ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 959 - (show annotations)
Sat Apr 14 16:16:58 2012 UTC (7 years, 4 months ago) by ph10
File MIME type: text/plain
File size: 441414 byte(s)
Update documentation to clarify that UTF-8/16 checking is done on complete 
strings before any other processing.
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 Starting with release 8.30, it is possible to compile two separate PCRE
29 libraries: the original, which supports 8-bit character strings
30 (including UTF-8 strings), and a second library that supports 16-bit
31 character strings (including UTF-16 strings). The build process allows
32 either one or both to be built. The majority of the work to make this
33 possible was done by Zoltan Herczeg.
35 The two libraries contain identical sets of functions, except that the
36 names in the 16-bit library start with pcre16_ instead of pcre_. To
37 avoid over-complication and reduce the documentation maintenance load,
38 most of the documentation describes the 8-bit library, with the differ-
39 ences for the 16-bit library described separately in the pcre16 page.
40 References to functions or structures of the form pcre[16]_xxx should
41 be read as meaning "pcre_xxx when using the 8-bit library and
42 pcre16_xxx when using the 16-bit library".
44 The current implementation of PCRE corresponds approximately with Perl
45 5.12, including support for UTF-8/16 encoded strings and Unicode gen-
46 eral category properties. However, UTF-8/16 and Unicode support has to
47 be explicitly enabled; it is not the default. The Unicode tables corre-
48 spond to Unicode release 6.0.0.
50 In addition to the Perl-compatible matching function, PCRE contains an
51 alternative function that matches the same compiled patterns in a dif-
52 ferent way. In certain circumstances, the alternative function has some
53 advantages. For a discussion of the two matching algorithms, see the
54 pcrematching page.
56 PCRE is written in C and released as a C library. A number of people
57 have written wrappers and interfaces of various kinds. In particular,
58 Google Inc. have provided a comprehensive C++ wrapper for the 8-bit
59 library. This is now included as part of the PCRE distribution. The
60 pcrecpp page has details of this interface. Other people's contribu-
61 tions can be found in the Contrib directory at the primary FTP site,
62 which is:
64 ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre
66 Details of exactly which Perl regular expression features are and are
67 not supported by PCRE are given in separate documents. See the pcrepat-
68 tern and pcrecompat pages. There is a syntax summary in the pcresyntax
69 page.
71 Some features of PCRE can be included, excluded, or changed when the
72 library is built. The pcre_config() function makes it possible for a
73 client to discover which features are available. The features them-
74 selves are described in the pcrebuild page. Documentation about build-
75 ing PCRE for various operating systems can be found in the README and
76 NON-UNIX-USE files in the source distribution.
78 The libraries contains a number of undocumented internal functions and
79 data tables that are used by more than one of the exported external
80 functions, but which are not intended for use by external callers.
81 Their names all begin with "_pcre_" or "_pcre16_", which hopefully will
82 not provoke any name clashes. In some environments, it is possible to
83 control which external symbols are exported when a shared library is
84 built, and in these cases the undocumented symbols are not exported.
89 The user documentation for PCRE comprises a number of different sec-
90 tions. In the "man" format, each of these is a separate "man page". In
91 the HTML format, each is a separate page, linked from the index page.
92 In the plain text format, all the sections, except the pcredemo sec-
93 tion, are concatenated, for ease of searching. The sections are as fol-
94 lows:
96 pcre this document
97 pcre16 details of the 16-bit library
98 pcre-config show PCRE installation configuration information
99 pcreapi details of PCRE's native C API
100 pcrebuild options for building PCRE
101 pcrecallout details of the callout feature
102 pcrecompat discussion of Perl compatibility
103 pcrecpp details of the C++ wrapper for the 8-bit library
104 pcredemo a demonstration C program that uses PCRE
105 pcregrep description of the pcregrep command (8-bit only)
106 pcrejit discussion of the just-in-time optimization support
107 pcrelimits details of size and other limits
108 pcrematching discussion of the two matching algorithms
109 pcrepartial details of the partial matching facility
110 pcrepattern syntax and semantics of supported
111 regular expressions
112 pcreperform discussion of performance issues
113 pcreposix the POSIX-compatible C API for the 8-bit library
114 pcreprecompile details of saving and re-using precompiled patterns
115 pcresample discussion of the pcredemo program
116 pcrestack discussion of stack usage
117 pcresyntax quick syntax reference
118 pcretest description of the pcretest testing command
119 pcreunicode discussion of Unicode and UTF-8/16 support
121 In addition, in the "man" and HTML formats, there is a short page for
122 each 8-bit C library function, listing its arguments and results.
127 Philip Hazel
128 University Computing Service
129 Cambridge CB2 3QH, England.
131 Putting an actual email address here seems to have been a spam magnet,
132 so I've taken it away. If you want to email me, use my two initials,
133 followed by the two digits 10, at the domain cam.ac.uk.
