...
2576 the number of the subpattern called "xxx" is 2. If the name is known to
2577 be unique (PCRE_DUPNAMES was not set), you can find the number from the
2578 name by calling pcre_get_stringnumber(). The first argument is the com-
2579 piled pattern, and the second is the name. The yield of the function is
2580 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
2581 subpattern of that name.
2583 Given the number, you can extract the substring directly, or use one of
2584 the functions described in the previous section. For convenience, there
2585 are also two functions that do the whole job.
2587 Most of the arguments of pcre_copy_named_substring() and
2588 pcre_get_named_substring() are the same as those for the similarly
2589 named functions that extract by number. As these are described in the
2590 previous section, they are not re-described here. There are just two
2591 differences:
2593 First, instead of a substring number, a substring name is given. Sec-
2594 ond, there is an extra argument, given at the start, which is a pointer
2595 to the compiled pattern. This is needed in order to gain access to the
2596 name-to-number translation table.
2598 These functions call pcre_get_stringnumber(), and if it succeeds, they
2599 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
2600 ate. NOTE: If PCRE_DUPNAMES is set and there are duplicate names, the
2601 behaviour may not be what you want (see the next section).
2603 Warning: If the pattern uses the (?| feature to set up multiple subpat-
2604 terns with the same number, as described in the section on duplicate
2605 subpattern numbers in the pcrepattern page, you cannot use names to
2606 distinguish the different subpatterns, because names are not included
2607 in the compiled code. The matching process uses only numbers. For this
2608 reason, the use of different names for subpatterns of the same number
2609 causes an error at compile time.
2614 int pcre_get_stringtable_entries(const pcre *code,
2615 const char *name, char **first, char **last);
2617 When a pattern is compiled with the PCRE_DUPNAMES option, names for
2618 subpatterns are not required to be unique. (Duplicate names are always
2619 allowed for subpatterns with the same number, created by using the (?|
2620 feature. Indeed, if such subpatterns are named, they are required to
2621 use the same names.)
2623 Normally, patterns with duplicate names are such that in any one match,
2624 only one of the named subpatterns participates. An example is shown in
2625 the pcrepattern documentation.
2627 When duplicates are present, pcre_copy_named_substring() and
2628 pcre_get_named_substring() return the first substring corresponding to
2629 the given name that is set. If none are set, PCRE_ERROR_NOSUBSTRING
2630 (-7) is returned; no data is returned. The pcre_get_stringnumber()
2631 function returns one of the numbers that are associated with the name,
2632 but it is not defined which it is.
2634 If you want to get full details of all captured substrings for a given
2635 name, you must use the pcre_get_stringtable_entries() function. The
2636 first argument is the compiled pattern, and the second is the name. The
2637 third and fourth are pointers to variables which are updated by the
2638 function. After it has run, they point to the first and last entries in
2639 the name-to-number table for the given name. The function itself
2640 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
2641 there are none. The format of the table is described above in the sec-
2642 tion entitled Information about a pattern above. Given all the rele-
2643 vant entries for the name, you can extract each of their numbers, and
2644 hence the captured data, if any.
2649 The traditional matching function uses a similar algorithm to Perl,
2650 which stops when it finds the first match, starting at a given point in
2651 the subject. If you want to find all possible matches, or the longest
2652 possible match, consider using the alternative matching function (see
2653 below) instead. If you cannot use the alternative function, but still
2654 need to find all possible matches, you can kludge it up by making use
2655 of the callout facility, which is described in the pcrecallout documen-
2656 tation.
2658 What you have to do is to insert a callout right at the end of the pat-
2659 tern. When your callout function is called, extract and save the cur-
2660 rent matched substring. Then return 1, which forces pcre_exec() to
2661 backtrack and try other alternatives. Ultimately, when it runs out of
2662 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
2667 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
2668 const char *subject, int length, int startoffset,
2669 int options, int *ovector, int ovecsize,
2670 int *workspace, int wscount);
2672 The function pcre_dfa_exec() is called to match a subject string
2673 against a compiled pattern, using a matching algorithm that scans the
2674 subject string just once, and does not backtrack. This has different
2675 characteristics to the normal algorithm, and is not compatible with
2676 Perl. Some of the features of PCRE patterns are not supported. Never-
2677 theless, there are times when this kind of matching can be useful. For
2678 a discussion of the two matching algorithms, and a list of features
2679 that pcre_dfa_exec() does not support, see the pcrematching documenta-
2680 tion.
2682 The arguments for the pcre_dfa_exec() function are the same as for
2683 pcre_exec(), plus two extras. The ovector argument is used in a differ-
2684 ent way, and this is described below. The other common arguments are
2685 used in the same way as for pcre_exec(), so their description is not
2686 repeated here.
2688 The two additional arguments provide workspace for the function. The
2689 workspace vector should contain at least 20 elements. It is used for
2690 keeping track of multiple paths through the pattern tree. More
2691 workspace will be needed for patterns and subjects where there are a
2692 lot of potential matches.
2694 Here is an example of a simple call to pcre_dfa_exec():
2696 int rc;
2697 int ovector[10];
2698 int wspace[20];
2699 rc = pcre_dfa_exec(
2700 re, /* result of pcre_compile() */
2701 NULL, /* we didn't study the pattern */
2702 "some string", /* the subject string */
2703 11, /* the length of the subject string */
2704 0, /* start at offset 0 in the subject */
2705 0, /* default options */
2706 ovector, /* vector of integers for substring information */
2707 10, /* number of elements (NOT size in bytes) */
2708 wspace, /* working space vector */
2709 20); /* number of elements (NOT size in bytes) */
2711 Option bits for pcre_dfa_exec()
2713 The unused bits of the options argument for pcre_dfa_exec() must be
2714 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
2719 four of these are exactly the same as for pcre_exec(), so their
2720 description is not repeated here.
2725 These have the same general effect as they do for pcre_exec(), but the
2726 details are slightly different. When PCRE_PARTIAL_HARD is set for
2727 pcre_dfa_exec(), it returns PCRE_ERROR_PARTIAL if the end of the sub-
2728 ject is reached and there is still at least one matching possibility
2729 that requires additional characters. This happens even if some complete
2730 matches have also been found. When PCRE_PARTIAL_SOFT is set, the return
2731 code PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end
2732 of the subject is reached, there have been no complete matches, but
2733 there is still at least one matching possibility. The portion of the
2734 string that was inspected when the longest partial match was found is
2735 set as the first matching string in both cases. There is a more
2736 detailed discussion of partial and multi-segment matching, with exam-
2737 ples, in the pcrepartial documentation.
2741 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
2742 stop as soon as it has found one match. Because of the way the alterna-
2743 tive algorithm works, this is necessarily the shortest possible match
2744 at the first possible matching point in the subject string.
2748 When pcre_dfa_exec() returns a partial match, it is possible to call it
2749 again, with additional subject characters, and have it continue with
2750 the same match. The PCRE_DFA_RESTART option requests this action; when
2751 it is set, the workspace and wscount options must reference the same
2752 vector as before because data about the match so far is left in them
2753 after a partial match. There is more discussion of this facility in the
2754 pcrepartial documentation.
2756 Successful returns from pcre_dfa_exec()
2758 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
2759 string in the subject. Note, however, that all the matches from one run
2760 of the function start at the same point in the subject. The shorter
2761 matches are all initial substrings of the longer matches. For example,
2762 if the pattern
2764 <.*>
2766 is matched against the string
2768 This is <something> <something else> <something further> no more
2770 the three matched strings are
2772 <something>
2773 <something> <something else>
2774 <something> <something else> <something further>
2776 On success, the yield of the function is a number greater than zero,
2777 which is the number of matched substrings. The substrings themselves
2778 are returned in ovector. Each string uses two elements; the first is
2779 the offset to the start, and the second is the offset to the end. In
2780 fact, all the strings have the same start offset. (Space could have
2781 been saved by giving this only once, but it was decided to retain some
2782 compatibility with the way pcre_exec() returns data, even though the
2783 meaning of the strings is different.)
2785 The strings are returned in reverse order of length; that is, the long-
2786 est matching string is given first. If there were too many matches to
2787 fit into ovector, the yield of the function is zero, and the vector is
2788 filled with the longest matches.
2790 Error returns from pcre_dfa_exec()
2792 The pcre_dfa_exec() function returns a negative number when it fails.
2793 Many of the errors are the same as for pcre_exec(), and these are
2794 described above. There are in addition the following errors that are
2795 specific to pcre_dfa_exec():
2799 This return is given if pcre_dfa_exec() encounters an item in the pat-
2800 tern that it does not support, for instance, the use of \C or a back
2801 reference.
2805 This return is given if pcre_dfa_exec() encounters a condition item
2806 that uses a back reference for the condition, or a test for recursion
2807 in a specific group. These are not supported.
2811 This return is given if pcre_dfa_exec() is called with an extra block
2812 that contains a setting of the match_limit field. This is not supported
2813 (it is meaningless).
2817 This return is given if pcre_dfa_exec() runs out of space in the
2818 workspace vector.
2822 When a recursive subpattern is processed, the matching function calls
2823 itself recursively, using private vectors for ovector and workspace.
2824 This error is given if the output vector is not large enough. This
2825 should be extremely rare, as a vector of size 1000 is used.
2830 pcrebuild(3), pcrecallout(3), pcrecpp(3)(3), pcrematching(3), pcrepar-
2831 tial(3), pcreposix(3), pcreprecompile(3), pcresample(3), pcrestack(3).
2836 Philip Hazel
2837 University Computing Service
2838 Cambridge CB2 3QH, England.
2843 Last updated: 13 August 2011
2844 Copyright (c) 1997-2011 University of Cambridge.
2845 ------------------------------------------------------------------------------
2851 NAME
2852 PCRE - Perl-compatible regular expressions
2857 int (*pcre_callout)(pcre_callout_block *);
2859 PCRE provides a feature called "callout", which is a means of temporar-
2860 ily passing control to the caller of PCRE in the middle of pattern
2861 matching. The caller of PCRE provides an external function by putting
2862 its entry point in the global variable pcre_callout. By default, this
2863 variable contains NULL, which disables all calling out.
2865 Within a regular expression, (?C) indicates the points at which the
2866 external function is to be called. Different callout points can be
2867 identified by putting a number less than 256 after the letter C. The
2868 default value is zero. For example, this pattern has two callout
2869 points:
2871 (?C1)abc(?C2)def
2873 If the PCRE_AUTO_CALLOUT option bit is set when pcre_compile() or
2874 pcre_compile2() is called, PCRE automatically inserts callouts, all
2875 with number 255, before each item in the pattern. For example, if
2876 PCRE_AUTO_CALLOUT is used with the pattern
2878 A(\d{2}|--)
2880 it is processed as if it were
2882 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
2884 Notice that there is a callout before and after each parenthesis and
2885 alternation bar. Automatic callouts can be used for tracking the
2886 progress of pattern matching. The pcretest command has an option that
2887 sets automatic callouts; when it is used, the output indicates how the
2888 pattern is matched. This is useful information when you are trying to
2889 optimize the performance of a particular pattern.
2894 You should be aware that, because of optimizations in the way PCRE
2895 matches patterns by default, callouts sometimes do not happen. For
2896 example, if the pattern is
2898 ab(?C4)cd
2900 PCRE knows that any matching string must contain the letter "d". If the
2901 subject string is "abyz", the lack of "d" means that matching doesn't
2902 ever start, and the callout is never reached. However, with "abyd",
2903 though the result is still no match, the callout is obeyed.
2905 If the pattern is studied, PCRE knows the minimum length of a matching
2906 string, and will immediately give a "no match" return without actually
2907 running a match if the subject is not long enough, or, for unanchored
2908 patterns, if it has been scanned far enough.
2910 You can disable these optimizations by passing the PCRE_NO_START_OPTI-
2911 MIZE option to pcre_compile(), pcre_exec(), or pcre_dfa_exec(), or by
2912 starting the pattern with (*NO_START_OPT). This slows down the matching
2913 process, but does ensure that callouts such as the example above are
2914 obeyed.
2919 During matching, when PCRE reaches a callout point, the external func-
2920 tion defined by pcre_callout is called (if it is set). This applies to
2921 both the pcre_exec() and the pcre_dfa_exec() matching functions. The
2922 only argument to the callout function is a pointer to a pcre_callout
2923 block. This structure contains the following fields:
2925 int version;
2926 int callout_number;
2927 int *offset_vector;
2928 const char *subject;
2929 int subject_length;
2930 int start_match;
2931 int current_position;
2932 int capture_top;
2933 int capture_last;
2934 void *callout_data;
2935 int pattern_position;
2936 int next_item_length;
2937 const unsigned char *mark;
2939 The version field is an integer containing the version number of the
2940 block format. The initial version was 0; the current version is 2. The
2941 version number will change again in future if additional fields are
2942 added, but the intention is never to remove any of the existing fields.
2944 The callout_number field contains the number of the callout, as com-
2945 piled into the pattern (that is, the number after ?C for manual call-
2946 outs, and 255 for automatically generated callouts).
2948 The offset_vector field is a pointer to the vector of offsets that was
2949 passed by the caller to pcre_exec() or pcre_dfa_exec(). When
2950 pcre_exec() is used, the contents can be inspected in order to extract
2951 substrings that have been matched so far, in the same way as for
2952 extracting substrings after a match has completed. For pcre_dfa_exec()
2953 this field is not useful.
2955 The subject and subject_length fields contain copies of the values that
2956 were passed to pcre_exec().
2958 The start_match field normally contains the offset within the subject
2959 at which the current match attempt started. However, if the escape
2960 sequence \K has been encountered, this value is changed to reflect the
2961 modified starting point. If the pattern is not anchored, the callout
2962 function may be called several times from the same point in the pattern
2963 for different starting points in the subject.
2965 The current_position field contains the offset within the subject of
2966 the current match pointer.
2968 When the pcre_exec() function is used, the capture_top field contains
2969 one more than the number of the highest numbered captured substring so
2970 far. If no substrings have been captured, the value of capture_top is
2971 one. This is always the case when pcre_dfa_exec() is used, because it
2972 does not support captured substrings.
2974 The capture_last field contains the number of the most recently cap-
2975 tured substring. If no substrings have been captured, its value is -1.
2976 This is always the case when pcre_dfa_exec() is used.
2978 The callout_data field contains a value that is passed to pcre_exec()
2979 or pcre_dfa_exec() specifically so that it can be passed back in call-
2980 outs. It is passed in the pcre_callout field of the pcre_extra data
2981 structure. If no such data was passed, the value of callout_data in a
2982 pcre_callout block is NULL. There is a description of the pcre_extra
2983 structure in the pcreapi documentation.
2985 The pattern_position field is present from version 1 of the pcre_call-
2986 out structure. It contains the offset to the next item to be matched in
2987 the pattern string.
2989 The next_item_length field is present from version 1 of the pcre_call-
2990 out structure. It contains the length of the next item to be matched in
2991 the pattern string. When the callout immediately precedes an alterna-
2992 tion bar, a closing parenthesis, or the end of the pattern, the length
2993 is zero. When the callout precedes an opening parenthesis, the length
2994 is that of the entire subpattern.
2996 The pattern_position and next_item_length fields are intended to help
2997 in distinguishing between different automatic callouts, which all have
2998 the same callout number. However, they are set for all callouts.
3000 The mark field is present from version 2 of the pcre_callout structure.
3001 In callouts from pcre_exec() it contains a pointer to the zero-termi-
3002 nated name of the most recently passed (*MARK) item in the match, or
3003 NULL if there are no (*MARK)s in the current matching path. In callouts
3004 from pcre_dfa_exec() this field always contains NULL.
3009 The external callout function returns an integer to PCRE. If the value
3010 is zero, matching proceeds as normal. If the value is greater than
3011 zero, matching fails at the current point, but the testing of other
3012 matching possibilities goes ahead, just as if a lookahead assertion had
3013 failed. If the value is less than zero, the match is abandoned, and
3014 pcre_exec() or pcre_dfa_exec() returns the negative value.
3016 Negative values should normally be chosen from the set of
3017 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
3018 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
3019 reserved for use by callout functions; it will never be used by PCRE
3020 itself.
3025 Philip Hazel
3026 University Computing Service
3027 Cambridge CB2 3QH, England.
3032 Last updated: 31 July 2011
3033 Copyright (c) 1997-2011 University of Cambridge.
3034 ------------------------------------------------------------------------------
3040 NAME
3041 PCRE - Perl-compatible regular expressions
3046 This document describes the differences in the ways that PCRE and Perl
3047 handle regular expressions. The differences described here are with
3048 respect to Perl versions 5.10 and above.
3050 1. PCRE has only a subset of Perl's UTF-8 and Unicode support. Details
3051 of what it does have are given in the pcreunicode page.
3053 2. PCRE allows repeat quantifiers only on parenthesized assertions, but
3054 they do not mean what you might think. For example, (?!a){3} does not
3055 assert that the next three characters are not "a". It just asserts that
3056 the next character is not "a" three times (in principle: PCRE optimizes
3057 this to run the assertion just once). Perl allows repeat quantifiers on
3058 other assertions such as \b, but these do not seem to have any use.
3060 3. Capturing subpatterns that occur inside negative lookahead asser-
3061 tions are counted, but their entries in the offsets vector are never
3062 set. Perl sets its numerical variables from any such patterns that are
3063 matched before the assertion fails to match something (thereby succeed-
3064 ing), but only if the negative lookahead assertion contains just one
3065 branch.
3067 4. Though binary zero characters are supported in the subject string,
3068 they are not allowed in a pattern string because it is passed as a nor-
3069 mal C string, terminated by zero. The escape sequence \0 can be used in
3070 the pattern to represent a binary zero.
3072 5. The following Perl escape sequences are not supported: \l, \u, \L,
3073 \U, and \N when followed by a character name or Unicode value. (\N on
3074 its own, matching a non-newline character, is supported.) In fact these
3075 are implemented by Perl's general string-handling and are not part of
3076 its pattern matching engine. If any of these are encountered by PCRE,
3077 an error is generated.
3079 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
3080 is built with Unicode character property support. The properties that
3081 can be tested with \p and \P are limited to the general category prop-
3082 erties such as Lu and Nd, script names such as Greek or Han, and the
3083 derived properties Any and L&. PCRE does support the Cs (surrogate)
3084 property, which Perl does not; the Perl documentation says "Because
3085 Perl hides the need for the user to understand the internal representa-
3086 tion of Unicode characters, there is no need to implement the somewhat
3087 messy concept of surrogates."
3089 7. PCRE implements a simpler version of \X than Perl, which changed to
3090 make \X match what Unicode calls an "extended grapheme cluster". This
3091 is more complicated than an extended Unicode sequence, which is what
3092 PCRE matches.
3094 8. PCRE does support the \Q...\E escape for quoting substrings. Charac-
3095 ters in between are treated as literals. This is slightly different
3096 from Perl in that $ and @ are also handled as literals inside the
3097 quotes. In Perl, they cause variable interpolation (but of course PCRE
3098 does not have variables). Note the following examples:
3100 Pattern PCRE matches Perl matches
3102 \Qabc$xyz\E abc$xyz abc followed by the
3103 contents of $xyz
3104 \Qabc\$xyz\E abc\$xyz abc\$xyz
3105 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3107 The \Q...\E sequence is recognized both inside and outside character
3108 classes.
3110 9. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
3111 constructions. However, there is support for recursive patterns. This
3112 is not available in Perl 5.8, but it is in Perl 5.10. Also, the PCRE
3113 "callout" feature allows an external function to be called during pat-
3114 tern matching. See the pcrecallout documentation for details.
3116 10. Subpatterns that are called recursively or as "subroutines" are
3117 always treated as atomic groups in PCRE. This is like Python, but
3118 unlike Perl. There is a discussion of an example that explains this in
3119 more detail in the section on recursion differences from Perl in the
3120 pcrepattern page.
3122 11. There are some differences that are concerned with the settings of
3123 captured strings when part of a pattern is repeated. For example,
3124 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
3125 unset, but in PCRE it is set to "b".
3127 12. PCRE's handling of duplicate subpattern numbers and duplicate sub-
3128 pattern names is not as general as Perl's. This is a consequence of the
3129 fact the PCRE works internally just with numbers, using an external ta-
3130 ble to translate between numbers and names. In particular, a pattern
3131 such as (?|(?<a>A)|(?<b)B), where the two capturing parentheses have
3132 the same number but different names, is not supported, and causes an
3133 error at compile time. If it were allowed, it would not be possible to
3134 distinguish which parentheses matched, because both names map to cap-
3135 turing subpattern number 1. To avoid this confusing situation, an error
3136 is given at compile time.
3138 13. Perl recognizes comments in some places that PCRE does not, for
3139 example, between the ( and ? at the start of a subpattern. If the /x
3140 modifier is set, Perl allows whitespace between ( and ? but PCRE never
3141 does, even if the PCRE_EXTENDED option is set.
3143 14. PCRE provides some extensions to the Perl regular expression facil-
3144 ities. Perl 5.10 includes new features that are not in earlier ver-
3145 sions of Perl, some of which (such as named parentheses) have been in
3146 PCRE for some time. This list is with respect to Perl 5.10:
3148 (a) Although lookbehind assertions in PCRE must match fixed length
3149 strings, each alternative branch of a lookbehind assertion can match a
3150 different length of string. Perl requires them all to have the same
3151 length.
3153 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
3154 meta-character matches only at the very end of the string.
3156 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
3157 cial meaning is faulted. Otherwise, like Perl, the backslash is quietly
3158 ignored. (Perl can be made to issue a warning.)
3160 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
3161 fiers is inverted, that is, by default they are not greedy, but if fol-
3162 lowed by a question mark they are.
3164 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
3165 tried only at the first matching position in the subject string.
3168 and PCRE_NO_AUTO_CAPTURE options for pcre_exec() have no Perl equiva-
3169 lents.
3171 (g) The \R escape sequence can be restricted to match only CR, LF, or
3172 CRLF by the PCRE_BSR_ANYCRLF option.
3174 (h) The callout facility is PCRE-specific.
3176 (i) The partial matching facility is PCRE-specific.
3178 (j) Patterns compiled by PCRE can be saved and re-used at a later time,
3179 even on different hosts that have the other endianness.
3181 (k) The alternative matching function (pcre_dfa_exec()) matches in a
3182 different way and is not Perl-compatible.
3184 (l) PCRE recognizes some special sequences such as (*CR) at the start
3185 of a pattern that set overall options that cannot be changed within the
3186 pattern.
3191 Philip Hazel
3192 University Computing Service
3193 Cambridge CB2 3QH, England.
3198 Last updated: 24 August 2011
3199 Copyright (c) 1997-2011 University of Cambridge.
3200 ------------------------------------------------------------------------------
3206 NAME
3207 PCRE - Perl-compatible regular expressions
3212 The syntax and semantics of the regular expressions that are supported
3213 by PCRE are described in detail below. There is a quick-reference syn-
3214 tax summary in the pcresyntax page. PCRE tries to match Perl syntax and
3215 semantics as closely as it can. PCRE also supports some alternative
3216 regular expression syntax (which does not conflict with the Perl syn-
3217 tax) in order to provide some compatibility with regular expressions in
3218 Python, .NET, and Oniguruma.
3220 Perl's regular expressions are described in its own documentation, and
3221 regular expressions in general are covered in a number of books, some
3222 of which have copious examples. Jeffrey Friedl's "Mastering Regular
3223 Expressions", published by O'Reilly, covers regular expressions in
3224 great detail. This description of PCRE's regular expressions is
3225 intended as reference material.
3227 The original operation of PCRE was on strings of one-byte characters.
3228 However, there is now also support for UTF-8 character strings. To use
3229 this, PCRE must be built to include UTF-8 support, and you must call
3230 pcre_compile() or pcre_compile2() with the PCRE_UTF8 option. There is
3231 also a special sequence that can be given at the start of a pattern:
3233 (*UTF8)
3235 Starting a pattern with this sequence is equivalent to setting the
3236 PCRE_UTF8 option. This feature is not Perl-compatible. How setting
3237 UTF-8 mode affects pattern matching is mentioned in several places
3238 below. There is also a summary of UTF-8 features in the pcreunicode
3239 page.
3241 Another special sequence that may appear at the start of a pattern or
3242 in combination with (*UTF8) is:
3244 (*UCP)
3246 This has the same effect as setting the PCRE_UCP option: it causes
3247 sequences such as \d and \w to use Unicode properties to determine
3248 character types, instead of recognizing only characters with codes less
3249 than 128 via a lookup table.
3251 If a pattern starts with (*NO_START_OPT), it has the same effect as
3252 setting the PCRE_NO_START_OPTIMIZE option either at compile or matching
3253 time. There are also some more of these special sequences that are con-
3254 cerned with the handling of newlines; they are described below.
3256 The remainder of this document discusses the patterns that are sup-
3257 ported by PCRE when its main matching function, pcre_exec(), is used.
3258 From release 6.0, PCRE offers a second matching function,
3259 pcre_dfa_exec(), which matches using a different algorithm that is not
3260 Perl-compatible. Some of the features discussed below are not available
3261 when pcre_dfa_exec() is used. The advantages and disadvantages of the
3262 alternative function, and how it differs from the normal function, are
3263 discussed in the pcrematching page.
3268 PCRE supports five different conventions for indicating line breaks in
3269 strings: a single CR (carriage return) character, a single LF (line-
3270 feed) character, the two-character sequence CRLF, any of the three pre-
3271 ceding, or any Unicode newline sequence. The pcreapi page has further
3272 discussion about newlines, and shows how to set the newline convention
3273 in the options arguments for the compiling and matching functions.
3275 It is also possible to specify a newline convention by starting a pat-
3276 tern string with one of the following five sequences:
3278 (*CR) carriage return
3279 (*LF) linefeed
3280 (*CRLF) carriage return, followed by linefeed
3281 (*ANYCRLF) any of the three above
3282 (*ANY) all Unicode newline sequences
3284 These override the default and the options given to pcre_compile() or
3285 pcre_compile2(). For example, on a Unix system where LF is the default
3286 newline sequence, the pattern
3288 (*CR)a.b
3290 changes the convention to CR. That pattern matches "a\nb" because LF is
3291 no longer a newline. Note that these special settings, which are not
3292 Perl-compatible, are recognized only at the very start of a pattern,
3293 and that they must be in upper case. If more than one of them is
3294 present, the last one is used.
3296 The newline convention affects the interpretation of the dot metachar-
3297 acter when PCRE_DOTALL is not set, and also the behaviour of \N. How-
3298 ever, it does not affect what the \R escape sequence matches. By
3299 default, this is any Unicode newline sequence, for Perl compatibility.
3300 However, this can be changed; see the description of \R in the section
3301 entitled "Newline sequences" below. A change of \R setting can be com-
3302 bined with a change of newline convention.
3307 A regular expression is a pattern that is matched against a subject
3308 string from left to right. Most characters stand for themselves in a
3309 pattern, and match the corresponding characters in the subject. As a
3310 trivial example, the pattern
3312 The quick brown fox
3314 matches a portion of a subject string that is identical to itself. When
3315 caseless matching is specified (the PCRE_CASELESS option), letters are
3316 matched independently of case. In UTF-8 mode, PCRE always understands
3317 the concept of case for characters whose values are less than 128, so
3318 caseless matching is always possible. For characters with higher val-
3319 ues, the concept of case is supported if PCRE is compiled with Unicode
3320 property support, but not otherwise. If you want to use caseless
3321 matching for characters 128 and above, you must ensure that PCRE is
3322 compiled with Unicode property support as well as with UTF-8 support.
3324 The power of regular expressions comes from the ability to include
3325 alternatives and repetitions in the pattern. These are encoded in the
3326 pattern by the use of metacharacters, which do not stand for themselves
3327 but instead are interpreted in some special way.
3329 There are two different sets of metacharacters: those that are recog-
3330 nized anywhere in the pattern except within square brackets, and those
3331 that are recognized within square brackets. Outside square brackets,
3332 the metacharacters are as follows:
3334 \ general escape character with several uses
3335 ^ assert start of string (or line, in multiline mode)
3336 $ assert end of string (or line, in multiline mode)
3337 . match any character except newline (by default)
3338 [ start character class definition
3339 | start of alternative branch
3340 ( start subpattern
3341 ) end subpattern
3342 ? extends the meaning of (
3343 also 0 or 1 quantifier
3344 also quantifier minimizer
3345 * 0 or more quantifier
3346 + 1 or more quantifier
3347 also "possessive quantifier"
3348 { start min/max quantifier
3350 Part of a pattern that is in square brackets is called a "character
3351 class". In a character class the only metacharacters are:
3353 \ general escape character
3354 ^ negate the class, but only if the first character
3355 - indicates character range
3356 [ POSIX character class (only if followed by POSIX
3357 syntax)
3358 ] terminates the character class
3360 The following sections describe the use of each of the metacharacters.
3365 The backslash character has several uses. Firstly, if it is followed by
3366 a character that is not a number or a letter, it takes away any special
3367 meaning that character may have. This use of backslash as an escape
3368 character applies both inside and outside character classes.
3370 For example, if you want to match a * character, you write \* in the
3371 pattern. This escaping action applies whether or not the following
3372 character would otherwise be interpreted as a metacharacter, so it is
3373 always safe to precede a non-alphanumeric with backslash to specify
3374 that it stands for itself. In particular, if you want to match a back-
3375 slash, you write \\.
3377 In UTF-8 mode, only ASCII numbers and letters have any special meaning
3378 after a backslash. All other characters (in particular, those whose
3379 codepoints are greater than 127) are treated as literals.
3381 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
3382 the pattern (other than in a character class) and characters between a
3383 # outside a character class and the next newline are ignored. An escap-
3384 ing backslash can be used to include a whitespace or # character as
3385 part of the pattern.
3387 If you want to remove the special meaning from a sequence of charac-
3388 ters, you can do so by putting them between \Q and \E. This is differ-
3389 ent from Perl in that $ and @ are handled as literals in \Q...\E
3390 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
3391 tion. Note the following examples:
3393 Pattern PCRE matches Perl matches
3395 \Qabc$xyz\E abc$xyz abc followed by the
3396 contents of $xyz
3397 \Qabc\$xyz\E abc\$xyz abc\$xyz
3398 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3400 The \Q...\E sequence is recognized both inside and outside character
3401 classes. An isolated \E that is not preceded by \Q is ignored. If \Q
3402 is not followed by \E later in the pattern, the literal interpretation
3403 continues to the end of the pattern (that is, \E is assumed at the
3404 end). If the isolated \Q is inside a character class, this causes an
3405 error, because the character class is not terminated.
3407 Non-printing characters
3409 A second use of backslash provides a way of encoding non-printing char-
3410 acters in patterns in a visible manner. There is no restriction on the
3411 appearance of non-printing characters, apart from the binary zero that
3412 terminates a pattern, but when a pattern is being prepared by text
3413 editing, it is often easier to use one of the following escape
3414 sequences than the binary character it represents:
3416 \a alarm, that is, the BEL character (hex 07)
3417 \cx "control-x", where x is any ASCII character
3418 \e escape (hex 1B)
3419 \f formfeed (hex 0C)
3420 \n linefeed (hex 0A)
3421 \r carriage return (hex 0D)
3422 \t tab (hex 09)
3423 \ddd character with octal code ddd, or back reference
3424 \xhh character with hex code hh
3425 \x{hhh..} character with hex code hhh..
3427 The precise effect of \cx is as follows: if x is a lower case letter,
3428 it is converted to upper case. Then bit 6 of the character (hex 40) is
3429 inverted. Thus \cz becomes hex 1A (z is 7A), but \c{ becomes hex 3B ({
3430 is 7B), while \c; becomes hex 7B (; is 3B). If the byte following \c
3431 has a value greater than 127, a compile-time error occurs. This locks
3432 out non-ASCII characters in both byte mode and UTF-8 mode. (When PCRE
3433 is compiled in EBCDIC mode, all byte values are valid. A lower case
3434 letter is converted to upper case, and then the 0xc0 bits are flipped.)
3436 After \x, from zero to two hexadecimal digits are read (letters can be
3437 in upper or lower case). Any number of hexadecimal digits may appear
3438 between \x{ and }, but the value of the character code must be less
3439 than 256 in non-UTF-8 mode, and less than 2**31 in UTF-8 mode. That is,
3440 the maximum value in hexadecimal is 7FFFFFFF. Note that this is bigger
3441 than the largest Unicode code point, which is 10FFFF.
3443 If characters other than hexadecimal digits appear between \x{ and },
3444 or if there is no terminating }, this form of escape is not recognized.
3445 Instead, the initial \x will be interpreted as a basic hexadecimal
3446 escape, with no following digits, giving a character whose value is
3447 zero.
3449 Characters whose value is less than 256 can be defined by either of the
3450 two syntaxes for \x. There is no difference in the way they are han-
3451 dled. For example, \xdc is exactly the same as \x{dc}.
3453 After \0 up to two further octal digits are read. If there are fewer
3454 than two digits, just those that are present are used. Thus the
3455 sequence \0\x\07 specifies two binary zeros followed by a BEL character
3456 (code value 7). Make sure you supply two digits after the initial zero
3457 if the pattern character that follows is itself an octal digit.