138 Last updated: 10 January 2012
139 Copyright (c) 1997-2012 University of Cambridge.
140 ------------------------------------------------------------------------------
143 PCRE(3) PCRE(3)
146 NAME
147 PCRE - Perl-compatible regular expressions
149 #include <pcre.h>
154 pcre16 *pcre16_compile(PCRE_SPTR16 pattern, int options,
155 const char **errptr, int *erroffset,
156 const unsigned char *tableptr);
158 pcre16 *pcre16_compile2(PCRE_SPTR16 pattern, int options,
159 int *errorcodeptr,
160 const char **errptr, int *erroffset,
161 const unsigned char *tableptr);
163 pcre16_extra *pcre16_study(const pcre16 *code, int options,
164 const char **errptr);
166 void pcre16_free_study(pcre16_extra *extra);
168 int pcre16_exec(const pcre16 *code, const pcre16_extra *extra,
169 PCRE_SPTR16 subject, int length, int startoffset,
170 int options, int *ovector, int ovecsize);
172 int pcre16_dfa_exec(const pcre16 *code, const pcre16_extra *extra,
173 PCRE_SPTR16 subject, int length, int startoffset,
174 int options, int *ovector, int ovecsize,
175 int *workspace, int wscount);
180 int pcre16_copy_named_substring(const pcre16 *code,
181 PCRE_SPTR16 subject, int *ovector,
182 int stringcount, PCRE_SPTR16 stringname,
183 PCRE_UCHAR16 *buffer, int buffersize);
185 int pcre16_copy_substring(PCRE_SPTR16 subject, int *ovector,
186 int stringcount, int stringnumber, PCRE_UCHAR16 *buffer,
187 int buffersize);
189 int pcre16_get_named_substring(const pcre16 *code,
190 PCRE_SPTR16 subject, int *ovector,
191 int stringcount, PCRE_SPTR16 stringname,
192 PCRE_SPTR16 *stringptr);
194 int pcre16_get_stringnumber(const pcre16 *code,
195 PCRE_SPTR16 name);
197 int pcre16_get_stringtable_entries(const pcre16 *code,
198 PCRE_SPTR16 name, PCRE_UCHAR16 **first, PCRE_UCHAR16 **last);
200 int pcre16_get_substring(PCRE_SPTR16 subject, int *ovector,
201 int stringcount, int stringnumber,
202 PCRE_SPTR16 *stringptr);
204 int pcre16_get_substring_list(PCRE_SPTR16 subject,
205 int *ovector, int stringcount, PCRE_SPTR16 **listptr);
207 void pcre16_free_substring(PCRE_SPTR16 stringptr);
209 void pcre16_free_substring_list(PCRE_SPTR16 *stringptr);
214 pcre16_jit_stack *pcre16_jit_stack_alloc(int startsize, int maxsize);
216 void pcre16_jit_stack_free(pcre16_jit_stack *stack);
218 void pcre16_assign_jit_stack(pcre16_extra *extra,
219 pcre16_jit_callback callback, void *data);
221 const unsigned char *pcre16_maketables(void);
223 int pcre16_fullinfo(const pcre16 *code, const pcre16_extra *extra,
224 int what, void *where);
226 int pcre16_refcount(pcre16 *code, int adjust);
228 int pcre16_config(int what, void *where);
230 const char *pcre16_version(void);
232 int pcre16_pattern_to_host_byte_order(pcre16 *code,
233 pcre16_extra *extra, const unsigned char *tables);
238 void *(*pcre16_malloc)(size_t);
240 void (*pcre16_free)(void *);
242 void *(*pcre16_stack_malloc)(size_t);
244 void (*pcre16_stack_free)(void *);
246 int (*pcre16_callout)(pcre16_callout_block *);
251 int pcre16_utf16_to_host_byte_order(PCRE_UCHAR16 *output,
252 PCRE_SPTR16 input, int length, int *byte_order,
253 int keep_boms);
258 Starting with release 8.30, it is possible to compile a PCRE library
259 that supports 16-bit character strings, including UTF-16 strings, as
260 well as or instead of the original 8-bit library. The majority of the
261 work to make this possible was done by Zoltan Herczeg. The two
262 libraries contain identical sets of functions, used in exactly the same
263 way. Only the names of the functions and the data types of their argu-
264 ments and results are different. To avoid over-complication and reduce
265 the documentation maintenance load, most of the PCRE documentation
266 describes the 8-bit library, with only occasional references to the
267 16-bit library. This page describes what is different when you use the
268 16-bit library.
270 WARNING: A single application can be linked with both libraries, but
271 you must take care when processing any particular pattern to use func-
272 tions from just one library. For example, if you want to study a pat-
273 tern that was compiled with pcre16_compile(), you must do so with
274 pcre16_study(), not pcre_study(), and you must free the study data with
275 pcre16_free_study().
280 There is only one header file, pcre.h. It contains prototypes for all
281 the functions in both libraries, as well as definitions of flags,
282 structures, error codes, etc.
287 In Unix-like systems, the 16-bit library is called libpcre16, and can
288 normally be accesss by adding -lpcre16 to the command for linking an
289 application that uses PCRE.
294 In the 8-bit library, strings are passed to PCRE library functions as
295 vectors of bytes with the C type "char *". In the 16-bit library,
296 strings are passed as vectors of unsigned 16-bit quantities. The macro
297 PCRE_UCHAR16 specifies an appropriate data type, and PCRE_SPTR16 is
298 defined as "const PCRE_UCHAR16 *". In very many environments, "short
299 int" is a 16-bit data type. When PCRE is built, it defines PCRE_UCHAR16
300 as "short int", but checks that it really is a 16-bit data type. If it
301 is not, the build fails with an error message telling the maintainer to
302 modify the definition appropriately.
307 The types of the opaque structures that are used for compiled 16-bit
308 patterns and JIT stacks are pcre16 and pcre16_jit_stack respectively.
309 The type of the user-accessible structure that is returned by
310 pcre16_study() is pcre16_extra, and the type of the structure that is
311 used for passing data to a callout function is pcre16_callout_block.
312 These structures contain the same fields, with the same names, as their
313 8-bit counterparts. The only difference is that pointers to character
314 strings are 16-bit instead of 8-bit types.
319 For every function in the 8-bit library there is a corresponding func-
320 tion in the 16-bit library with a name that starts with pcre16_ instead
321 of pcre_. The prototypes are listed above. In addition, there is one
322 extra function, pcre16_utf16_to_host_byte_order(). This is a utility
323 function that converts a UTF-16 character string to host byte order if
324 necessary. The other 16-bit functions expect the strings they are
325 passed to be in host byte order.
327 The input and output arguments of pcre16_utf16_to_host_byte_order() may
328 point to the same address, that is, conversion in place is supported.
329 The output buffer must be at least as long as the input.
331 The length argument specifies the number of 16-bit data units in the
332 input string; a negative value specifies a zero-terminated string.
334 If byte_order is NULL, it is assumed that the string starts off in host
335 byte order. This may be changed by byte-order marks (BOMs) anywhere in
336 the string (commonly as the first character).
338 If byte_order is not NULL, a non-zero value of the integer to which it
339 points means that the input starts off in host byte order, otherwise
340 the opposite order is assumed. Again, BOMs in the string can change
341 this. The final byte order is passed back at the end of processing.
343 If keep_boms is not zero, byte-order mark characters (0xfeff) are
344 copied into the output string. Otherwise they are discarded.