3459 The handling of a backslash followed by a digit other than 0 is compli-
3460 cated. Outside a character class, PCRE reads it and any following dig-
3461 its as a decimal number. If the number is less than 10, or if there
3462 have been at least that many previous capturing left parentheses in the
3463 expression, the entire sequence is taken as a back reference. A
3464 description of how this works is given later, following the discussion
3465 of parenthesized subpatterns.
3467 Inside a character class, or if the decimal number is greater than 9
3468 and there have not been that many capturing subpatterns, PCRE re-reads
3469 up to three octal digits following the backslash, and uses them to gen-
3470 erate a data character. Any subsequent digits stand for themselves. In
3471 non-UTF-8 mode, the value of a character specified in octal must be
3472 less than \400. In UTF-8 mode, values up to \777 are permitted. For
3473 example:
3475 \040 is another way of writing a space
3476 \40 is the same, provided there are fewer than 40
3477 previous capturing subpatterns
3478 \7 is always a back reference
3479 \11 might be a back reference, or another way of
3480 writing a tab
3481 \011 is always a tab
3482 \0113 is a tab followed by the character "3"
3483 \113 might be a back reference, otherwise the
3484 character with octal code 113
3485 \377 might be a back reference, otherwise
3486 the byte consisting entirely of 1 bits
3487 \81 is either a back reference, or a binary zero
3488 followed by the two characters "8" and "1"
3490 Note that octal values of 100 or greater must not be introduced by a
3491 leading zero, because no more than three octal digits are ever read.
3493 All the sequences that define a single character value can be used both
3494 inside and outside character classes. In addition, inside a character
3495 class, the sequence \b is interpreted as the backspace character (hex
3496 08). The sequences \B, \N, \R, and \X are not special inside a charac-
3497 ter class. Like any other unrecognized escape sequences, they are
3498 treated as the literal characters "B", "N", "R", and "X" by default,
3499 but cause an error if the PCRE_EXTRA option is set. Outside a character
3500 class, these sequences have different meanings.
3502 Absolute and relative back references
3504 The sequence \g followed by an unsigned or a negative number, option-
3505 ally enclosed in braces, is an absolute or relative back reference. A
3506 named back reference can be coded as \g{name}. Back references are dis-
3507 cussed later, following the discussion of parenthesized subpatterns.
3509 Absolute and relative subroutine calls
3511 For compatibility with Oniguruma, the non-Perl syntax \g followed by a
3512 name or a number enclosed either in angle brackets or single quotes, is
3513 an alternative syntax for referencing a subpattern as a "subroutine".
3514 Details are discussed later. Note that \g{...} (Perl syntax) and
3515 \g<...> (Oniguruma syntax) are not synonymous. The former is a back
3516 reference; the latter is a subroutine call.
3518 Generic character types
3520 Another use of backslash is for specifying generic character types:
3522 \d any decimal digit
3523 \D any character that is not a decimal digit
3524 \h any horizontal whitespace character
3525 \H any character that is not a horizontal whitespace character
3526 \s any whitespace character
3527 \S any character that is not a whitespace character
3528 \v any vertical whitespace character
3529 \V any character that is not a vertical whitespace character
3530 \w any "word" character
3531 \W any "non-word" character
3533 There is also the single sequence \N, which matches a non-newline char-
3534 acter. This is the same as the "." metacharacter when PCRE_DOTALL is
3535 not set.
3537 Each pair of lower and upper case escape sequences partitions the com-
3538 plete set of characters into two disjoint sets. Any given character
3539 matches one, and only one, of each pair. The sequences can appear both
3540 inside and outside character classes. They each match one character of
3541 the appropriate type. If the current matching point is at the end of
3542 the subject string, all of them fail, because there is no character to
3543 match.
3545 For compatibility with Perl, \s does not match the VT character (code
3546 11). This makes it different from the the POSIX "space" class. The \s
3547 characters are HT (9), LF (10), FF (12), CR (13), and space (32). If
3548 "use locale;" is included in a Perl script, \s may match the VT charac-
3549 ter. In PCRE, it never does.
3551 A "word" character is an underscore or any character that is a letter
3552 or digit. By default, the definition of letters and digits is con-
3553 trolled by PCRE's low-valued character tables, and may vary if locale-
3554 specific matching is taking place (see "Locale support" in the pcreapi
3555 page). For example, in a French locale such as "fr_FR" in Unix-like
3556 systems, or "french" in Windows, some character codes greater than 128
3557 are used for accented letters, and these are then matched by \w. The
3558 use of locales with Unicode is discouraged.
3560 By default, in UTF-8 mode, characters with values greater than 128
3561 never match \d, \s, or \w, and always match \D, \S, and \W. These
3562 sequences retain their original meanings from before UTF-8 support was
3563 available, mainly for efficiency reasons. However, if PCRE is compiled
3564 with Unicode property support, and the PCRE_UCP option is set, the be-
3565 haviour is changed so that Unicode properties are used to determine
3566 character types, as follows:
3568 \d any character that \p{Nd} matches (decimal digit)
3569 \s any character that \p{Z} matches, plus HT, LF, FF, CR
3570 \w any character that \p{L} or \p{N} matches, plus underscore
3572 The upper case escapes match the inverse sets of characters. Note that
3573 \d matches only decimal digits, whereas \w matches any Unicode digit,
3574 as well as any Unicode letter, and underscore. Note also that PCRE_UCP
3575 affects \b, and \B because they are defined in terms of \w and \W.
3576 Matching these sequences is noticeably slower when PCRE_UCP is set.
3578 The sequences \h, \H, \v, and \V are features that were added to Perl
3579 at release 5.10. In contrast to the other sequences, which match only
3580 ASCII characters by default, these always match certain high-valued
3581 codepoints in UTF-8 mode, whether or not PCRE_UCP is set. The horizon-
3582 tal space characters are:
3584 U+0009 Horizontal tab
3585 U+0020 Space
3586 U+00A0 Non-break space
3587 U+1680 Ogham space mark
3588 U+180E Mongolian vowel separator
3589 U+2000 En quad
3590 U+2001 Em quad
3591 U+2002 En space
3592 U+2003 Em space
3593 U+2004 Three-per-em space
3594 U+2005 Four-per-em space
3595 U+2006 Six-per-em space
3596 U+2007 Figure space
3597 U+2008 Punctuation space
3598 U+2009 Thin space
3599 U+200A Hair space
3600 U+202F Narrow no-break space
3601 U+205F Medium mathematical space
3602 U+3000 Ideographic space
3604 The vertical space characters are:
3606 U+000A Linefeed
3607 U+000B Vertical tab
3608 U+000C Formfeed
3609 U+000D Carriage return
3610 U+0085 Next line
3611 U+2028 Line separator
3612 U+2029 Paragraph separator
3614 Newline sequences
3616 Outside a character class, by default, the escape sequence \R matches
3617 any Unicode newline sequence. In non-UTF-8 mode \R is equivalent to the
3618 following:
3620 (?>\r\n|\n|\x0b|\f|\r|\x85)
3622 This is an example of an "atomic group", details of which are given
3623 below. This particular group matches either the two-character sequence
3624 CR followed by LF, or one of the single characters LF (linefeed,
3625 U+000A), VT (vertical tab, U+000B), FF (formfeed, U+000C), CR (carriage
3626 return, U+000D), or NEL (next line, U+0085). The two-character sequence
3627 is treated as a single unit that cannot be split.
3629 In UTF-8 mode, two additional characters whose codepoints are greater
3630 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
3631 rator, U+2029). Unicode character property support is not needed for
3632 these characters to be recognized.
3634 It is possible to restrict \R to match only CR, LF, or CRLF (instead of
3635 the complete set of Unicode line endings) by setting the option
3636 PCRE_BSR_ANYCRLF either at compile time or when the pattern is matched.
3637 (BSR is an abbrevation for "backslash R".) This can be made the default
3638 when PCRE is built; if this is the case, the other behaviour can be
3639 requested via the PCRE_BSR_UNICODE option. It is also possible to
3640 specify these settings by starting a pattern string with one of the
3641 following sequences:
3643 (*BSR_ANYCRLF) CR, LF, or CRLF only
3644 (*BSR_UNICODE) any Unicode newline sequence
3646 These override the default and the options given to pcre_compile() or
3647 pcre_compile2(), but they can be overridden by options given to
3648 pcre_exec() or pcre_dfa_exec(). Note that these special settings, which
3649 are not Perl-compatible, are recognized only at the very start of a
3650 pattern, and that they must be in upper case. If more than one of them
3651 is present, the last one is used. They can be combined with a change of
3652 newline convention; for example, a pattern can start with:
3656 They can also be combined with the (*UTF8) or (*UCP) special sequences.
3657 Inside a character class, \R is treated as an unrecognized escape
3658 sequence, and so matches the letter "R" by default, but causes an error
3659 if PCRE_EXTRA is set.
3661 Unicode character properties
3663 When PCRE is built with Unicode character property support, three addi-
3664 tional escape sequences that match characters with specific properties
3665 are available. When not in UTF-8 mode, these sequences are of course
3666 limited to testing characters whose codepoints are less than 256, but
3667 they do work in this mode. The extra escape sequences are:
3669 \p{xx} a character with the xx property
3670 \P{xx} a character without the xx property
3671 \X an extended Unicode sequence
3673 The property names represented by xx above are limited to the Unicode
3674 script names, the general category properties, "Any", which matches any
3675 character (including newline), and some special PCRE properties
3676 (described in the next section). Other Perl properties such as "InMu-
3677 sicalSymbols" are not currently supported by PCRE. Note that \P{Any}
3678 does not match any characters, so always causes a match failure.
3680 Sets of Unicode characters are defined as belonging to certain scripts.
3681 A character from one of these sets can be matched using a script name.
3682 For example:
3684 \p{Greek}
3685 \P{Han}
3687 Those that are not part of an identified script are lumped together as
3688 "Common". The current list of scripts is:
3690 Arabic, Armenian, Avestan, Balinese, Bamum, Bengali, Bopomofo, Braille,
3691 Buginese, Buhid, Canadian_Aboriginal, Carian, Cham, Cherokee, Common,
3692 Coptic, Cuneiform, Cypriot, Cyrillic, Deseret, Devanagari, Egyp-
3693 tian_Hieroglyphs, Ethiopic, Georgian, Glagolitic, Gothic, Greek,
3694 Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hiragana, Impe-
3695 rial_Aramaic, Inherited, Inscriptional_Pahlavi, Inscriptional_Parthian,
3696 Javanese, Kaithi, Kannada, Katakana, Kayah_Li, Kharoshthi, Khmer, Lao,
3697 Latin, Lepcha, Limbu, Linear_B, Lisu, Lycian, Lydian, Malayalam,
3698 Meetei_Mayek, Mongolian, Myanmar, New_Tai_Lue, Nko, Ogham, Old_Italic,
3699 Old_Persian, Old_South_Arabian, Old_Turkic, Ol_Chiki, Oriya, Osmanya,
3700 Phags_Pa, Phoenician, Rejang, Runic, Samaritan, Saurashtra, Shavian,
3701 Sinhala, Sundanese, Syloti_Nagri, Syriac, Tagalog, Tagbanwa, Tai_Le,
3702 Tai_Tham, Tai_Viet, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh,
3703 Ugaritic, Vai, Yi.
3705 Each character has exactly one Unicode general category property, spec-
3706 ified by a two-letter abbreviation. For compatibility with Perl, nega-
3707 tion can be specified by including a circumflex between the opening
3708 brace and the property name. For example, \p{^Lu} is the same as
3709 \P{Lu}.
3711 If only one letter is specified with \p or \P, it includes all the gen-
3712 eral category properties that start with that letter. In this case, in
3713 the absence of negation, the curly brackets in the escape sequence are
3714 optional; these two examples have the same effect:
3716 \p{L}
3717 \pL
3719 The following general category property codes are supported:
3721 C Other
3722 Cc Control
3723 Cf Format
3724 Cn Unassigned
3725 Co Private use
3726 Cs Surrogate
3728 L Letter
3729 Ll Lower case letter
3730 Lm Modifier letter
3731 Lo Other letter
3732 Lt Title case letter
3733 Lu Upper case letter
3735 M Mark
3736 Mc Spacing mark
3737 Me Enclosing mark
3738 Mn Non-spacing mark
3740 N Number
3741 Nd Decimal number
3742 Nl Letter number
3743 No Other number
3745 P Punctuation
3746 Pc Connector punctuation
3747 Pd Dash punctuation
3748 Pe Close punctuation
3749 Pf Final punctuation
3750 Pi Initial punctuation
3751 Po Other punctuation
3752 Ps Open punctuation
3754 S Symbol
3755 Sc Currency symbol
3756 Sk Modifier symbol
3757 Sm Mathematical symbol
3758 So Other symbol
3760 Z Separator
3761 Zl Line separator
3762 Zp Paragraph separator
3763 Zs Space separator
3765 The special property L& is also supported: it matches a character that
3766 has the Lu, Ll, or Lt property, in other words, a letter that is not
3767 classified as a modifier or "other".
3769 The Cs (Surrogate) property applies only to characters in the range
3770 U+D800 to U+DFFF. Such characters are not valid in UTF-8 strings (see
3771 RFC 3629) and so cannot be tested by PCRE, unless UTF-8 validity check-
3772 ing has been turned off (see the discussion of PCRE_NO_UTF8_CHECK in
3773 the pcreapi page). Perl does not support the Cs property.
3775 The long synonyms for property names that Perl supports (such as
3776 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
3777 any of these properties with "Is".
3779 No character that is in the Unicode table has the Cn (unassigned) prop-
3780 erty. Instead, this property is assumed for any code point that is not
3781 in the Unicode table.
3783 Specifying caseless matching does not affect these escape sequences.
3784 For example, \p{Lu} always matches only upper case letters.
3786 The \X escape matches any number of Unicode characters that form an
3787 extended Unicode sequence. \X is equivalent to
3789 (?>\PM\pM*)
3791 That is, it matches a character without the "mark" property, followed
3792 by zero or more characters with the "mark" property, and treats the
3793 sequence as an atomic group (see below). Characters with the "mark"
3794 property are typically accents that affect the preceding character.
3795 None of them have codepoints less than 256, so in non-UTF-8 mode \X
3796 matches any one character.
3798 Note that recent versions of Perl have changed \X to match what Unicode
3799 calls an "extended grapheme cluster", which has a more complicated def-
3800 inition.
3802 Matching characters by Unicode property is not fast, because PCRE has
3803 to search a structure that contains data for over fifteen thousand
3804 characters. That is why the traditional escape sequences such as \d and
3805 \w do not use Unicode properties in PCRE by default, though you can
3806 make them do so by setting the PCRE_UCP option for pcre_compile() or by
3807 starting the pattern with (*UCP).
3809 PCRE's additional properties
3811 As well as the standard Unicode properties described in the previous
3812 section, PCRE supports four more that make it possible to convert tra-
3813 ditional escape sequences such as \w and \s and POSIX character classes
3814 to use Unicode properties. PCRE uses these non-standard, non-Perl prop-
3815 erties internally when PCRE_UCP is set. They are:
3817 Xan Any alphanumeric character
3818 Xps Any POSIX space character
3819 Xsp Any Perl space character
3820 Xwd Any Perl "word" character
3822 Xan matches characters that have either the L (letter) or the N (num-
3823 ber) property. Xps matches the characters tab, linefeed, vertical tab,
3824 formfeed, or carriage return, and any other character that has the Z
3825 (separator) property. Xsp is the same as Xps, except that vertical tab
3826 is excluded. Xwd matches the same characters as Xan, plus underscore.
3828 Resetting the match start
3830 The escape sequence \K causes any previously matched characters not to
3831 be included in the final matched sequence. For example, the pattern:
3833 foo\Kbar
3835 matches "foobar", but reports that it has matched "bar". This feature
3836 is similar to a lookbehind assertion (described below). However, in
3837 this case, the part of the subject before the real match does not have
3838 to be of fixed length, as lookbehind assertions do. The use of \K does
3839 not interfere with the setting of captured substrings. For example,
3840 when the pattern
3842 (foo)\Kbar
3844 matches "foobar", the first substring is still set to "foo".
3846 Perl documents that the use of \K within assertions is "not well
3847 defined". In PCRE, \K is acted upon when it occurs inside positive
3848 assertions, but is ignored in negative assertions.
3850 Simple assertions
3852 The final use of backslash is for certain simple assertions. An asser-
3853 tion specifies a condition that has to be met at a particular point in
3854 a match, without consuming any characters from the subject string. The
3855 use of subpatterns for more complicated assertions is described below.
3856 The backslashed assertions are:
3858 \b matches at a word boundary
3859 \B matches when not at a word boundary
3860 \A matches at the start of the subject
3861 \Z matches at the end of the subject
3862 also matches before a newline at the end of the subject
3863 \z matches only at the end of the subject
3864 \G matches at the first matching position in the subject
3866 Inside a character class, \b has a different meaning; it matches the
3867 backspace character. If any other of these assertions appears in a
3868 character class, by default it matches the corresponding literal char-
3869 acter (for example, \B matches the letter B). However, if the
3870 PCRE_EXTRA option is set, an "invalid escape sequence" error is gener-
3871 ated instead.
3873 A word boundary is a position in the subject string where the current
3874 character and the previous character do not both match \w or \W (i.e.
3875 one matches \w and the other matches \W), or the start or end of the
3876 string if the first or last character matches \w, respectively. In
3877 UTF-8 mode, the meanings of \w and \W can be changed by setting the
3878 PCRE_UCP option. When this is done, it also affects \b and \B. Neither
3879 PCRE nor Perl has a separate "start of word" or "end of word" metase-
3880 quence. However, whatever follows \b normally determines which it is.
3881 For example, the fragment \ba matches "a" at the start of a word.
3883 The \A, \Z, and \z assertions differ from the traditional circumflex
3884 and dollar (described in the next section) in that they only ever match
3885 at the very start and end of the subject string, whatever options are
3886 set. Thus, they are independent of multiline mode. These three asser-
3887 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
3888 affect only the behaviour of the circumflex and dollar metacharacters.
3889 However, if the startoffset argument of pcre_exec() is non-zero, indi-
3890 cating that matching is to start at a point other than the beginning of
3891 the subject, \A can never match. The difference between \Z and \z is
3892 that \Z matches before a newline at the end of the string as well as at
3893 the very end, whereas \z matches only at the end.
3895 The \G assertion is true only when the current matching position is at
3896 the start point of the match, as specified by the startoffset argument
3897 of pcre_exec(). It differs from \A when the value of startoffset is
3898 non-zero. By calling pcre_exec() multiple times with appropriate argu-
3899 ments, you can mimic Perl's /g option, and it is in this kind of imple-
3900 mentation where \G can be useful.
3902 Note, however, that PCRE's interpretation of \G, as the start of the
3903 current match, is subtly different from Perl's, which defines it as the
3904 end of the previous match. In Perl, these can be different when the
3905 previously matched string was empty. Because PCRE does just one match
3906 at a time, it cannot reproduce this behaviour.
3908 If all the alternatives of a pattern begin with \G, the expression is
3909 anchored to the starting match position, and the "anchored" flag is set
3910 in the compiled regular expression.
3915 Outside a character class, in the default matching mode, the circumflex
3916 character is an assertion that is true only if the current matching
3917 point is at the start of the subject string. If the startoffset argu-
3918 ment of pcre_exec() is non-zero, circumflex can never match if the
3919 PCRE_MULTILINE option is unset. Inside a character class, circumflex
3920 has an entirely different meaning (see below).
3922 Circumflex need not be the first character of the pattern if a number
3923 of alternatives are involved, but it should be the first thing in each
3924 alternative in which it appears if the pattern is ever to match that
3925 branch. If all possible alternatives start with a circumflex, that is,
3926 if the pattern is constrained to match only at the start of the sub-
3927 ject, it is said to be an "anchored" pattern. (There are also other
3928 constructs that can cause a pattern to be anchored.)
3930 A dollar character is an assertion that is true only if the current
3931 matching point is at the end of the subject string, or immediately
3932 before a newline at the end of the string (by default). Dollar need not
3933 be the last character of the pattern if a number of alternatives are
3934 involved, but it should be the last item in any branch in which it
3935 appears. Dollar has no special meaning in a character class.
3937 The meaning of dollar can be changed so that it matches only at the
3938 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
3939 compile time. This does not affect the \Z assertion.
3941 The meanings of the circumflex and dollar characters are changed if the
3942 PCRE_MULTILINE option is set. When this is the case, a circumflex
3943 matches immediately after internal newlines as well as at the start of
3944 the subject string. It does not match after a newline that ends the
3945 string. A dollar matches before any newlines in the string, as well as
3946 at the very end, when PCRE_MULTILINE is set. When newline is specified
3947 as the two-character sequence CRLF, isolated CR and LF characters do
3948 not indicate newlines.
3950 For example, the pattern /^abc$/ matches the subject string "def\nabc"
3951 (where \n represents a newline) in multiline mode, but not otherwise.
3952 Consequently, patterns that are anchored in single line mode because
3953 all branches start with ^ are not anchored in multiline mode, and a
3954 match for circumflex is possible when the startoffset argument of
3955 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
3956 PCRE_MULTILINE is set.
3958 Note that the sequences \A, \Z, and \z can be used to match the start
3959 and end of the subject in both modes, and if all branches of a pattern
3960 start with \A it is always anchored, whether or not PCRE_MULTILINE is
3961 set.
3966 Outside a character class, a dot in the pattern matches any one charac-
3967 ter in the subject string except (by default) a character that signi-
3968 fies the end of a line. In UTF-8 mode, the matched character may be
3969 more than one byte long.
3971 When a line ending is defined as a single character, dot never matches
3972 that character; when the two-character sequence CRLF is used, dot does
3973 not match CR if it is immediately followed by LF, but otherwise it
3974 matches all characters (including isolated CRs and LFs). When any Uni-
3975 code line endings are being recognized, dot does not match CR or LF or
3976 any of the other line ending characters.
3978 The behaviour of dot with regard to newlines can be changed. If the
3979 PCRE_DOTALL option is set, a dot matches any one character, without
3980 exception. If the two-character sequence CRLF is present in the subject
3981 string, it takes two dots to match it.
3983 The handling of dot is entirely independent of the handling of circum-
3984 flex and dollar, the only relationship being that they both involve
3985 newlines. Dot has no special meaning in a character class.
3987 The escape sequence \N behaves like a dot, except that it is not
3988 affected by the PCRE_DOTALL option. In other words, it matches any
3989 character except one that signifies the end of a line.
3994 Outside a character class, the escape sequence \C matches any one byte,
3995 both in and out of UTF-8 mode. Unlike a dot, it always matches any
3996 line-ending characters. The feature is provided in Perl in order to
3997 match individual bytes in UTF-8 mode. Because it breaks up UTF-8 char-
3998 acters into individual bytes, the rest of the string may start with a
3999 malformed UTF-8 character. For this reason, the \C escape sequence is
4000 best avoided.
4002 PCRE does not allow \C to appear in lookbehind assertions (described
4003 below), because in UTF-8 mode this would make it impossible to calcu-
4004 late the length of the lookbehind.
4009 An opening square bracket introduces a character class, terminated by a
4010 closing square bracket. A closing square bracket on its own is not spe-
4011 cial by default. However, if the PCRE_JAVASCRIPT_COMPAT option is set,
4012 a lone closing square bracket causes a compile-time error. If a closing
4013 square bracket is required as a member of the class, it should be the
4014 first data character in the class (after an initial circumflex, if
4015 present) or escaped with a backslash.
4017 A character class matches a single character in the subject. In UTF-8
4018 mode, the character may be more than one byte long. A matched character
4019 must be in the set of characters defined by the class, unless the first
4020 character in the class definition is a circumflex, in which case the
4021 subject character must not be in the set defined by the class. If a
4022 circumflex is actually required as a member of the class, ensure it is
4023 not the first character, or escape it with a backslash.
4025 For example, the character class [aeiou] matches any lower case vowel,
4026 while [^aeiou] matches any character that is not a lower case vowel.
4027 Note that a circumflex is just a convenient notation for specifying the
4028 characters that are in the class by enumerating those that are not. A
4029 class that starts with a circumflex is not an assertion; it still con-
4030 sumes a character from the subject string, and therefore it fails if
4031 the current pointer is at the end of the string.
4033 In UTF-8 mode, characters with values greater than 255 can be included
4034 in a class as a literal string of bytes, or by using the \x{ escaping
4035 mechanism.
4037 When caseless matching is set, any letters in a class represent both
4038 their upper case and lower case versions, so for example, a caseless
4039 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
4040 match "A", whereas a caseful version would. In UTF-8 mode, PCRE always
4041 understands the concept of case for characters whose values are less
4042 than 128, so caseless matching is always possible. For characters with
4043 higher values, the concept of case is supported if PCRE is compiled
4044 with Unicode property support, but not otherwise. If you want to use
4045 caseless matching in UTF8-mode for characters 128 and above, you must
4046 ensure that PCRE is compiled with Unicode property support as well as
4047 with UTF-8 support.
4049 Characters that might indicate line breaks are never treated in any
4050 special way when matching character classes, whatever line-ending
4051 sequence is in use, and whatever setting of the PCRE_DOTALL and
4052 PCRE_MULTILINE options is used. A class such as [^a] always matches one
4053 of these characters.
4055 The minus (hyphen) character can be used to specify a range of charac-
4056 ters in a character class. For example, [d-m] matches any letter
4057 between d and m, inclusive. If a minus character is required in a
4058 class, it must be escaped with a backslash or appear in a position
4059 where it cannot be interpreted as indicating a range, typically as the
4060 first or last character in the class.
4062 It is not possible to have the literal character "]" as the end charac-
4063 ter of a range. A pattern such as [W-]46] is interpreted as a class of
4064 two characters ("W" and "-") followed by a literal string "46]", so it
4065 would match "W46]" or "-46]". However, if the "]" is escaped with a
4066 backslash it is interpreted as the end of range, so [W-\]46] is inter-
4067 preted as a class containing a range followed by two other characters.
4068 The octal or hexadecimal representation of "]" can also be used to end
4069 a range.
4071 Ranges operate in the collating sequence of character values. They can
4072 also be used for characters specified numerically, for example
4073 [\000-\037]. In UTF-8 mode, ranges can include characters whose values
4074 are greater than 255, for example [\x{100}-\x{2ff}].
4076 If a range that includes letters is used when caseless matching is set,
4077 it matches the letters in either case. For example, [W-c] is equivalent
4078 to [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if
4079 character tables for a French locale are in use, [\xc8-\xcb] matches
4080 accented E characters in both cases. In UTF-8 mode, PCRE supports the
4081 concept of case for characters with values greater than 128 only when
4082 it is compiled with Unicode property support.
4084 The character escape sequences \d, \D, \h, \H, \p, \P, \s, \S, \v, \V,
4085 \w, and \W may appear in a character class, and add the characters that
4086 they match to the class. For example, [\dABCDEF] matches any hexadeci-
4087 mal digit. In UTF-8 mode, the PCRE_UCP option affects the meanings of
4088 \d, \s, \w and their upper case partners, just as it does when they
4089 appear outside a character class, as described in the section entitled
4090 "Generic character types" above. The escape sequence \b has a different
4091 meaning inside a character class; it matches the backspace character.
4092 The sequences \B, \N, \R, and \X are not special inside a character
4093 class. Like any other unrecognized escape sequences, they are treated
4094 as the literal characters "B", "N", "R", and "X" by default, but cause
4095 an error if the PCRE_EXTRA option is set.
4097 A circumflex can conveniently be used with the upper case character
4098 types to specify a more restricted set of characters than the matching
4099 lower case type. For example, the class [^\W_] matches any letter or
4100 digit, but not underscore, whereas [\w] includes underscore. A positive
4101 character class should be read as "something OR something OR ..." and a
4102 negative class as "NOT something AND NOT something AND NOT ...".
4104 The only metacharacters that are recognized in character classes are
4105 backslash, hyphen (only where it can be interpreted as specifying a
4106 range), circumflex (only at the start), opening square bracket (only
4107 when it can be interpreted as introducing a POSIX class name - see the
4108 next section), and the terminating closing square bracket. However,
4109 escaping other non-alphanumeric characters does no harm.
4114 Perl supports the POSIX notation for character classes. This uses names
4115 enclosed by [: and :] within the enclosing square brackets. PCRE also
4116 supports this notation. For example,
4118 [01[:alpha:]%]
4120 matches "0", "1", any alphabetic character, or "%". The supported class
4121 names are:
4123 alnum letters and digits
4124 alpha letters
4125 ascii character codes 0 - 127
4126 blank space or tab only
4127 cntrl control characters
4128 digit decimal digits (same as \d)
4129 graph printing characters, excluding space
4130 lower lower case letters
4131 print printing characters, including space
4132 punct printing characters, excluding letters and digits and space
4133 space white space (not quite the same as \s)
4134 upper upper case letters
4135 word "word" characters (same as \w)
4136 xdigit hexadecimal digits
4138 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
4139 and space (32). Notice that this list includes the VT character (code
4140 11). This makes "space" different to \s, which does not include VT (for
4141 Perl compatibility).
4143 The name "word" is a Perl extension, and "blank" is a GNU extension
4144 from Perl 5.8. Another Perl extension is negation, which is indicated
4145 by a ^ character after the colon. For example,
4147 [12[:^digit:]]
4149 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
4150 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
4151 these are not supported, and an error is given if they are encountered.
4153 By default, in UTF-8 mode, characters with values greater than 128 do
4154 not match any of the POSIX character classes. However, if the PCRE_UCP
4155 option is passed to pcre_compile(), some of the classes are changed so
4156 that Unicode character properties are used. This is achieved by replac-
4157 ing the POSIX classes by other sequences, as follows:
4159 [:alnum:] becomes \p{Xan}
4160 [:alpha:] becomes \p{L}
4161 [:blank:] becomes \h
4162 [:digit:] becomes \p{Nd}
4163 [:lower:] becomes \p{Ll}
4164 [:space:] becomes \p{Xps}
4165 [:upper:] becomes \p{Lu}
4166 [:word:] becomes \p{Xwd}
4168 Negated versions, such as [:^alpha:] use \P instead of \p. The other
4169 POSIX classes are unchanged, and match only characters with code points
4170 less than 128.
4175 Vertical bar characters are used to separate alternative patterns. For
4176 example, the pattern
4178 gilbert|sullivan
4180 matches either "gilbert" or "sullivan". Any number of alternatives may
4181 appear, and an empty alternative is permitted (matching the empty
4182 string). The matching process tries each alternative in turn, from left
4183 to right, and the first one that succeeds is used. If the alternatives
4184 are within a subpattern (defined below), "succeeds" means matching the
4185 rest of the main pattern as well as the alternative in the subpattern.
4190 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
4191 PCRE_EXTENDED options (which are Perl-compatible) can be changed from
4192 within the pattern by a sequence of Perl option letters enclosed
4193 between "(?" and ")". The option letters are
4195 i for PCRE_CASELESS
4197 s for PCRE_DOTALL
4198 x for PCRE_EXTENDED
4200 For example, (?im) sets caseless, multiline matching. It is also possi-
4201 ble to unset these options by preceding the letter with a hyphen, and a
4202 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
4204 is also permitted. If a letter appears both before and after the
4205 hyphen, the option is unset.
4208 can be changed in the same way as the Perl-compatible options by using
4209 the characters J, U and X respectively.
4211 When one of these option changes occurs at top level (that is, not
4212 inside subpattern parentheses), the change applies to the remainder of
4213 the pattern that follows. If the change is placed right at the start of
4214 a pattern, PCRE extracts it into the global options (and it will there-
4215 fore show up in data extracted by the pcre_fullinfo() function).
4217 An option change within a subpattern (see below for a description of
4218 subpatterns) affects only that part of the subpattern that follows it,
4219 so
4221 (a(?i)b)c
4223 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
4224 used). By this means, options can be made to have different settings
4225 in different parts of the pattern. Any changes made in one alternative
4226 do carry on into subsequent branches within the same subpattern. For
4227 example,
4229 (a(?i)b|c)
4231 matches "ab", "aB", "c", and "C", even though when matching "C" the
4232 first branch is abandoned before the option setting. This is because
4233 the effects of option settings happen at compile time. There would be
4234 some very weird behaviour otherwise.
4236 Note: There are other PCRE-specific options that can be set by the
4237 application when the compile or match functions are called. In some
4238 cases the pattern can contain special leading sequences such as (*CRLF)
4239 to override what the application has set or what has been defaulted.
4240 Details are given in the section entitled "Newline sequences" above.
4241 There are also the (*UTF8) and (*UCP) leading sequences that can be
4242 used to set UTF-8 and Unicode property modes; they are equivalent to
4243 setting the PCRE_UTF8 and the PCRE_UCP options, respectively.
4248 Subpatterns are delimited by parentheses (round brackets), which can be
4249 nested. Turning part of a pattern into a subpattern does two things:
4251 1. It localizes a set of alternatives. For example, the pattern
4253 cat(aract|erpillar|)
4255 matches "cataract", "caterpillar", or "cat". Without the parentheses,
4256 it would match "cataract", "erpillar" or an empty string.
4258 2. It sets up the subpattern as a capturing subpattern. This means
4259 that, when the whole pattern matches, that portion of the subject
4260 string that matched the subpattern is passed back to the caller via the
4261 ovector argument of pcre_exec(). Opening parentheses are counted from
4262 left to right (starting from 1) to obtain numbers for the capturing
4263 subpatterns. For example, if the string "the red king" is matched
4264 against the pattern
4266 the ((red|white) (king|queen))
4268 the captured substrings are "red king", "red", and "king", and are num-
4269 bered 1, 2, and 3, respectively.