346 The result of the function is the number of 16-bit units placed into
347 the output buffer, including the zero terminator if the string was
348 zero-terminated.
353 The offsets within subject strings that are returned by the matching
354 functions are in 16-bit units rather than bytes.
359 The name-to-number translation table that is maintained for named sub-
360 patterns uses 16-bit characters. The pcre16_get_stringtable_entries()
361 function returns the length of each entry in the table as the number of
362 16-bit data units.
367 There are two new general option names, PCRE_UTF16 and
368 PCRE_NO_UTF16_CHECK, which correspond to PCRE_UTF8 and
369 PCRE_NO_UTF8_CHECK in the 8-bit library. In fact, these new options
370 define the same bits in the options word. There is a discussion about
371 the validity of UTF-16 strings in the pcreunicode page.
373 For the pcre16_config() function there is an option PCRE_CONFIG_UTF16
374 that returns 1 if UTF-16 support is configured, otherwise 0. If this
375 option is given to pcre_config(), or if the PCRE_CONFIG_UTF8 option is
376 given to pcre16_config(), the result is the PCRE_ERROR_BADOPTION error.
381 In 16-bit mode, when PCRE_UTF16 is not set, character values are
382 treated in the same way as in 8-bit, non UTF-8 mode, except, of course,
383 that they can range from 0 to 0xffff instead of 0 to 0xff. Character
384 types for characters less than 0xff can therefore be influenced by the
385 locale in the same way as before. Characters greater than 0xff have
386 only one case, and no "type" (such as letter or digit).
388 In UTF-16 mode, the character code is Unicode, in the range 0 to
389 0x10ffff, with the exception of values in the range 0xd800 to 0xdfff
390 because those are "surrogate" values that are used in pairs to encode
391 values greater than 0xffff.
393 A UTF-16 string can indicate its endianness by special code knows as a
394 byte-order mark (BOM). The PCRE functions do not handle this, expecting
395 strings to be in host byte order. A utility function called
396 pcre16_utf16_to_host_byte_order() is provided to help with this (see
397 above).
403 spond to their 8-bit counterparts. The error PCRE_ERROR_BADMODE is
404 given when a compiled pattern is passed to a function that processes
405 patterns in the other mode, for example, if a pattern compiled with
406 pcre_compile() is passed to pcre16_exec().
408 There are new error codes whose names begin with PCRE_UTF16_ERR for
409 invalid UTF-16 strings, corresponding to the PCRE_UTF8_ERR codes for
410 UTF-8 strings that are described in the section entitled "Reason codes
411 for invalid UTF-8 strings" in the main pcreapi page. The UTF-16 errors
412 are:
414 PCRE_UTF16_ERR1 Missing low surrogate at end of string
415 PCRE_UTF16_ERR2 Invalid low surrogate follows high surrogate
416 PCRE_UTF16_ERR3 Isolated low surrogate
417 PCRE_UTF16_ERR4 Invalid character 0xfffe
422 If there is an error while compiling a pattern, the error text that is
423 passed back by pcre16_compile() or pcre16_compile2() is still an 8-bit
424 character string, zero-terminated.
429 The subject and mark fields in the callout block that is passed to a
430 callout function point to 16-bit vectors.
435 The pcretest program continues to operate with 8-bit input and output
436 files, but it can be used for testing the 16-bit library. If it is run
437 with the command line option -16, patterns and subject strings are con-
438 verted from 8-bit to 16-bit before being passed to PCRE, and the 16-bit
439 library functions are used instead of the 8-bit ones. Returned 16-bit
440 strings are converted to 8-bit for output. If the 8-bit library was not
441 compiled, pcretest defaults to 16-bit and the -16 option is ignored.
443 When PCRE is being built, the RunTest script that is called by "make
444 check" uses the pcretest -C option to discover which of the 8-bit and
445 16-bit libraries has been built, and runs the tests appropriately.
450 Not all the features of the 8-bit library are available with the 16-bit
451 library. The C++ and POSIX wrapper functions support only the 8-bit
452 library, and the pcregrep program is at present 8-bit only.
457 Philip Hazel
458 University Computing Service
459 Cambridge CB2 3QH, England.
464 Last updated: 14 April 2012
465 Copyright (c) 1997-2012 University of Cambridge.
466 ------------------------------------------------------------------------------
472 NAME
473 PCRE - Perl-compatible regular expressions
478 This document describes the optional features of PCRE that can be
479 selected when the library is compiled. It assumes use of the configure
480 script, where the optional features are selected or deselected by pro-
481 viding options to configure before running the make command. However,
482 the same options can be selected in both Unix-like and non-Unix-like
483 environments using the GUI facility of cmake-gui if you are using CMake
484 instead of configure to build PCRE.
486 There is a lot more information about building PCRE in non-Unix-like
487 environments in the file called NON_UNIX_USE, which is part of the PCRE
488 distribution. You should consult this file as well as the README file
489 if you are building in a non-Unix-like environment.
491 The complete list of options for configure (which includes the standard
492 ones such as the selection of the installation directory) can be
493 obtained by running
495 ./configure --help
497 The following sections include descriptions of options whose names
498 begin with --enable or --disable. These settings specify changes to the
499 defaults for the configure command. Because of the way that configure
500 works, --enable and --disable always come in pairs, so the complemen-
501 tary option always exists as well, but as it specifies the default, it
502 is not described.
507 By default, a library called libpcre is built, containing functions
508 that take string arguments contained in vectors of bytes, either as
509 single-byte characters, or interpreted as UTF-8 strings. You can also
510 build a separate library, called libpcre16, in which strings are con-
511 tained in vectors of 16-bit data units and interpreted either as sin-
512 gle-unit characters or UTF-16 strings, by adding
514 --enable-pcre16
516 to the configure command. If you do not want the 8-bit library, add
518 --disable-pcre8
520 as well. At least one of the two libraries must be built. Note that the
521 C++ and POSIX wrappers are for the 8-bit library only, and that pcre-
522 grep is an 8-bit program. None of these are built if you select only
523 the 16-bit library.
528 The PCRE building process uses libtool to build both shared and static
529 Unix libraries by default. You can suppress one of these by adding one
530 of
532 --disable-shared
533 --disable-static
535 to the configure command, as required.