4271 The fact that plain parentheses fulfil two functions is not always
4272 helpful. There are often times when a grouping subpattern is required
4273 without a capturing requirement. If an opening parenthesis is followed
4274 by a question mark and a colon, the subpattern does not do any captur-
4275 ing, and is not counted when computing the number of any subsequent
4276 capturing subpatterns. For example, if the string "the white queen" is
4277 matched against the pattern
4279 the ((?:red|white) (king|queen))
4281 the captured substrings are "white queen" and "queen", and are numbered
4282 1 and 2. The maximum number of capturing subpatterns is 65535.
4284 As a convenient shorthand, if any option settings are required at the
4285 start of a non-capturing subpattern, the option letters may appear
4286 between the "?" and the ":". Thus the two patterns
4288 (?i:saturday|sunday)
4289 (?:(?i)saturday|sunday)
4291 match exactly the same set of strings. Because alternative branches are
4292 tried from left to right, and options are not reset until the end of
4293 the subpattern is reached, an option setting in one branch does affect
4294 subsequent branches, so the above patterns match "SUNDAY" as well as
4295 "Saturday".
4300 Perl 5.10 introduced a feature whereby each alternative in a subpattern
4301 uses the same numbers for its capturing parentheses. Such a subpattern
4302 starts with (?| and is itself a non-capturing subpattern. For example,
4303 consider this pattern:
4305 (?|(Sat)ur|(Sun))day
4307 Because the two alternatives are inside a (?| group, both sets of cap-
4308 turing parentheses are numbered one. Thus, when the pattern matches,
4309 you can look at captured substring number one, whichever alternative
4310 matched. This construct is useful when you want to capture part, but
4311 not all, of one of a number of alternatives. Inside a (?| group, paren-
4312 theses are numbered as usual, but the number is reset at the start of
4313 each branch. The numbers of any capturing parentheses that follow the
4314 subpattern start after the highest number used in any branch. The fol-
4315 lowing example is taken from the Perl documentation. The numbers under-
4316 neath show in which buffer the captured content will be stored.
4318 # before ---------------branch-reset----------- after
4319 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
4320 # 1 2 2 3 2 3 4
4322 A back reference to a numbered subpattern uses the most recent value
4323 that is set for that number by any subpattern. The following pattern
4324 matches "abcabc" or "defdef":
4326 /(?|(abc)|(def))\1/
4328 In contrast, a recursive or "subroutine" call to a numbered subpattern
4329 always refers to the first one in the pattern with the given number.
4330 The following pattern matches "abcabc" or "defabc":
4332 /(?|(abc)|(def))(?1)/
4334 If a condition test for a subpattern's having matched refers to a non-
4335 unique number, the test is true if any of the subpatterns of that num-
4336 ber have matched.
4338 An alternative approach to using this "branch reset" feature is to use
4339 duplicate named subpatterns, as described in the next section.
4344 Identifying capturing parentheses by number is simple, but it can be
4345 very hard to keep track of the numbers in complicated regular expres-
4346 sions. Furthermore, if an expression is modified, the numbers may
4347 change. To help with this difficulty, PCRE supports the naming of sub-
4348 patterns. This feature was not added to Perl until release 5.10. Python
4349 had the feature earlier, and PCRE introduced it at release 4.0, using
4350 the Python syntax. PCRE now supports both the Perl and the Python syn-
4351 tax. Perl allows identically numbered subpatterns to have different
4352 names, but PCRE does not.
4354 In PCRE, a subpattern can be named in one of three ways: (?<name>...)
4355 or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
4356 to capturing parentheses from other parts of the pattern, such as back
4357 references, recursion, and conditions, can be made by name as well as
4358 by number.
4360 Names consist of up to 32 alphanumeric characters and underscores.
4361 Named capturing parentheses are still allocated numbers as well as
4362 names, exactly as if the names were not present. The PCRE API provides
4363 function calls for extracting the name-to-number translation table from
4364 a compiled pattern. There is also a convenience function for extracting
4365 a captured substring by name.
4367 By default, a name must be unique within a pattern, but it is possible
4368 to relax this constraint by setting the PCRE_DUPNAMES option at compile
4369 time. (Duplicate names are also always permitted for subpatterns with
4370 the same number, set up as described in the previous section.) Dupli-
4371 cate names can be useful for patterns where only one instance of the
4372 named parentheses can match. Suppose you want to match the name of a
4373 weekday, either as a 3-letter abbreviation or as the full name, and in
4374 both cases you want to extract the abbreviation. This pattern (ignoring
4375 the line breaks) does the job:
4377 (?<DN>Mon|Fri|Sun)(?:day)?|
4378 (?<DN>Tue)(?:sday)?|
4379 (?<DN>Wed)(?:nesday)?|
4380 (?<DN>Thu)(?:rsday)?|
4381 (?<DN>Sat)(?:urday)?
4383 There are five capturing substrings, but only one is ever set after a
4384 match. (An alternative way of solving this problem is to use a "branch
4385 reset" subpattern, as described in the previous section.)
4387 The convenience function for extracting the data by name returns the
4388 substring for the first (and in this example, the only) subpattern of
4389 that name that matched. This saves searching to find which numbered
4390 subpattern it was.
4392 If you make a back reference to a non-unique named subpattern from
4393 elsewhere in the pattern, the one that corresponds to the first occur-
4394 rence of the name is used. In the absence of duplicate numbers (see the
4395 previous section) this is the one with the lowest number. If you use a
4396 named reference in a condition test (see the section about conditions
4397 below), either to check whether a subpattern has matched, or to check
4398 for recursion, all subpatterns with the same name are tested. If the
4399 condition is true for any one of them, the overall condition is true.
4400 This is the same behaviour as testing by number. For further details of
4401 the interfaces for handling named subpatterns, see the pcreapi documen-
4402 tation.
4404 Warning: You cannot use different names to distinguish between two sub-
4405 patterns with the same number because PCRE uses only the numbers when
4406 matching. For this reason, an error is given at compile time if differ-
4407 ent names are given to subpatterns with the same number. However, you
4408 can give the same name to subpatterns with the same number, even when
4409 PCRE_DUPNAMES is not set.
4414 Repetition is specified by quantifiers, which can follow any of the
4415 following items:
4417 a literal data character
4418 the dot metacharacter
4419 the \C escape sequence
4420 the \X escape sequence (in UTF-8 mode with Unicode properties)
4421 the \R escape sequence
4422 an escape such as \d or \pL that matches a single character
4423 a character class
4424 a back reference (see next section)
4425 a parenthesized subpattern (including assertions)
4426 a recursive or "subroutine" call to a subpattern
4428 The general repetition quantifier specifies a minimum and maximum num-
4429 ber of permitted matches, by giving the two numbers in curly brackets
4430 (braces), separated by a comma. The numbers must be less than 65536,
4431 and the first must be less than or equal to the second. For example:
4433 z{2,4}
4435 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
4436 special character. If the second number is omitted, but the comma is
4437 present, there is no upper limit; if the second number and the comma
4438 are both omitted, the quantifier specifies an exact number of required
4439 matches. Thus
4441 [aeiou]{3,}
4443 matches at least 3 successive vowels, but may match many more, while
4445 \d{8}
4447 matches exactly 8 digits. An opening curly bracket that appears in a
4448 position where a quantifier is not allowed, or one that does not match
4449 the syntax of a quantifier, is taken as a literal character. For exam-
4450 ple, {,6} is not a quantifier, but a literal string of four characters.
4452 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to
4453 individual bytes. Thus, for example, \x{100}{2} matches two UTF-8 char-
4454 acters, each of which is represented by a two-byte sequence. Similarly,
4455 when Unicode property support is available, \X{3} matches three Unicode
4456 extended sequences, each of which may be several bytes long (and they
4457 may be of different lengths).
4459 The quantifier {0} is permitted, causing the expression to behave as if
4460 the previous item and the quantifier were not present. This may be use-
4461 ful for subpatterns that are referenced as subroutines from elsewhere
4462 in the pattern (but see also the section entitled "Defining subpatterns
4463 for use by reference only" below). Items other than subpatterns that
4464 have a {0} quantifier are omitted from the compiled pattern.
4466 For convenience, the three most common quantifiers have single-charac-
4467 ter abbreviations:
4469 * is equivalent to {0,}
4470 + is equivalent to {1,}
4471 ? is equivalent to {0,1}
4473 It is possible to construct infinite loops by following a subpattern
4474 that can match no characters with a quantifier that has no upper limit,
4475 for example:
4477 (a?)*
4479 Earlier versions of Perl and PCRE used to give an error at compile time
4480 for such patterns. However, because there are cases where this can be
4481 useful, such patterns are now accepted, but if any repetition of the
4482 subpattern does in fact match no characters, the loop is forcibly bro-
4483 ken.
4485 By default, the quantifiers are "greedy", that is, they match as much
4486 as possible (up to the maximum number of permitted times), without
4487 causing the rest of the pattern to fail. The classic example of where
4488 this gives problems is in trying to match comments in C programs. These
4489 appear between /* and */ and within the comment, individual * and /
4490 characters may appear. An attempt to match C comments by applying the
4491 pattern
4493 /\*.*\*/
4495 to the string
4497 /* first comment */ not comment /* second comment */
4499 fails, because it matches the entire string owing to the greediness of
4500 the .* item.
4502 However, if a quantifier is followed by a question mark, it ceases to
4503 be greedy, and instead matches the minimum number of times possible, so
4504 the pattern
4506 /\*.*?\*/
4508 does the right thing with the C comments. The meaning of the various
4509 quantifiers is not otherwise changed, just the preferred number of
4510 matches. Do not confuse this use of question mark with its use as a
4511 quantifier in its own right. Because it has two uses, it can sometimes
4512 appear doubled, as in
4514 \d??\d
4516 which matches one digit by preference, but can match two if that is the
4517 only way the rest of the pattern matches.
4519 If the PCRE_UNGREEDY option is set (an option that is not available in
4520 Perl), the quantifiers are not greedy by default, but individual ones
4521 can be made greedy by following them with a question mark. In other
4522 words, it inverts the default behaviour.
4524 When a parenthesized subpattern is quantified with a minimum repeat
4525 count that is greater than 1 or with a limited maximum, more memory is
4526 required for the compiled pattern, in proportion to the size of the
4527 minimum or maximum.
4529 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
4530 alent to Perl's /s) is set, thus allowing the dot to match newlines,
4531 the pattern is implicitly anchored, because whatever follows will be
4532 tried against every character position in the subject string, so there
4533 is no point in retrying the overall match at any position after the
4534 first. PCRE normally treats such a pattern as though it were preceded
4535 by \A.
4537 In cases where it is known that the subject string contains no new-
4538 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
4539 mization, or alternatively using ^ to indicate anchoring explicitly.
4541 However, there is one situation where the optimization cannot be used.
4542 When .* is inside capturing parentheses that are the subject of a back
4543 reference elsewhere in the pattern, a match at the start may fail where
4544 a later one succeeds. Consider, for example:
4546 (.*)abc\1
4548 If the subject is "xyz123abc123" the match point is the fourth charac-
4549 ter. For this reason, such a pattern is not implicitly anchored.
4551 When a capturing subpattern is repeated, the value captured is the sub-
4552 string that matched the final iteration. For example, after
4554 (tweedle[dume]{3}\s*)+
4556 has matched "tweedledum tweedledee" the value of the captured substring
4557 is "tweedledee". However, if there are nested capturing subpatterns,
4558 the corresponding captured values may have been set in previous itera-
4559 tions. For example, after
4561 /(a|(b))+/
4563 matches "aba" the value of the second captured substring is "b".
4568 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
4569 repetition, failure of what follows normally causes the repeated item
4570 to be re-evaluated to see if a different number of repeats allows the
4571 rest of the pattern to match. Sometimes it is useful to prevent this,
4572 either to change the nature of the match, or to cause it fail earlier
4573 than it otherwise might, when the author of the pattern knows there is
4574 no point in carrying on.
4576 Consider, for example, the pattern \d+foo when applied to the subject
4577 line
4579 123456bar
4581 After matching all 6 digits and then failing to match "foo", the normal
4582 action of the matcher is to try again with only 5 digits matching the
4583 \d+ item, and then with 4, and so on, before ultimately failing.
4584 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
4585 the means for specifying that once a subpattern has matched, it is not
4586 to be re-evaluated in this way.
4588 If we use atomic grouping for the previous example, the matcher gives
4589 up immediately on failing to match "foo" the first time. The notation
4590 is a kind of special parenthesis, starting with (?> as in this example:
4592 (?>\d+)foo
4594 This kind of parenthesis "locks up" the part of the pattern it con-
4595 tains once it has matched, and a failure further into the pattern is
4596 prevented from backtracking into it. Backtracking past it to previous
4597 items, however, works as normal.
4599 An alternative description is that a subpattern of this type matches
4600 the string of characters that an identical standalone pattern would
4601 match, if anchored at the current point in the subject string.
4603 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
4604 such as the above example can be thought of as a maximizing repeat that
4605 must swallow everything it can. So, while both \d+ and \d+? are pre-
4606 pared to adjust the number of digits they match in order to make the
4607 rest of the pattern match, (?>\d+) can only match an entire sequence of
4608 digits.
4610 Atomic groups in general can of course contain arbitrarily complicated
4611 subpatterns, and can be nested. However, when the subpattern for an
4612 atomic group is just a single repeated item, as in the example above, a
4613 simpler notation, called a "possessive quantifier" can be used. This
4614 consists of an additional + character following a quantifier. Using
4615 this notation, the previous example can be rewritten as
4617 \d++foo
4619 Note that a possessive quantifier can be used with an entire group, for
4620 example:
4622 (abc|xyz){2,3}+
4624 Possessive quantifiers are always greedy; the setting of the
4625 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
4626 simpler forms of atomic group. However, there is no difference in the
4627 meaning of a possessive quantifier and the equivalent atomic group,
4628 though there may be a performance difference; possessive quantifiers
4629 should be slightly faster.
4631 The possessive quantifier syntax is an extension to the Perl 5.8 syn-
4632 tax. Jeffrey Friedl originated the idea (and the name) in the first
4633 edition of his book. Mike McCloskey liked it, so implemented it when he
4634 built Sun's Java package, and PCRE copied it from there. It ultimately
4635 found its way into Perl at release 5.10.
4637 PCRE has an optimization that automatically "possessifies" certain sim-
4638 ple pattern constructs. For example, the sequence A+B is treated as
4639 A++B because there is no point in backtracking into a sequence of A's
4640 when B must follow.
4642 When a pattern contains an unlimited repeat inside a subpattern that
4643 can itself be repeated an unlimited number of times, the use of an
4644 atomic group is the only way to avoid some failing matches taking a
4645 very long time indeed. The pattern
4647 (\D+|<\d+>)*[!?]
4649 matches an unlimited number of substrings that either consist of non-
4650 digits, or digits enclosed in <>, followed by either ! or ?. When it
4651 matches, it runs quickly. However, if it is applied to
4653 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
4655 it takes a long time before reporting failure. This is because the
4656 string can be divided between the internal \D+ repeat and the external
4657 * repeat in a large number of ways, and all have to be tried. (The
4658 example uses [!?] rather than a single character at the end, because
4659 both PCRE and Perl have an optimization that allows for fast failure
4660 when a single character is used. They remember the last single charac-
4661 ter that is required for a match, and fail early if it is not present
4662 in the string.) If the pattern is changed so that it uses an atomic
4663 group, like this:
4665 ((?>\D+)|<\d+>)*[!?]
4667 sequences of non-digits cannot be broken, and failure happens quickly.
4672 Outside a character class, a backslash followed by a digit greater than
4673 0 (and possibly further digits) is a back reference to a capturing sub-
4674 pattern earlier (that is, to its left) in the pattern, provided there
4675 have been that many previous capturing left parentheses.
4677 However, if the decimal number following the backslash is less than 10,
4678 it is always taken as a back reference, and causes an error only if
4679 there are not that many capturing left parentheses in the entire pat-
4680 tern. In other words, the parentheses that are referenced need not be
4681 to the left of the reference for numbers less than 10. A "forward back
4682 reference" of this type can make sense when a repetition is involved
4683 and the subpattern to the right has participated in an earlier itera-
4684 tion.
4686 It is not possible to have a numerical "forward back reference" to a
4687 subpattern whose number is 10 or more using this syntax because a
4688 sequence such as \50 is interpreted as a character defined in octal.
4689 See the subsection entitled "Non-printing characters" above for further
4690 details of the handling of digits following a backslash. There is no
4691 such problem when named parentheses are used. A back reference to any
4692 subpattern is possible using named parentheses (see below).