540 By default, if the 8-bit library is being built, the configure script
541 will search for a C++ compiler and C++ header files. If it finds them,
542 it automatically builds the C++ wrapper library (which supports only
543 8-bit strings). You can disable this by adding
545 --disable-cpp
547 to the configure command.
550 UTF-8 and UTF-16 SUPPORT
552 To build PCRE with support for UTF Unicode character strings, add
554 --enable-utf
556 to the configure command. This setting applies to both libraries,
557 adding support for UTF-8 to the 8-bit library and support for UTF-16 to
558 the 16-bit library. There are no separate options for enabling UTF-8
559 and UTF-16 independently because that would allow ridiculous settings
560 such as requesting UTF-16 support while building only the 8-bit
561 library. It is not possible to build one library with UTF support and
562 the other without in the same configuration. (For backwards compatibil-
563 ity, --enable-utf8 is a synonym of --enable-utf.)
565 Of itself, this setting does not make PCRE treat strings as UTF-8 or
566 UTF-16. As well as compiling PCRE with this option, you also have have
567 to set the PCRE_UTF8 or PCRE_UTF16 option when you call one of the pat-
568 tern compiling functions.
570 If you set --enable-utf when compiling in an EBCDIC environment, PCRE
571 expects its input to be either ASCII or UTF-8 (depending on the runtime
572 option). It is not possible to support both EBCDIC and UTF-8 codes in
573 the same version of the library. Consequently, --enable-utf and
574 --enable-ebcdic are mutually exclusive.
579 UTF support allows the libraries to process character codepoints up to
580 0x10ffff in the strings that they handle. On its own, however, it does
581 not provide any facilities for accessing the properties of such charac-
582 ters. If you want to be able to use the pattern escapes \P, \p, and \X,
583 which refer to Unicode character properties, you must add
585 --enable-unicode-properties
587 to the configure command. This implies UTF support, even if you have
588 not explicitly requested it.
590 Including Unicode property support adds around 30K of tables to the
591 PCRE library. Only the general category properties such as Lu and Nd
592 are supported. Details are given in the pcrepattern documentation.
597 Just-in-time compiler support is included in the build by specifying
599 --enable-jit
601 This support is available only for certain hardware architectures. If
602 this option is set for an unsupported architecture, a compile time
603 error occurs. See the pcrejit documentation for a discussion of JIT
604 usage. When JIT support is enabled, pcregrep automatically makes use of
605 it, unless you add
607 --disable-pcregrep-jit
609 to the "configure" command.
614 By default, PCRE interprets the linefeed (LF) character as indicating
615 the end of a line. This is the normal newline character on Unix-like
616 systems. You can compile PCRE to use carriage return (CR) instead, by
617 adding
619 --enable-newline-is-cr
621 to the configure command. There is also a --enable-newline-is-lf
622 option, which explicitly specifies linefeed as the newline character.
624 Alternatively, you can specify that line endings are to be indicated by
625 the two character sequence CRLF. If you want this, add
627 --enable-newline-is-crlf
629 to the configure command. There is a fourth option, specified by
631 --enable-newline-is-anycrlf
633 which causes PCRE to recognize any of the three sequences CR, LF, or
634 CRLF as indicating a line ending. Finally, a fifth option, specified by
636 --enable-newline-is-any
638 causes PCRE to recognize any Unicode newline sequence.
640 Whatever line ending convention is selected when PCRE is built can be
641 overridden when the library functions are called. At build time it is
642 conventional to use the standard for your operating system.
647 By default, the sequence \R in a pattern matches any Unicode newline
648 sequence, whatever has been selected as the line ending sequence. If
649 you specify
651 --enable-bsr-anycrlf
653 the default is changed so that \R matches only CR, LF, or CRLF. What-
654 ever is selected when PCRE is built can be overridden when the library
655 functions are called.
660 When the 8-bit library is called through the POSIX interface (see the
661 pcreposix documentation), additional working storage is required for
662 holding the pointers to capturing substrings, because PCRE requires
663 three integers per substring, whereas the POSIX interface provides only
664 two. If the number of expected substrings is small, the wrapper func-
665 tion uses space on the stack, because this is faster than using mal-
666 loc() for each call. The default threshold above which the stack is no
667 longer used is 10; it can be changed by adding a setting such as
669 --with-posix-malloc-threshold=20
671 to the configure command.
676 Within a compiled pattern, offset values are used to point from one
677 part to another (for example, from an opening parenthesis to an alter-
678 nation metacharacter). By default, two-byte values are used for these
679 offsets, leading to a maximum size for a compiled pattern of around
680 64K. This is sufficient to handle all but the most gigantic patterns.
681 Nevertheless, some people do want to process truly enormous patterns,
682 so it is possible to compile PCRE to use three-byte or four-byte off-
683 sets by adding a setting such as
685 --with-link-size=3
687 to the configure command. The value given must be 2, 3, or 4. For the
688 16-bit library, a value of 3 is rounded up to 4. Using longer offsets
689 slows down the operation of PCRE because it has to load additional data
690 when handling them.
695 When matching with the pcre_exec() function, PCRE implements backtrack-
696 ing by making recursive calls to an internal function called match().
697 In environments where the size of the stack is limited, this can se-
698 verely limit PCRE's operation. (The Unix environment does not usually
699 suffer from this problem, but it may sometimes be necessary to increase
700 the maximum stack size. There is a discussion in the pcrestack docu-
701 mentation.) An alternative approach to recursion that uses memory from
702 the heap to remember data, instead of using recursive function calls,
703 has been implemented to work round the problem of limited stack size.
704 If you want to build a version of PCRE that works this way, add
706 --disable-stack-for-recursion
708 to the configure command. With this configuration, PCRE will use the
709 pcre_stack_malloc and pcre_stack_free variables to call memory manage-
710 ment functions. By default these point to malloc() and free(), but you
711 can replace the pointers so that your own functions are used instead.
713 Separate functions are provided rather than using pcre_malloc and
714 pcre_free because the usage is very predictable: the block sizes
715 requested are always the same, and the blocks are always freed in
716 reverse order. A calling program might be able to implement optimized
717 functions that perform better than malloc() and free(). PCRE runs
718 noticeably more slowly when built in this way. This option affects only
719 the pcre_exec() function; it is not relevant for pcre_dfa_exec().
724 Internally, PCRE has a function called match(), which it calls repeat-
725 edly (sometimes recursively) when matching a pattern with the
726 pcre_exec() function. By controlling the maximum number of times this
727 function may be called during a single matching operation, a limit can
728 be placed on the resources used by a single call to pcre_exec(). The
729 limit can be changed at run time, as described in the pcreapi documen-
730 tation. The default is 10 million, but this can be changed by adding a
731 setting such as
733 --with-match-limit=500000
735 to the configure command. This setting has no effect on the
736 pcre_dfa_exec() matching function.
738 In some environments it is desirable to limit the depth of recursive
739 calls of match() more strictly than the total number of calls, in order
740 to restrict the maximum amount of stack (or heap, if --disable-stack-
741 for-recursion is specified) that is used. A second limit controls this;
742 it defaults to the value that is set for --with-match-limit, which
743 imposes no additional constraints. However, you can set a lower limit
744 by adding, for example,
746 --with-match-limit-recursion=10000
748 to the configure command. This value can also be overridden at run
749 time.
754 PCRE uses fixed tables for processing characters whose code values are
755 less than 256. By default, PCRE is built with a set of tables that are
756 distributed in the file pcre_chartables.c.dist. These tables are for
757 ASCII codes only. If you add
759 --enable-rebuild-chartables
761 to the configure command, the distributed tables are no longer used.
762 Instead, a program called dftables is compiled and run. This outputs
763 the source for new set of tables, created in the default locale of your
764 C runtime system. (This method of replacing the tables does not work if
765 you are cross compiling, because dftables is run on the local host. If
766 you need to create alternative tables when cross compiling, you will
767 have to do so "by hand".)
772 PCRE assumes by default that it will run in an environment where the
773 character code is ASCII (or Unicode, which is a superset of ASCII).
774 This is the case for most computer operating systems. PCRE can, how-
775 ever, be compiled to run in an EBCDIC environment by adding
777 --enable-ebcdic
779 to the configure command. This setting implies --enable-rebuild-charta-
780 bles. You should only use it if you know that you are in an EBCDIC
781 environment (for example, an IBM mainframe operating system). The
782 --enable-ebcdic option is incompatible with --enable-utf.
787 By default, pcregrep reads all files as plain text. You can build it so
788 that it recognizes files whose names end in .gz or .bz2, and reads them
789 with libz or libbz2, respectively, by adding one or both of
791 --enable-pcregrep-libz
792 --enable-pcregrep-libbz2
794 to the configure command. These options naturally require that the rel-
795 evant libraries are installed on your system. Configuration will fail
796 if they are not.
801 pcregrep uses an internal buffer to hold a "window" on the file it is
802 scanning, in order to be able to output "before" and "after" lines when
803 it finds a match. The size of the buffer is controlled by a parameter
804 whose default value is 20K. The buffer itself is three times this size,
805 but because of the way it is used for holding "before" lines, the long-
806 est line that is guaranteed to be processable is the parameter size.
807 You can change the default parameter value by adding, for example,
809 --with-pcregrep-bufsize=50K
811 to the configure command. The caller of pcregrep can, however, override
812 this value by specifying a run-time option.
817 If you add
819 --enable-pcretest-libreadline
821 to the configure command, pcretest is linked with the libreadline
822 library, and when its input is from a terminal, it reads it using the
823 readline() function. This provides line-editing and history facilities.
824 Note that libreadline is GPL-licensed, so if you distribute a binary of
825 pcretest linked in this way, there may be licensing issues.
827 Setting this option causes the -lreadline option to be added to the
828 pcretest build. In many operating environments with a sytem-installed
829 libreadline this is sufficient. However, in some environments (e.g. if
830 an unmodified distribution version of readline is in use), some extra
831 configuration may be necessary. The INSTALL file for libreadline says
832 this:
834 "Readline uses the termcap functions, but does not link with the
835 termcap or curses library itself, allowing applications which link
836 with readline the to choose an appropriate library."
838 If your environment has not been set up so that an appropriate library
839 is automatically included, you may need to add something like
841 LIBS="-ncurses"
843 immediately before the configure command.
848 pcreapi(3), pcre16, pcre_config(3).
853 Philip Hazel
854 University Computing Service
855 Cambridge CB2 3QH, England.
860 Last updated: 07 January 2012
861 Copyright (c) 1997-2012 University of Cambridge.
862 ------------------------------------------------------------------------------
868 NAME
869 PCRE - Perl-compatible regular expressions
874 This document describes the two different algorithms that are available
875 in PCRE for matching a compiled regular expression against a given sub-
876 ject string. The "standard" algorithm is the one provided by the
877 pcre_exec() and pcre16_exec() functions. These work in the same was as
878 Perl's matching function, and provide a Perl-compatible matching opera-
879 tion. The just-in-time (JIT) optimization that is described in the
880 pcrejit documentation is compatible with these functions.
882 An alternative algorithm is provided by the pcre_dfa_exec() and
883 pcre16_dfa_exec() functions; they operate in a different way, and are
884 not Perl-compatible. This alternative has advantages and disadvantages
885 compared with the standard algorithm, and these are described below.
887 When there is only one possible way in which a given subject string can
888 match a pattern, the two algorithms give the same answer. A difference
889 arises, however, when there are multiple possibilities. For example, if
890 the pattern
892 ^<.*>
894 is matched against the string
896 <something> <something else> <something further>
898 there are three possible answers. The standard algorithm finds only one
899 of them, whereas the alternative algorithm finds all three.
904 The set of strings that are matched by a regular expression can be rep-
905 resented as a tree structure. An unlimited repetition in the pattern
906 makes the tree of infinite size, but it is still a tree. Matching the
907 pattern to a given subject string (from a given starting point) can be
908 thought of as a search of the tree. There are two ways to search a
909 tree: depth-first and breadth-first, and these correspond to the two
910 matching algorithms provided by PCRE.