4694 Another way of avoiding the ambiguity inherent in the use of digits
4695 following a backslash is to use the \g escape sequence. This escape
4696 must be followed by an unsigned number or a negative number, optionally
4697 enclosed in braces. These examples are all identical:
4699 (ring), \1
4700 (ring), \g1
4701 (ring), \g{1}
4703 An unsigned number specifies an absolute reference without the ambigu-
4704 ity that is present in the older syntax. It is also useful when literal
4705 digits follow the reference. A negative number is a relative reference.
4706 Consider this example:
4708 (abc(def)ghi)\g{-1}
4710 The sequence \g{-1} is a reference to the most recently started captur-
4711 ing subpattern before \g, that is, is it equivalent to \2 in this exam-
4712 ple. Similarly, \g{-2} would be equivalent to \1. The use of relative
4713 references can be helpful in long patterns, and also in patterns that
4714 are created by joining together fragments that contain references
4715 within themselves.
4717 A back reference matches whatever actually matched the capturing sub-
4718 pattern in the current subject string, rather than anything matching
4719 the subpattern itself (see "Subpatterns as subroutines" below for a way
4720 of doing that). So the pattern
4722 (sens|respons)e and \1ibility
4724 matches "sense and sensibility" and "response and responsibility", but
4725 not "sense and responsibility". If caseful matching is in force at the
4726 time of the back reference, the case of letters is relevant. For exam-
4727 ple,
4729 ((?i)rah)\s+\1
4731 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
4732 original capturing subpattern is matched caselessly.
4734 There are several different ways of writing back references to named
4735 subpatterns. The .NET syntax \k{name} and the Perl syntax \k<name> or
4736 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's
4737 unified back reference syntax, in which \g can be used for both numeric
4738 and named references, is also supported. We could rewrite the above
4739 example in any of the following ways:
4741 (?<p1>(?i)rah)\s+\k<p1>
4742 (?'p1'(?i)rah)\s+\k{p1}
4743 (?P<p1>(?i)rah)\s+(?P=p1)
4744 (?<p1>(?i)rah)\s+\g{p1}
4746 A subpattern that is referenced by name may appear in the pattern
4747 before or after the reference.
4749 There may be more than one back reference to the same subpattern. If a
4750 subpattern has not actually been used in a particular match, any back
4751 references to it always fail by default. For example, the pattern
4753 (a|(bc))\2
4755 always fails if it starts to match "a" rather than "bc". However, if
4756 the PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back refer-
4757 ence to an unset value matches an empty string.
4759 Because there may be many capturing parentheses in a pattern, all dig-
4760 its following a backslash are taken as part of a potential back refer-
4761 ence number. If the pattern continues with a digit character, some
4762 delimiter must be used to terminate the back reference. If the
4763 PCRE_EXTENDED option is set, this can be whitespace. Otherwise, the \g{
4764 syntax or an empty comment (see "Comments" below) can be used.
4766 Recursive back references
4768 A back reference that occurs inside the parentheses to which it refers
4769 fails when the subpattern is first used, so, for example, (a\1) never
4770 matches. However, such references can be useful inside repeated sub-
4771 patterns. For example, the pattern
4773 (a|b\1)+
4775 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
4776 ation of the subpattern, the back reference matches the character
4777 string corresponding to the previous iteration. In order for this to
4778 work, the pattern must be such that the first iteration does not need
4779 to match the back reference. This can be done using alternation, as in
4780 the example above, or by a quantifier with a minimum of zero.
4782 Back references of this type cause the group that they reference to be
4783 treated as an atomic group. Once the whole group has been matched, a
4784 subsequent matching failure cannot cause backtracking into the middle
4785 of the group.
4790 An assertion is a test on the characters following or preceding the
4791 current matching point that does not actually consume any characters.
4792 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
4793 described above.
4795 More complicated assertions are coded as subpatterns. There are two
4796 kinds: those that look ahead of the current position in the subject
4797 string, and those that look behind it. An assertion subpattern is
4798 matched in the normal way, except that it does not cause the current
4799 matching position to be changed.
4801 Assertion subpatterns are not capturing subpatterns. If such an asser-
4802 tion contains capturing subpatterns within it, these are counted for
4803 the purposes of numbering the capturing subpatterns in the whole pat-
4804 tern. However, substring capturing is carried out only for positive
4805 assertions, because it does not make sense for negative assertions.
4807 For compatibility with Perl, assertion subpatterns may be repeated;
4808 though it makes no sense to assert the same thing several times, the
4809 side effect of capturing parentheses may occasionally be useful. In
4810 practice, there only three cases:
4812 (1) If the quantifier is {0}, the assertion is never obeyed during
4813 matching. However, it may contain internal capturing parenthesized
4814 groups that are called from elsewhere via the subroutine mechanism.
4816 (2) If quantifier is {0,n} where n is greater than zero, it is treated
4817 as if it were {0,1}. At run time, the rest of the pattern match is
4818 tried with and without the assertion, the order depending on the greed-
4819 iness of the quantifier.
4821 (3) If the minimum repetition is greater than zero, the quantifier is
4822 ignored. The assertion is obeyed just once when encountered during
4823 matching.
4825 Lookahead assertions
4827 Lookahead assertions start with (?= for positive assertions and (?! for
4828 negative assertions. For example,
4830 \w+(?=;)
4832 matches a word followed by a semicolon, but does not include the semi-
4833 colon in the match, and
4835 foo(?!bar)
4837 matches any occurrence of "foo" that is not followed by "bar". Note
4838 that the apparently similar pattern
4840 (?!foo)bar
4842 does not find an occurrence of "bar" that is preceded by something
4843 other than "foo"; it finds any occurrence of "bar" whatsoever, because
4844 the assertion (?!foo) is always true when the next three characters are
4845 "bar". A lookbehind assertion is needed to achieve the other effect.
4847 If you want to force a matching failure at some point in a pattern, the
4848 most convenient way to do it is with (?!) because an empty string
4849 always matches, so an assertion that requires there not to be an empty
4850 string must always fail. The backtracking control verb (*FAIL) or (*F)
4851 is a synonym for (?!).
4853 Lookbehind assertions
4855 Lookbehind assertions start with (?<= for positive assertions and (?<!
4856 for negative assertions. For example,
4858 (?<!foo)bar
4860 does find an occurrence of "bar" that is not preceded by "foo". The
4861 contents of a lookbehind assertion are restricted such that all the
4862 strings it matches must have a fixed length. However, if there are sev-
4863 eral top-level alternatives, they do not all have to have the same
4864 fixed length. Thus
4866 (?<=bullock|donkey)
4868 is permitted, but
4870 (?<!dogs?|cats?)
4872 causes an error at compile time. Branches that match different length
4873 strings are permitted only at the top level of a lookbehind assertion.
4874 This is an extension compared with Perl, which requires all branches to
4875 match the same length of string. An assertion such as
4877 (?<=ab(c|de))
4879 is not permitted, because its single top-level branch can match two
4880 different lengths, but it is acceptable to PCRE if rewritten to use two
4881 top-level branches:
4883 (?<=abc|abde)
4885 In some cases, the escape sequence \K (see above) can be used instead
4886 of a lookbehind assertion to get round the fixed-length restriction.
4888 The implementation of lookbehind assertions is, for each alternative,
4889 to temporarily move the current position back by the fixed length and
4890 then try to match. If there are insufficient characters before the cur-
4891 rent position, the assertion fails.
4893 PCRE does not allow the \C escape (which matches a single byte in UTF-8
4894 mode) to appear in lookbehind assertions, because it makes it impossi-
4895 ble to calculate the length of the lookbehind. The \X and \R escapes,
4896 which can match different numbers of bytes, are also not permitted.
4898 "Subroutine" calls (see below) such as (?2) or (?&X) are permitted in
4899 lookbehinds, as long as the subpattern matches a fixed-length string.
4900 Recursion, however, is not supported.
4902 Possessive quantifiers can be used in conjunction with lookbehind
4903 assertions to specify efficient matching of fixed-length strings at the
4904 end of subject strings. Consider a simple pattern such as
4906 abcd$
4908 when applied to a long string that does not match. Because matching
4909 proceeds from left to right, PCRE will look for each "a" in the subject
4910 and then see if what follows matches the rest of the pattern. If the
4911 pattern is specified as
4913 ^.*abcd$
4915 the initial .* matches the entire string at first, but when this fails
4916 (because there is no following "a"), it backtracks to match all but the
4917 last character, then all but the last two characters, and so on. Once
4918 again the search for "a" covers the entire string, from right to left,
4919 so we are no better off. However, if the pattern is written as
4921 ^.*+(?<=abcd)
4923 there can be no backtracking for the .*+ item; it can match only the
4924 entire string. The subsequent lookbehind assertion does a single test
4925 on the last four characters. If it fails, the match fails immediately.
4926 For long strings, this approach makes a significant difference to the
4927 processing time.
4929 Using multiple assertions
4931 Several assertions (of any sort) may occur in succession. For example,
4933 (?<=\d{3})(?<!999)foo
4935 matches "foo" preceded by three digits that are not "999". Notice that
4936 each of the assertions is applied independently at the same point in
4937 the subject string. First there is a check that the previous three
4938 characters are all digits, and then there is a check that the same
4939 three characters are not "999". This pattern does not match "foo" pre-
4940 ceded by six characters, the first of which are digits and the last
4941 three of which are not "999". For example, it doesn't match "123abc-
4942 foo". A pattern to do that is
4944 (?<=\d{3}...)(?<!999)foo
4946 This time the first assertion looks at the preceding six characters,
4947 checking that the first three are digits, and then the second assertion
4948 checks that the preceding three characters are not "999".
4950 Assertions can be nested in any combination. For example,
4952 (?<=(?<!foo)bar)baz
4954 matches an occurrence of "baz" that is preceded by "bar" which in turn
4955 is not preceded by "foo", while
4957 (?<=\d{3}(?!999)...)foo
4959 is another pattern that matches "foo" preceded by three digits and any
4960 three characters that are not "999".
4965 It is possible to cause the matching process to obey a subpattern con-
4966 ditionally or to choose between two alternative subpatterns, depending
4967 on the result of an assertion, or whether a specific capturing subpat-
4968 tern has already been matched. The two possible forms of conditional
4969 subpattern are:
4971 (?(condition)yes-pattern)
4972 (?(condition)yes-pattern|no-pattern)
4974 If the condition is satisfied, the yes-pattern is used; otherwise the
4975 no-pattern (if present) is used. If there are more than two alterna-
4976 tives in the subpattern, a compile-time error occurs. Each of the two
4977 alternatives may itself contain nested subpatterns of any form, includ-
4978 ing conditional subpatterns; the restriction to two alternatives
4979 applies only at the level of the condition. This pattern fragment is an
4980 example where the alternatives are complex:
4982 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
4985 There are four kinds of condition: references to subpatterns, refer-
4986 ences to recursion, a pseudo-condition called DEFINE, and assertions.
4988 Checking for a used subpattern by number
4990 If the text between the parentheses consists of a sequence of digits,
4991 the condition is true if a capturing subpattern of that number has pre-
4992 viously matched. If there is more than one capturing subpattern with
4993 the same number (see the earlier section about duplicate subpattern
4994 numbers), the condition is true if any of them have matched. An alter-
4995 native notation is to precede the digits with a plus or minus sign. In
4996 this case, the subpattern number is relative rather than absolute. The
4997 most recently opened parentheses can be referenced by (?(-1), the next
4998 most recent by (?(-2), and so on. Inside loops it can also make sense
4999 to refer to subsequent groups. The next parentheses to be opened can be
5000 referenced as (?(+1), and so on. (The value zero in any of these forms
5001 is not used; it provokes a compile-time error.)
5003 Consider the following pattern, which contains non-significant white
5004 space to make it more readable (assume the PCRE_EXTENDED option) and to
5005 divide it into three parts for ease of discussion:
5007 ( \( )? [^()]+ (?(1) \) )
5009 The first part matches an optional opening parenthesis, and if that
5010 character is present, sets it as the first captured substring. The sec-
5011 ond part matches one or more characters that are not parentheses. The
5012 third part is a conditional subpattern that tests whether or not the
5013 first set of parentheses matched. If they did, that is, if subject
5014 started with an opening parenthesis, the condition is true, and so the
5015 yes-pattern is executed and a closing parenthesis is required. Other-
5016 wise, since no-pattern is not present, the subpattern matches nothing.
5017 In other words, this pattern matches a sequence of non-parentheses,
5018 optionally enclosed in parentheses.
5020 If you were embedding this pattern in a larger one, you could use a
5021 relative reference:
5023 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
5025 This makes the fragment independent of the parentheses in the larger
5026 pattern.
5028 Checking for a used subpattern by name
5030 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
5031 used subpattern by name. For compatibility with earlier versions of
5032 PCRE, which had this facility before Perl, the syntax (?(name)...) is
5033 also recognized. However, there is a possible ambiguity with this syn-
5034 tax, because subpattern names may consist entirely of digits. PCRE
5035 looks first for a named subpattern; if it cannot find one and the name
5036 consists entirely of digits, PCRE looks for a subpattern of that num-
5037 ber, which must be greater than zero. Using subpattern names that con-
5038 sist entirely of digits is not recommended.
5040 Rewriting the above example to use a named subpattern gives this:
5042 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
5044 If the name used in a condition of this kind is a duplicate, the test
5045 is applied to all subpatterns of the same name, and is true if any one
5046 of them has matched.
5048 Checking for pattern recursion
5050 If the condition is the string (R), and there is no subpattern with the
5051 name R, the condition is true if a recursive call to the whole pattern
5052 or any subpattern has been made. If digits or a name preceded by amper-
5053 sand follow the letter R, for example:
5055 (?(R3)...) or (?(R&name)...)
5057 the condition is true if the most recent recursion is into a subpattern
5058 whose number or name is given. This condition does not check the entire
5059 recursion stack. If the name used in a condition of this kind is a
5060 duplicate, the test is applied to all subpatterns of the same name, and
5061 is true if any one of them is the most recent recursion.
5063 At "top level", all these recursion test conditions are false. The
5064 syntax for recursive patterns is described below.
5066 Defining subpatterns for use by reference only
5068 If the condition is the string (DEFINE), and there is no subpattern
5069 with the name DEFINE, the condition is always false. In this case,
5070 there may be only one alternative in the subpattern. It is always
5071 skipped if control reaches this point in the pattern; the idea of
5072 DEFINE is that it can be used to define "subroutines" that can be ref-
5073 erenced from elsewhere. (The use of "subroutines" is described below.)
5074 For example, a pattern to match an IPv4 address such as
5075 "" could be written like this (ignore whitespace and line
5076 breaks):
5078 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
5079 \b (?&byte) (\.(?&byte)){3} \b
5081 The first part of the pattern is a DEFINE group inside which a another
5082 group named "byte" is defined. This matches an individual component of
5083 an IPv4 address (a number less than 256). When matching takes place,
5084 this part of the pattern is skipped because DEFINE acts like a false
5085 condition. The rest of the pattern uses references to the named group
5086 to match the four dot-separated components of an IPv4 address, insist-
5087 ing on a word boundary at each end.
5089 Assertion conditions
5091 If the condition is not in any of the above formats, it must be an
5092 assertion. This may be a positive or negative lookahead or lookbehind
5093 assertion. Consider this pattern, again containing non-significant
5094 white space, and with the two alternatives on the second line:
5096 (?(?=[^a-z]*[a-z])
5097 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
5099 The condition is a positive lookahead assertion that matches an
5100 optional sequence of non-letters followed by a letter. In other words,
5101 it tests for the presence of at least one letter in the subject. If a
5102 letter is found, the subject is matched against the first alternative;
5103 otherwise it is matched against the second. This pattern matches
5104 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
5105 letters and dd are digits.
5110 There are two ways of including comments in patterns that are processed
5111 by PCRE. In both cases, the start of the comment must not be in a char-
5112 acter class, nor in the middle of any other sequence of related charac-
5113 ters such as (?: or a subpattern name or number. The characters that
5114 make up a comment play no part in the pattern matching.
5116 The sequence (?# marks the start of a comment that continues up to the
5117 next closing parenthesis. Nested parentheses are not permitted. If the
5118 PCRE_EXTENDED option is set, an unescaped # character also introduces a
5119 comment, which in this case continues to immediately after the next
5120 newline character or character sequence in the pattern. Which charac-
5121 ters are interpreted as newlines is controlled by the options passed to
5122 pcre_compile() or by a special sequence at the start of the pattern, as
5123 described in the section entitled "Newline conventions" above. Note
5124 that the end of this type of comment is a literal newline sequence in
5125 the pattern; escape sequences that happen to represent a newline do not
5126 count. For example, consider this pattern when PCRE_EXTENDED is set,
5127 and the default newline convention is in force:
5129 abc #comment \n still comment
5131 On encountering the # character, pcre_compile() skips along, looking
5132 for a newline in the pattern. The sequence \n is still literal at this
5133 stage, so it does not terminate the comment. Only an actual character
5134 with the code value 0x0a (the default newline) does so.