915 In the terminology of Jeffrey Friedl's book "Mastering Regular Expres-
916 sions", the standard algorithm is an "NFA algorithm". It conducts a
917 depth-first search of the pattern tree. That is, it proceeds along a
918 single path through the tree, checking that the subject matches what is
919 required. When there is a mismatch, the algorithm tries any alterna-
920 tives at the current point, and if they all fail, it backs up to the
921 previous branch point in the tree, and tries the next alternative
922 branch at that level. This often involves backing up (moving to the
923 left) in the subject string as well. The order in which repetition
924 branches are tried is controlled by the greedy or ungreedy nature of
925 the quantifier.
927 If a leaf node is reached, a matching string has been found, and at
928 that point the algorithm stops. Thus, if there is more than one possi-
929 ble match, this algorithm returns the first one that it finds. Whether
930 this is the shortest, the longest, or some intermediate length depends
931 on the way the greedy and ungreedy repetition quantifiers are specified
932 in the pattern.
934 Because it ends up with a single path through the tree, it is rela-
935 tively straightforward for this algorithm to keep track of the sub-
936 strings that are matched by portions of the pattern in parentheses.
937 This provides support for capturing parentheses and back references.
942 This algorithm conducts a breadth-first search of the tree. Starting
943 from the first matching point in the subject, it scans the subject
944 string from left to right, once, character by character, and as it does
945 this, it remembers all the paths through the tree that represent valid
946 matches. In Friedl's terminology, this is a kind of "DFA algorithm",
947 though it is not implemented as a traditional finite state machine (it
948 keeps multiple states active simultaneously).
950 Although the general principle of this matching algorithm is that it
951 scans the subject string only once, without backtracking, there is one
952 exception: when a lookaround assertion is encountered, the characters
953 following or preceding the current point have to be independently
954 inspected.
956 The scan continues until either the end of the subject is reached, or
957 there are no more unterminated paths. At this point, terminated paths
958 represent the different matching possibilities (if there are none, the
959 match has failed). Thus, if there is more than one possible match,
960 this algorithm finds all of them, and in particular, it finds the long-
961 est. The matches are returned in decreasing order of length. There is
962 an option to stop the algorithm after the first match (which is neces-
963 sarily the shortest) is found.
965 Note that all the matches that are found start at the same point in the
966 subject. If the pattern
968 cat(er(pillar)?)?
970 is matched against the string "the caterpillar catchment", the result
971 will be the three strings "caterpillar", "cater", and "cat" that start
972 at the fifth character of the subject. The algorithm does not automati-
973 cally move on to find matches that start at later positions.
975 There are a number of features of PCRE regular expressions that are not
976 supported by the alternative matching algorithm. They are as follows:
978 1. Because the algorithm finds all possible matches, the greedy or
979 ungreedy nature of repetition quantifiers is not relevant. Greedy and
980 ungreedy quantifiers are treated in exactly the same way. However, pos-
981 sessive quantifiers can make a difference when what follows could also
982 match what is quantified, for example in a pattern like this:
984 ^a++\w!
986 This pattern matches "aaab!" but not "aaa!", which would be matched by
987 a non-possessive quantifier. Similarly, if an atomic group is present,
988 it is matched as if it were a standalone pattern at the current point,
989 and the longest match is then "locked in" for the rest of the overall
990 pattern.
992 2. When dealing with multiple paths through the tree simultaneously, it
993 is not straightforward to keep track of captured substrings for the
994 different matching possibilities, and PCRE's implementation of this
995 algorithm does not attempt to do this. This means that no captured sub-
996 strings are available.
998 3. Because no substrings are captured, back references within the pat-
999 tern are not supported, and cause errors if encountered.
1001 4. For the same reason, conditional expressions that use a backrefer-
1002 ence as the condition or test for a specific group recursion are not
1003 supported.
1005 5. Because many paths through the tree may be active, the \K escape
1006 sequence, which resets the start of the match when encountered (but may
1007 be on some paths and not on others), is not supported. It causes an
1008 error if encountered.
1010 6. Callouts are supported, but the value of the capture_top field is
1011 always 1, and the value of the capture_last field is always -1.
1013 7. The \C escape sequence, which (in the standard algorithm) always
1014 matches a single data unit, even in UTF-8 or UTF-16 modes, is not sup-
1015 ported in these modes, because the alternative algorithm moves through
1016 the subject string one character (not data unit) at a time, for all
1017 active paths through the tree.
1019 8. Except for (*FAIL), the backtracking control verbs such as (*PRUNE)
1020 are not supported. (*FAIL) is supported, and behaves like a failing
1021 negative assertion.
1026 Using the alternative matching algorithm provides the following advan-
1027 tages:
1029 1. All possible matches (at a single point in the subject) are automat-
1030 ically found, and in particular, the longest match is found. To find
1031 more than one match using the standard algorithm, you have to do kludgy
1032 things with callouts.
1034 2. Because the alternative algorithm scans the subject string just
1035 once, and never needs to backtrack (except for lookbehinds), it is pos-
1036 sible to pass very long subject strings to the matching function in
1037 several pieces, checking for partial matching each time. Although it is
1038 possible to do multi-segment matching using the standard algorithm by
1039 retaining partially matched substrings, it is more complicated. The
1040 pcrepartial documentation gives details of partial matching and dis-
1041 cusses multi-segment matching.
1046 The alternative algorithm suffers from a number of disadvantages:
1048 1. It is substantially slower than the standard algorithm. This is
1049 partly because it has to search for all possible matches, but is also
1050 because it is less susceptible to optimization.
1052 2. Capturing parentheses and back references are not supported.
1054 3. Although atomic groups are supported, their use does not provide the
1055 performance advantage that it does for the standard algorithm.
1060 Philip Hazel
1061 University Computing Service
1062 Cambridge CB2 3QH, England.
1067 Last updated: 08 January 2012
1068 Copyright (c) 1997-2012 University of Cambridge.
1069 ------------------------------------------------------------------------------
1075 NAME
1076 PCRE - Perl-compatible regular expressions
1078 #include <pcre.h>
1083 pcre *pcre_compile(const char *pattern, int options,
1084 const char **errptr, int *erroffset,
1085 const unsigned char *tableptr);
1087 pcre *pcre_compile2(const char *pattern, int options,
1088 int *errorcodeptr,
1089 const char **errptr, int *erroffset,
1090 const unsigned char *tableptr);
1092 pcre_extra *pcre_study(const pcre *code, int options,
1093 const char **errptr);
1095 void pcre_free_study(pcre_extra *extra);
1097 int pcre_exec(const pcre *code, const pcre_extra *extra,
1098 const char *subject, int length, int startoffset,
1099 int options, int *ovector, int ovecsize);
1101 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
1102 const char *subject, int length, int startoffset,
1103 int options, int *ovector, int ovecsize,
1104 int *workspace, int wscount);
1109 int pcre_copy_named_substring(const pcre *code,
1110 const char *subject, int *ovector,
1111 int stringcount, const char *stringname,
1112 char *buffer, int buffersize);
1114 int pcre_copy_substring(const char *subject, int *ovector,
1115 int stringcount, int stringnumber, char *buffer,
1116 int buffersize);
1118 int pcre_get_named_substring(const pcre *code,
1119 const char *subject, int *ovector,
1120 int stringcount, const char *stringname,
1121 const char **stringptr);
1123 int pcre_get_stringnumber(const pcre *code,
1124 const char *name);
1126 int pcre_get_stringtable_entries(const pcre *code,
1127 const char *name, char **first, char **last);
1129 int pcre_get_substring(const char *subject, int *ovector,
1130 int stringcount, int stringnumber,
1131 const char **stringptr);
1133 int pcre_get_substring_list(const char *subject,
1134 int *ovector, int stringcount, const char ***listptr);
1136 void pcre_free_substring(const char *stringptr);
1138 void pcre_free_substring_list(const char **stringptr);
1143 pcre_jit_stack *pcre_jit_stack_alloc(int startsize, int maxsize);
1145 void pcre_jit_stack_free(pcre_jit_stack *stack);
1147 void pcre_assign_jit_stack(pcre_extra *extra,
1148 pcre_jit_callback callback, void *data);
1150 const unsigned char *pcre_maketables(void);
1152 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1153 int what, void *where);
1155 int pcre_refcount(pcre *code, int adjust);
1157 int pcre_config(int what, void *where);
1159 const char *pcre_version(void);
1161 int pcre_pattern_to_host_byte_order(pcre *code,
1162 pcre_extra *extra, const unsigned char *tables);
1167 void *(*pcre_malloc)(size_t);
1169 void (*pcre_free)(void *);
1171 void *(*pcre_stack_malloc)(size_t);
1173 void (*pcre_stack_free)(void *);
1175 int (*pcre_callout)(pcre_callout_block *);
1180 From release 8.30, PCRE can be compiled as a library for handling
1181 16-bit character strings as well as, or instead of, the original
1182 library that handles 8-bit character strings. To avoid too much compli-
1183 cation, this document describes the 8-bit versions of the functions,
1184 with only occasional references to the 16-bit library.
1186 The 16-bit functions operate in the same way as their 8-bit counter-
1187 parts; they just use different data types for their arguments and
1188 results, and their names start with pcre16_ instead of pcre_. For every
1189 option that has UTF8 in its name (for example, PCRE_UTF8), there is a
1190 corresponding 16-bit name with UTF8 replaced by UTF16. This facility is
1191 in fact just cosmetic; the 16-bit option names define the same bit val-
1192 ues.
1194 References to bytes and UTF-8 in this document should be read as refer-
1195 ences to 16-bit data quantities and UTF-16 when using the 16-bit
1196 library, unless specified otherwise. More details of the specific dif-
1197 ferences for the 16-bit library are given in the pcre16 page.
1202 PCRE has its own native API, which is described in this document. There
1203 are also some wrapper functions (for the 8-bit library only) that cor-
1204 respond to the POSIX regular expression API, but they do not give
1205 access to all the functionality. They are described in the pcreposix
1206 documentation. Both of these APIs define a set of C function calls. A
1207 C++ wrapper (again for the 8-bit library only) is also distributed with
1208 PCRE. It is documented in the pcrecpp page.
1210 The native API C function prototypes are defined in the header file
1211 pcre.h, and on Unix-like systems the (8-bit) library itself is called
1212 libpcre. It can normally be accessed by adding -lpcre to the command
1213 for linking an application that uses PCRE. The header file defines the
1214 macros PCRE_MAJOR and PCRE_MINOR to contain the major and minor release
1215 numbers for the library. Applications can use these to include support
1216 for different releases of PCRE.
1218 In a Windows environment, if you want to statically link an application
1219 program against a non-dll pcre.a file, you must define PCRE_STATIC
1220 before including pcre.h or pcrecpp.h, because otherwise the pcre_mal-
1221 loc() and pcre_free() exported functions will be declared
1222 __declspec(dllimport), with unwanted results.
1224 The functions pcre_compile(), pcre_compile2(), pcre_study(), and
1225 pcre_exec() are used for compiling and matching regular expressions in
1226 a Perl-compatible manner. A sample program that demonstrates the sim-
1227 plest way of using them is provided in the file called pcredemo.c in
1228 the PCRE source distribution. A listing of this program is given in the
1229 pcredemo documentation, and the pcresample documentation describes how
1230 to compile and run it.
1232 Just-in-time compiler support is an optional feature of PCRE that can
1233 be built in appropriate hardware environments. It greatly speeds up the
1234 matching performance of many patterns. Simple programs can easily
1235 request that it be used if available, by setting an option that is
1236 ignored when it is not relevant. More complicated programs might need
1237 to make use of the functions pcre_jit_stack_alloc(),
1238 pcre_jit_stack_free(), and pcre_assign_jit_stack() in order to control
1239 the JIT code's memory usage. These functions are discussed in the
1240 pcrejit documentation.
1242 A second matching function, pcre_dfa_exec(), which is not Perl-compati-
1243 ble, is also provided. This uses a different algorithm for the match-
1244 ing. The alternative algorithm finds all possible matches (at a given
1245 point in the subject), and scans the subject just once (unless there
1246 are lookbehind assertions). However, this algorithm does not return
1247 captured substrings. A description of the two matching algorithms and
1248 their advantages and disadvantages is given in the pcrematching docu-
1249 mentation.