5139 Consider the problem of matching a string in parentheses, allowing for
5140 unlimited nested parentheses. Without the use of recursion, the best
5141 that can be done is to use a pattern that matches up to some fixed
5142 depth of nesting. It is not possible to handle an arbitrary nesting
5143 depth.
5145 For some time, Perl has provided a facility that allows regular expres-
5146 sions to recurse (amongst other things). It does this by interpolating
5147 Perl code in the expression at run time, and the code can refer to the
5148 expression itself. A Perl pattern using code interpolation to solve the
5149 parentheses problem can be created like this:
5151 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
5153 The (?p{...}) item interpolates Perl code at run time, and in this case
5154 refers recursively to the pattern in which it appears.
5156 Obviously, PCRE cannot support the interpolation of Perl code. Instead,
5157 it supports special syntax for recursion of the entire pattern, and
5158 also for individual subpattern recursion. After its introduction in
5159 PCRE and Python, this kind of recursion was subsequently introduced
5160 into Perl at release 5.10.
5162 A special item that consists of (? followed by a number greater than
5163 zero and a closing parenthesis is a recursive call of the subpattern of
5164 the given number, provided that it occurs inside that subpattern. (If
5165 not, it is a "subroutine" call, which is described in the next sec-
5166 tion.) The special item (?R) or (?0) is a recursive call of the entire
5167 regular expression.
5169 This PCRE pattern solves the nested parentheses problem (assume the
5170 PCRE_EXTENDED option is set so that white space is ignored):
5172 \( ( [^()]++ | (?R) )* \)
5174 First it matches an opening parenthesis. Then it matches any number of
5175 substrings which can either be a sequence of non-parentheses, or a
5176 recursive match of the pattern itself (that is, a correctly parenthe-
5177 sized substring). Finally there is a closing parenthesis. Note the use
5178 of a possessive quantifier to avoid backtracking into sequences of non-
5179 parentheses.
5181 If this were part of a larger pattern, you would not want to recurse
5182 the entire pattern, so instead you could use this:
5184 ( \( ( [^()]++ | (?1) )* \) )
5186 We have put the pattern into parentheses, and caused the recursion to
5187 refer to them instead of the whole pattern.
5189 In a larger pattern, keeping track of parenthesis numbers can be
5190 tricky. This is made easier by the use of relative references. Instead
5191 of (?1) in the pattern above you can write (?-2) to refer to the second
5192 most recently opened parentheses preceding the recursion. In other
5193 words, a negative number counts capturing parentheses leftwards from
5194 the point at which it is encountered.
5196 It is also possible to refer to subsequently opened parentheses, by
5197 writing references such as (?+2). However, these cannot be recursive
5198 because the reference is not inside the parentheses that are refer-
5199 enced. They are always "subroutine" calls, as described in the next
5200 section.
5202 An alternative approach is to use named parentheses instead. The Perl
5203 syntax for this is (?&name); PCRE's earlier syntax (?P>name) is also
5204 supported. We could rewrite the above example as follows:
5206 (?<pn> \( ( [^()]++ | (?&pn) )* \) )
5208 If there is more than one subpattern with the same name, the earliest
5209 one is used.
5211 This particular example pattern that we have been looking at contains
5212 nested unlimited repeats, and so the use of a possessive quantifier for
5213 matching strings of non-parentheses is important when applying the pat-
5214 tern to strings that do not match. For example, when this pattern is
5215 applied to
5217 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
5219 it yields "no match" quickly. However, if a possessive quantifier is
5220 not used, the match runs for a very long time indeed because there are
5221 so many different ways the + and * repeats can carve up the subject,
5222 and all have to be tested before failure can be reported.
5224 At the end of a match, the values of capturing parentheses are those
5225 from the outermost level. If you want to obtain intermediate values, a
5226 callout function can be used (see below and the pcrecallout documenta-
5227 tion). If the pattern above is matched against
5229 (ab(cd)ef)
5231 the value for the inner capturing parentheses (numbered 2) is "ef",
5232 which is the last value taken on at the top level. If a capturing sub-
5233 pattern is not matched at the top level, its final value is unset, even
5234 if it is (temporarily) set at a deeper level.
5236 If there are more than 15 capturing parentheses in a pattern, PCRE has
5237 to obtain extra memory to store data during a recursion, which it does
5238 by using pcre_malloc, freeing it via pcre_free afterwards. If no memory
5239 can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
5241 Do not confuse the (?R) item with the condition (R), which tests for
5242 recursion. Consider this pattern, which matches text in angle brack-
5243 ets, allowing for arbitrary nesting. Only digits are allowed in nested
5244 brackets (that is, when recursing), whereas any characters are permit-
5245 ted at the outer level.
5247 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
5249 In this pattern, (?(R) is the start of a conditional subpattern, with
5250 two different alternatives for the recursive and non-recursive cases.
5251 The (?R) item is the actual recursive call.
5253 Recursion difference from Perl
5255 In PCRE (like Python, but unlike Perl), a recursive subpattern call is
5256 always treated as an atomic group. That is, once it has matched some of
5257 the subject string, it is never re-entered, even if it contains untried
5258 alternatives and there is a subsequent matching failure. This can be
5259 illustrated by the following pattern, which purports to match a palin-
5260 dromic string that contains an odd number of characters (for example,
5261 "a", "aba", "abcba", "abcdcba"):
5263 ^(.|(.)(?1)\2)$
5265 The idea is that it either matches a single character, or two identical
5266 characters surrounding a sub-palindrome. In Perl, this pattern works;
5267 in PCRE it does not if the pattern is longer than three characters.
5268 Consider the subject string "abcba":
5270 At the top level, the first character is matched, but as it is not at
5271 the end of the string, the first alternative fails; the second alterna-
5272 tive is taken and the recursion kicks in. The recursive call to subpat-
5273 tern 1 successfully matches the next character ("b"). (Note that the
5274 beginning and end of line tests are not part of the recursion).
5276 Back at the top level, the next character ("c") is compared with what
5277 subpattern 2 matched, which was "a". This fails. Because the recursion
5278 is treated as an atomic group, there are now no backtracking points,
5279 and so the entire match fails. (Perl is able, at this point, to re-
5280 enter the recursion and try the second alternative.) However, if the
5281 pattern is written with the alternatives in the other order, things are
5282 different:
5284 ^((.)(?1)\2|.)$
5286 This time, the recursing alternative is tried first, and continues to
5287 recurse until it runs out of characters, at which point the recursion
5288 fails. But this time we do have another alternative to try at the
5289 higher level. That is the big difference: in the previous case the
5290 remaining alternative is at a deeper recursion level, which PCRE cannot
5291 use.
5293 To change the pattern so that it matches all palindromic strings, not
5294 just those with an odd number of characters, it is tempting to change
5295 the pattern to this:
5297 ^((.)(?1)\2|.?)$
5299 Again, this works in Perl, but not in PCRE, and for the same reason.
5300 When a deeper recursion has matched a single character, it cannot be
5301 entered again in order to match an empty string. The solution is to
5302 separate the two cases, and write out the odd and even cases as alter-
5303 natives at the higher level:
5305 ^(?:((.)(?1)\2|)|((.)(?3)\4|.))
5307 If you want to match typical palindromic phrases, the pattern has to
5308 ignore all non-word characters, which can be done like this:
5310 ^\W*+(?:((.)\W*+(?1)\W*+\2|)|((.)\W*+(?3)\W*+\4|\W*+.\W*+))\W*+$
5312 If run with the PCRE_CASELESS option, this pattern matches phrases such
5313 as "A man, a plan, a canal: Panama!" and it works well in both PCRE and
5314 Perl. Note the use of the possessive quantifier *+ to avoid backtrack-
5315 ing into sequences of non-word characters. Without this, PCRE takes a
5316 great deal longer (ten times or more) to match typical phrases, and
5317 Perl takes so long that you think it has gone into a loop.
5319 WARNING: The palindrome-matching patterns above work only if the sub-
5320 ject string does not start with a palindrome that is shorter than the
5321 entire string. For example, although "abcba" is correctly matched, if
5322 the subject is "ababa", PCRE finds the palindrome "aba" at the start,
5323 then fails at top level because the end of the string does not follow.
5324 Once again, it cannot jump back into the recursion to try other alter-
5325 natives, so the entire match fails.
5330 If the syntax for a recursive subpattern reference (either by number or
5331 by name) is used outside the parentheses to which it refers, it oper-
5332 ates like a subroutine in a programming language. The "called" subpat-
5333 tern may be defined before or after the reference. A numbered reference
5334 can be absolute or relative, as in these examples:
5336 (...(absolute)...)...(?2)...
5337 (...(relative)...)...(?-1)...
5338 (...(?+1)...(relative)...
5340 An earlier example pointed out that the pattern
5342 (sens|respons)e and \1ibility
5344 matches "sense and sensibility" and "response and responsibility", but
5345 not "sense and responsibility". If instead the pattern
5347 (sens|respons)e and (?1)ibility
5349 is used, it does match "sense and responsibility" as well as the other
5350 two strings. Another example is given in the discussion of DEFINE
5351 above.
5353 Like recursive subpatterns, a subroutine call is always treated as an
5354 atomic group. That is, once it has matched some of the subject string,
5355 it is never re-entered, even if it contains untried alternatives and
5356 there is a subsequent matching failure. Any capturing parentheses that
5357 are set during the subroutine call revert to their previous values
5358 afterwards.
5360 When a subpattern is used as a subroutine, processing options such as
5361 case-independence are fixed when the subpattern is defined. They cannot
5362 be changed for different calls. For example, consider this pattern:
5364 (abc)(?i:(?-1))
5366 It matches "abcabc". It does not match "abcABC" because the change of
5367 processing option does not affect the called subpattern.
5372 For compatibility with Oniguruma, the non-Perl syntax \g followed by a
5373 name or a number enclosed either in angle brackets or single quotes, is
5374 an alternative syntax for referencing a subpattern as a subroutine,
5375 possibly recursively. Here are two of the examples used above, rewrit-
5376 ten using this syntax:
5378 (?<pn> \( ( (?>[^()]+) | \g<pn> )* \) )
5379 (sens|respons)e and \g'1'ibility
5381 PCRE supports an extension to Oniguruma: if a number is preceded by a
5382 plus or a minus sign it is taken as a relative reference. For example:
5384 (abc)(?i:\g<-1>)
5386 Note that \g{...} (Perl syntax) and \g<...> (Oniguruma syntax) are not
5387 synonymous. The former is a back reference; the latter is a subroutine
5388 call.
5393 Perl has a feature whereby using the sequence (?{...}) causes arbitrary
5394 Perl code to be obeyed in the middle of matching a regular expression.
5395 This makes it possible, amongst other things, to extract different sub-
5396 strings that match the same pair of parentheses when there is a repeti-
5397 tion.
5399 PCRE provides a similar feature, but of course it cannot obey arbitrary
5400 Perl code. The feature is called "callout". The caller of PCRE provides
5401 an external function by putting its entry point in the global variable
5402 pcre_callout. By default, this variable contains NULL, which disables
5403 all calling out.
5405 Within a regular expression, (?C) indicates the points at which the
5406 external function is to be called. If you want to identify different
5407 callout points, you can put a number less than 256 after the letter C.
5408 The default value is zero. For example, this pattern has two callout
5409 points:
5411 (?C1)abc(?C2)def
5413 If the PCRE_AUTO_CALLOUT flag is passed to pcre_compile(), callouts are
5414 automatically installed before each item in the pattern. They are all
5415 numbered 255.
5417 During matching, when PCRE reaches a callout point (and pcre_callout is
5418 set), the external function is called. It is provided with the number
5419 of the callout, the position in the pattern, and, optionally, one item
5420 of data originally supplied by the caller of pcre_exec(). The callout
5421 function may cause matching to proceed, to backtrack, or to fail alto-
5422 gether. A complete description of the interface to the callout function
5423 is given in the pcrecallout documentation.
5428 Perl 5.10 introduced a number of "Special Backtracking Control Verbs",
5429 which are described in the Perl documentation as "experimental and sub-
5430 ject to change or removal in a future version of Perl". It goes on to
5431 say: "Their usage in production code should be noted to avoid problems
5432 during upgrades." The same remarks apply to the PCRE features described
5433 in this section.
5435 Since these verbs are specifically related to backtracking, most of
5436 them can be used only when the pattern is to be matched using
5437 pcre_exec(), which uses a backtracking algorithm. With the exception of
5438 (*FAIL), which behaves like a failing negative assertion, they cause an
5439 error if encountered by pcre_dfa_exec().
5441 If any of these verbs are used in an assertion or subroutine subpattern
5442 (including recursive subpatterns), their effect is confined to that
5443 subpattern; it does not extend to the surrounding pattern, with one
5444 exception: a *MARK that is encountered in a positive assertion is
5445 passed back (compare capturing parentheses in assertions). Note that
5446 such subpatterns are processed as anchored at the point where they are
5447 tested.
5449 The new verbs make use of what was previously invalid syntax: an open-
5450 ing parenthesis followed by an asterisk. They are generally of the form
5451 (*VERB) or (*VERB:NAME). Some may take either form, with differing be-
5452 haviour, depending on whether or not an argument is present. An name is
5453 a sequence of letters, digits, and underscores. If the name is empty,
5454 that is, if the closing parenthesis immediately follows the colon, the
5455 effect is as if the colon were not there. Any number of these verbs may
5456 occur in a pattern.
5458 PCRE contains some optimizations that are used to speed up matching by
5459 running some checks at the start of each match attempt. For example, it
5460 may know the minimum length of matching subject, or that a particular
5461 character must be present. When one of these optimizations suppresses
5462 the running of a match, any included backtracking verbs will not, of
5463 course, be processed. You can suppress the start-of-match optimizations
5464 by setting the PCRE_NO_START_OPTIMIZE option when calling pcre_com-
5465 pile() or pcre_exec(), or by starting the pattern with (*NO_START_OPT).
5467 Verbs that act immediately
5469 The following verbs act as soon as they are encountered. They may not
5470 be followed by a name.
5472 (*ACCEPT)
5474 This verb causes the match to end successfully, skipping the remainder
5475 of the pattern. When inside a recursion, only the innermost pattern is
5476 ended immediately. If (*ACCEPT) is inside capturing parentheses, the
5477 data so far is captured. (This feature was added to PCRE at release
5478 8.00.) For example:
5480 A((?:A|B(*ACCEPT)|C)D)
5482 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is cap-
5483 tured by the outer parentheses.
5485 (*FAIL) or (*F)
5487 This verb causes the match to fail, forcing backtracking to occur. It
5488 is equivalent to (?!) but easier to read. The Perl documentation notes
5489 that it is probably useful only when combined with (?{}) or (??{}).
5490 Those are, of course, Perl features that are not present in PCRE. The
5491 nearest equivalent is the callout feature, as for example in this pat-
5492 tern:
5494 a+(?C)(*FAIL)
5496 A match with the string "aaaa" always fails, but the callout is taken
5497 before each backtrack happens (in this example, 10 times).
5499 Recording which path was taken
5501 There is one verb whose main purpose is to track how a match was
5502 arrived at, though it also has a secondary use in conjunction with
5503 advancing the match starting point (see (*SKIP) below).
5505 (*MARK:NAME) or (*:NAME)
5507 A name is always required with this verb. There may be as many
5508 instances of (*MARK) as you like in a pattern, and their names do not
5509 have to be unique.
5511 When a match succeeds, the name of the last-encountered (*MARK) is
5512 passed back to the caller via the pcre_extra data structure, as
5513 described in the section on pcre_extra in the pcreapi documentation. No
5514 data is returned for a partial match. Here is an example of pcretest
5515 output, where the /K modifier requests the retrieval and outputting of
5516 (*MARK) data:
5518 /X(*MARK:A)Y|X(*MARK:B)Z/K
5519 XY
5520 0: XY
5521 MK: A
5522 XZ
5523 0: XZ
5524 MK: B
5526 The (*MARK) name is tagged with "MK:" in this output, and in this exam-
5527 ple it indicates which of the two alternatives matched. This is a more
5528 efficient way of obtaining this information than putting each alterna-
5529 tive in its own capturing parentheses.
5531 If (*MARK) is encountered in a positive assertion, its name is recorded
5532 and passed back if it is the last-encountered. This does not happen for
5533 negative assetions.
5535 A name may also be returned after a failed match if the final path
5536 through the pattern involves (*MARK). However, unless (*MARK) used in
5537 conjunction with (*COMMIT), this is unlikely to happen for an unan-
5538 chored pattern because, as the starting point for matching is advanced,
5539 the final check is often with an empty string, causing a failure before
5540 (*MARK) is reached. For example:
5542 /X(*MARK:A)Y|X(*MARK:B)Z/K
5543 XP
5544 No match
5546 There are three potential starting points for this match (starting with
5547 X, starting with P, and with an empty string). If the pattern is
5548 anchored, the result is different:
5550 /^X(*MARK:A)Y|^X(*MARK:B)Z/K
5551 XP
5552 No match, mark = B
5554 PCRE's start-of-match optimizations can also interfere with this. For
5555 example, if, as a result of a call to pcre_study(), it knows the mini-
5556 mum subject length for a match, a shorter subject will not be scanned
5557 at all.
5559 Note that similar anomalies (though different in detail) exist in Perl,
5560 no doubt for the same reasons. The use of (*MARK) data after a failed
5561 match of an unanchored pattern is not recommended, unless (*COMMIT) is
5562 involved.
5564 Verbs that act after backtracking
5566 The following verbs do nothing when they are encountered. Matching con-
5567 tinues with what follows, but if there is no subsequent match, causing
5568 a backtrack to the verb, a failure is forced. That is, backtracking
5569 cannot pass to the left of the verb. However, when one of these verbs
5570 appears inside an atomic group, its effect is confined to that group,
5571 because once the group has been matched, there is never any backtrack-
5572 ing into it. In this situation, backtracking can "jump back" to the
5573 left of the entire atomic group. (Remember also, as stated above, that
5574 this localization also applies in subroutine calls and assertions.)
5576 These verbs differ in exactly what kind of failure occurs when back-
5577 tracking reaches them.
5579 (*COMMIT)
5581 This verb, which may not be followed by a name, causes the whole match
5582 to fail outright if the rest of the pattern does not match. Even if the
5583 pattern is unanchored, no further attempts to find a match by advancing
5584 the starting point take place. Once (*COMMIT) has been passed,
5585 pcre_exec() is committed to finding a match at the current starting
5586 point, or not at all. For example:
5588 a+(*COMMIT)b
5590 This matches "xxaab" but not "aacaab". It can be thought of as a kind
5591 of dynamic anchor, or "I've started, so I must finish." The name of the
5592 most recently passed (*MARK) in the path is passed back when (*COMMIT)
5593 forces a match failure.
5595 Note that (*COMMIT) at the start of a pattern is not the same as an
5596 anchor, unless PCRE's start-of-match optimizations are turned off, as
5597 shown in this pcretest example:
5599 /(*COMMIT)abc/
5600 xyzabc
5601 0: abc
5602 xyzabc\Y
5603 No match
5605 PCRE knows that any match must start with "a", so the optimization
5606 skips along the subject to "a" before running the first match attempt,
5607 which succeeds. When the optimization is disabled by the \Y escape in
5608 the second subject, the match starts at "x" and so the (*COMMIT) causes
5609 it to fail without trying any other starting points.
5611 (*PRUNE) or (*PRUNE:NAME)
5613 This verb causes the match to fail at the current starting position in
5614 the subject if the rest of the pattern does not match. If the pattern
5615 is unanchored, the normal "bumpalong" advance to the next starting
5616 character then happens. Backtracking can occur as usual to the left of
5617 (*PRUNE), before it is reached, or when matching to the right of
5618 (*PRUNE), but if there is no match to the right, backtracking cannot
5619 cross (*PRUNE). In simple cases, the use of (*PRUNE) is just an alter-
5620 native to an atomic group or possessive quantifier, but there are some
5621 uses of (*PRUNE) that cannot be expressed in any other way. The behav-
5622 iour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE) when the
5623 match fails completely; the name is passed back if this is the final
5624 attempt. (*PRUNE:NAME) does not pass back a name if the match suc-
5625 ceeds. In an anchored pattern (*PRUNE) has the same effect as (*COM-
5626 MIT).
5628 (*SKIP)
5630 This verb, when given without a name, is like (*PRUNE), except that if
5631 the pattern is unanchored, the "bumpalong" advance is not to the next
5632 character, but to the position in the subject where (*SKIP) was encoun-
5633 tered. (*SKIP) signifies that whatever text was matched leading up to
5634 it cannot be part of a successful match. Consider:
5636 a+(*SKIP)b
5638 If the subject is "aaaac...", after the first match attempt fails
5639 (starting at the first character in the string), the starting point
5640 skips on to start the next attempt at "c". Note that a possessive quan-
5641 tifer does not have the same effect as this example; although it would
5642 suppress backtracking during the first match attempt, the second
5643 attempt would start at the second character instead of skipping on to
5644 "c".
5646 (*SKIP:NAME)
5648 When (*SKIP) has an associated name, its behaviour is modified. If the
5649 following pattern fails to match, the previous path through the pattern
5650 is searched for the most recent (*MARK) that has the same name. If one
5651 is found, the "bumpalong" advance is to the subject position that cor-
5652 responds to that (*MARK) instead of to where (*SKIP) was encountered.
5653 If no (*MARK) with a matching name is found, normal "bumpalong" of one
5654 character happens (the (*SKIP) is ignored).
5656 (*THEN) or (*THEN:NAME)
5658 This verb causes a skip to the next alternation in the innermost
5659 enclosing group if the rest of the pattern does not match. That is, it
5660 cancels pending backtracking, but only within the current alternation.
5661 Its name comes from the observation that it can be used for a pattern-
5662 based if-then-else block:
5664 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
5666 If the COND1 pattern matches, FOO is tried (and possibly further items
5667 after the end of the group if FOO succeeds); on failure the matcher
5668 skips to the second alternative and tries COND2, without backtracking
5669 into COND1. The behaviour of (*THEN:NAME) is exactly the same as
5670 (*MARK:NAME)(*THEN) if the overall match fails. If (*THEN) is not
5671 directly inside an alternation, it acts like (*PRUNE).
5673 The above verbs provide four different "strengths" of control when sub-
5674 sequent matching fails. (*THEN) is the weakest, carrying on the match
5675 at the next alternation. (*PRUNE) comes next, failing the match at the
5676 current starting position, but allowing an advance to the next charac-
5677 ter (for an unanchored pattern). (*SKIP) is similar, except that the
5678 advance may be more than one character. (*COMMIT) is the strongest,
5679 causing the entire match to fail.
5681 If more than one is present in a pattern, the "stongest" one wins. For
5682 example, consider this pattern, where A, B, etc. are complex pattern
5683 fragments:
5685 (A(*COMMIT)B(*THEN)C|D)
5687 Once A has matched, PCRE is committed to this match, at the current
5688 starting position. If subsequently B matches, but C does not, the nor-
5689 mal (*THEN) action of trying the next alternation (that is, D) does not
5690 happen because (*COMMIT) overrides.
5695 pcreapi(3), pcrecallout(3), pcrematching(3), pcresyntax(3), pcre(3).
5700 Philip Hazel
5701 University Computing Service
5702 Cambridge CB2 3QH, England.
5707 Last updated: 24 August 2011
5708 Copyright (c) 1997-2011 University of Cambridge.
5709 ------------------------------------------------------------------------------
5715 NAME
5716 PCRE - Perl-compatible regular expressions
5721 The full syntax and semantics of the regular expressions that are sup-
5722 ported by PCRE are described in the pcrepattern documentation. This
5723 document contains just a quick-reference summary of the syntax.
5728 \x where x is non-alphanumeric is a literal x
5729 \Q...\E treat enclosed characters as literal
5734 \a alarm, that is, the BEL character (hex 07)
5735 \cx "control-x", where x is any ASCII character
5736 \e escape (hex 1B)
5737 \f formfeed (hex 0C)
5738 \n newline (hex 0A)
5739 \r carriage return (hex 0D)
5740 \t tab (hex 09)
5741 \ddd character with octal code ddd, or backreference
5742 \xhh character with hex code hh
5743 \x{hhh..} character with hex code hhh..
5748 . any character except newline;
5749 in dotall mode, any character whatsoever
5750 \C one byte, even in UTF-8 mode (best avoided)
5751 \d a decimal digit
5752 \D a character that is not a decimal digit
5753 \h a horizontal whitespace character
5754 \H a character that is not a horizontal whitespace character
5755 \N a character that is not a newline
5756 \p{xx} a character with the xx property
5757 \P{xx} a character without the xx property
5758 \R a newline sequence
5759 \s a whitespace character
5760 \S a character that is not a whitespace character
5761 \v a vertical whitespace character
5762 \V a character that is not a vertical whitespace character
5763 \w a "word" character
5764 \W a "non-word" character
5765 \X an extended Unicode sequence
5767 In PCRE, by default, \d, \D, \s, \S, \w, and \W recognize only ASCII
5768 characters, even in UTF-8 mode. However, this can be changed by setting
5769 the PCRE_UCP option.
5774 C Other
5775 Cc Control
5776 Cf Format
5777 Cn Unassigned
5778 Co Private use
5779 Cs Surrogate
5781 L Letter
5782 Ll Lower case letter
5783 Lm Modifier letter
5784 Lo Other letter
5785 Lt Title case letter
5786 Lu Upper case letter
5787 L& Ll, Lu, or Lt
5789 M Mark
5790 Mc Spacing mark
5791 Me Enclosing mark
5792 Mn Non-spacing mark
5794 N Number
5795 Nd Decimal number
5796 Nl Letter number
5797 No Other number
5799 P Punctuation
5800 Pc Connector punctuation
5801 Pd Dash punctuation
5802 Pe Close punctuation
5803 Pf Final punctuation
5804 Pi Initial punctuation
5805 Po Other punctuation
5806 Ps Open punctuation
5808 S Symbol
5809 Sc Currency symbol
5810 Sk Modifier symbol
5811 Sm Mathematical symbol
5812 So Other symbol
5814 Z Separator
5815 Zl Line separator
5816 Zp Paragraph separator
5817 Zs Space separator
5822 Xan Alphanumeric: union of properties L and N
5823 Xps POSIX space: property Z or tab, NL, VT, FF, CR
5824 Xsp Perl space: property Z or tab, NL, FF, CR
5825 Xwd Perl word: property Xan or underscore
5830 Arabic, Armenian, Avestan, Balinese, Bamum, Bengali, Bopomofo, Braille,
5831 Buginese, Buhid, Canadian_Aboriginal, Carian, Cham, Cherokee, Common,
5832 Coptic, Cuneiform, Cypriot, Cyrillic, Deseret, Devanagari, Egyp-
5833 tian_Hieroglyphs, Ethiopic, Georgian, Glagolitic, Gothic, Greek,
5834 Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hiragana, Impe-
5835 rial_Aramaic, Inherited, Inscriptional_Pahlavi, Inscriptional_Parthian,
5836 Javanese, Kaithi, Kannada, Katakana, Kayah_Li, Kharoshthi, Khmer, Lao,
5837 Latin, Lepcha, Limbu, Linear_B, Lisu, Lycian, Lydian, Malayalam,
5838 Meetei_Mayek, Mongolian, Myanmar, New_Tai_Lue, Nko, Ogham, Old_Italic,
5839 Old_Persian, Old_South_Arabian, Old_Turkic, Ol_Chiki, Oriya, Osmanya,
5840 Phags_Pa, Phoenician, Rejang, Runic, Samaritan, Saurashtra, Shavian,
5841 Sinhala, Sundanese, Syloti_Nagri, Syriac, Tagalog, Tagbanwa, Tai_Le,
5842 Tai_Tham, Tai_Viet, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh,
5843 Ugaritic, Vai, Yi.
5848 [...] positive character class
5849 [^...] negative character class
5850 [x-y] range (can be used for hex characters)
5851 [[:xxx:]] positive POSIX named set
5852 [[:^xxx:]] negative POSIX named set
5854 alnum alphanumeric
5855 alpha alphabetic
5856 ascii 0-127
5857 blank space or tab
5858 cntrl control character
5859 digit decimal digit
5860 graph printing, excluding space
5861 lower lower case letter
5862 print printing, including space
5863 punct printing, excluding alphanumeric
5864 space whitespace
5865 upper upper case letter
5866 word same as \w
5867 xdigit hexadecimal digit
5869 In PCRE, POSIX character set names recognize only ASCII characters by
5870 default, but some of them use Unicode properties if PCRE_UCP is set.
5871 You can use \Q...\E inside a character class.
5876 ? 0 or 1, greedy
5877 ?+ 0 or 1, possessive
5878 ?? 0 or 1, lazy
5879 * 0 or more, greedy
5880 *+ 0 or more, possessive
5881 *? 0 or more, lazy
5882 + 1 or more, greedy
5883 ++ 1 or more, possessive
5884 +? 1 or more, lazy
5885 {n} exactly n
5886 {n,m} at least n, no more than m, greedy
5887 {n,m}+ at least n, no more than m, possessive
5888 {n,m}? at least n, no more than m, lazy
5889 {n,} n or more, greedy
5890 {n,}+ n or more, possessive
5891 {n,}? n or more, lazy
5896 \b word boundary
5897 \B not a word boundary
5898 ^ start of subject
5899 also after internal newline in multiline mode
5900 \A start of subject
5901 $ end of subject
5902 also before newline at end of subject
5903 also before internal newline in multiline mode
5904 \Z end of subject
5905 also before newline at end of subject
5906 \z end of subject
5907 \G first matching position in subject
5912 \K reset start of match
5917 expr|expr|expr...
5922 (...) capturing group
5923 (?<name>...) named capturing group (Perl)
5924 (?'name'...) named capturing group (Perl)
5925 (?P<name>...) named capturing group (Python)
5926 (?:...) non-capturing group
5927 (?|...) non-capturing group; reset group numbers for
5928 capturing groups in each alternative
5933 (?>...) atomic, non-capturing group
5938 (?#....) comment (not nestable)
5943 (?i) caseless
5944 (?J) allow duplicate names
5945 (?m) multiline
5946 (?s) single line (dotall)
5947 (?U) default ungreedy (lazy)
5948 (?x) extended (ignore white space)
5949 (?-...) unset option(s)
5951 The following are recognized only at the start of a pattern or after
5952 one of the newline-setting options with similar syntax:
5954 (*NO_START_OPT) no start-match optimization (PCRE_NO_START_OPTIMIZE)
5955 (*UTF8) set UTF-8 mode (PCRE_UTF8)
5956 (*UCP) set PCRE_UCP (use Unicode properties for \d etc)
5961 (?=...) positive look ahead
5962 (?!...) negative look ahead
5963 (?<=...) positive look behind
5964 (?<!...) negative look behind
5966 Each top-level branch of a look behind must be of a fixed length.
5971 \n reference by number (can be ambiguous)
5972 \gn reference by number
5973 \g{n} reference by number
5974 \g{-n} relative reference by number
5975 \k<name> reference by name (Perl)
5976 \k'name' reference by name (Perl)
5977 \g{name} reference by name (Perl)
5978 \k{name} reference by name (.NET)
5979 (?P=name) reference by name (Python)
5984 (?R) recurse whole pattern
5985 (?n) call subpattern by absolute number
5986 (?+n) call subpattern by relative number
5987 (?-n) call subpattern by relative number
5988 (?&name) call subpattern by name (Perl)
5989 (?P>name) call subpattern by name (Python)
5990 \g<name> call subpattern by name (Oniguruma)
5991 \g'name' call subpattern by name (Oniguruma)
5992 \g<n> call subpattern by absolute number (Oniguruma)
5993 \g'n' call subpattern by absolute number (Oniguruma)
5994 \g<+n> call subpattern by relative number (PCRE extension)
5995 \g'+n' call subpattern by relative number (PCRE extension)
5996 \g<-n> call subpattern by relative number (PCRE extension)
5997 \g'-n' call subpattern by relative number (PCRE extension)
6002 (?(condition)yes-pattern)
6003 (?(condition)yes-pattern|no-pattern)
6005 (?(n)... absolute reference condition
6006 (?(+n)... relative reference condition
6007 (?(-n)... relative reference condition
6008 (?(<name>)... named reference condition (Perl)
6009 (?('name')... named reference condition (Perl)
6010 (?(name)... named reference condition (PCRE)
6011 (?(R)... overall recursion condition
6012 (?(Rn)... specific group recursion condition
6013 (?(R&name)... specific recursion condition
6014 (?(DEFINE)... define subpattern for reference
6015 (?(assert)... assertion condition
6020 The following act immediately they are reached:
6022 (*ACCEPT) force successful match
6023 (*FAIL) force backtrack; synonym (*F)
6025 The following act only when a subsequent match failure causes a back-
6026 track to reach them. They all force a match failure, but they differ in
6027 what happens afterwards. Those that advance the start-of-match point do
6028 so only if the pattern is not anchored.
6030 (*COMMIT) overall failure, no advance of starting point
6031 (*PRUNE) advance to next starting character
6032 (*SKIP) advance start to current matching position
6033 (*THEN) local failure, backtrack to next alternation