1251 In addition to the main compiling and matching functions, there are
1252 convenience functions for extracting captured substrings from a subject
1253 string that is matched by pcre_exec(). They are:
1255 pcre_copy_substring()
1256 pcre_copy_named_substring()
1257 pcre_get_substring()
1258 pcre_get_named_substring()
1259 pcre_get_substring_list()
1260 pcre_get_stringnumber()
1261 pcre_get_stringtable_entries()
1263 pcre_free_substring() and pcre_free_substring_list() are also provided,
1264 to free the memory used for extracted strings.
1266 The function pcre_maketables() is used to build a set of character
1267 tables in the current locale for passing to pcre_compile(),
1268 pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
1269 provided for specialist use. Most commonly, no special tables are
1270 passed, in which case internal tables that are generated when PCRE is
1271 built are used.
1273 The function pcre_fullinfo() is used to find out information about a
1274 compiled pattern. The function pcre_version() returns a pointer to a
1275 string containing the version of PCRE and its date of release.
1277 The function pcre_refcount() maintains a reference count in a data
1278 block containing a compiled pattern. This is provided for the benefit
1279 of object-oriented applications.
1281 The global variables pcre_malloc and pcre_free initially contain the
1282 entry points of the standard malloc() and free() functions, respec-
1283 tively. PCRE calls the memory management functions via these variables,
1284 so a calling program can replace them if it wishes to intercept the
1285 calls. This should be done before calling any PCRE functions.
1287 The global variables pcre_stack_malloc and pcre_stack_free are also
1288 indirections to memory management functions. These special functions
1289 are used only when PCRE is compiled to use the heap for remembering
1290 data, instead of recursive function calls, when running the pcre_exec()
1291 function. See the pcrebuild documentation for details of how to do
1292 this. It is a non-standard way of building PCRE, for use in environ-
1293 ments that have limited stacks. Because of the greater use of memory
1294 management, it runs more slowly. Separate functions are provided so
1295 that special-purpose external code can be used for this case. When
1296 used, these functions are always called in a stack-like manner (last
1297 obtained, first freed), and always for memory blocks of the same size.
1298 There is a discussion about PCRE's stack usage in the pcrestack docu-
1299 mentation.
1301 The global variable pcre_callout initially contains NULL. It can be set
1302 by the caller to a "callout" function, which PCRE will then call at
1303 specified points during a matching operation. Details are given in the
1304 pcrecallout documentation.
1309 PCRE supports five different conventions for indicating line breaks in
1310 strings: a single CR (carriage return) character, a single LF (line-
1311 feed) character, the two-character sequence CRLF, any of the three pre-
1312 ceding, or any Unicode newline sequence. The Unicode newline sequences
1313 are the three just mentioned, plus the single characters VT (vertical
1314 tab, U+000B), FF (formfeed, U+000C), NEL (next line, U+0085), LS (line
1315 separator, U+2028), and PS (paragraph separator, U+2029).
1317 Each of the first three conventions is used by at least one operating
1318 system as its standard newline sequence. When PCRE is built, a default
1319 can be specified. The default default is LF, which is the Unix stan-
1320 dard. When PCRE is run, the default can be overridden, either when a
1321 pattern is compiled, or when it is matched.
1323 At compile time, the newline convention can be specified by the options
1324 argument of pcre_compile(), or it can be specified by special text at
1325 the start of the pattern itself; this overrides any other settings. See
1326 the pcrepattern page for details of the special character sequences.
1328 In the PCRE documentation the word "newline" is used to mean "the char-
1329 acter or pair of characters that indicate a line break". The choice of
1330 newline convention affects the handling of the dot, circumflex, and
1331 dollar metacharacters, the handling of #-comments in /x mode, and, when
1332 CRLF is a recognized line ending sequence, the match position advance-
1333 ment for a non-anchored pattern. There is more detail about this in the
1334 section on pcre_exec() options below.
1336 The choice of newline convention does not affect the interpretation of
1337 the \n or \r escape sequences, nor does it affect what \R matches,
1338 which is controlled in a similar way, but by separate options.
1343 The PCRE functions can be used in multi-threading applications, with
1344 the proviso that the memory management functions pointed to by
1345 pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
1346 callout function pointed to by pcre_callout, are shared by all threads.
1348 The compiled form of a regular expression is not altered during match-
1349 ing, so the same compiled pattern can safely be used by several threads
1350 at once.
1352 If the just-in-time optimization feature is being used, it needs sepa-
1353 rate memory stack areas for each thread. See the pcrejit documentation
1354 for more details.
1359 The compiled form of a regular expression can be saved and re-used at a
1360 later time, possibly by a different program, and even on a host other
1361 than the one on which it was compiled. Details are given in the
1362 pcreprecompile documentation, which includes a description of the
1363 pcre_pattern_to_host_byte_order() function. However, compiling a regu-
1364 lar expression with one version of PCRE for use with a different ver-
1365 sion is not guaranteed to work and may cause crashes.
1370 int pcre_config(int what, void *where);
1372 The function pcre_config() makes it possible for a PCRE client to dis-
1373 cover which optional features have been compiled into the PCRE library.
1374 The pcrebuild documentation has more details about these optional fea-
1375 tures.
1377 The first argument for pcre_config() is an integer, specifying which
1378 information is required; the second argument is a pointer to a variable
1379 into which the information is placed. The returned value is zero on
1380 success, or the negative error code PCRE_ERROR_BADOPTION if the value
1381 in the first argument is not recognized. The following information is
1382 available:
1386 The output is an integer that is set to one if UTF-8 support is avail-
1387 able; otherwise it is set to zero. If this option is given to the
1388 16-bit version of this function, pcre16_config(), the result is
1393 The output is an integer that is set to one if UTF-16 support is avail-
1394 able; otherwise it is set to zero. This value should normally be given
1395 to the 16-bit version of this function, pcre16_config(). If it is given
1396 to the 8-bit version of this function, the result is PCRE_ERROR_BADOP-
1397 TION.
1401 The output is an integer that is set to one if support for Unicode
1402 character properties is available; otherwise it is set to zero.
1406 The output is an integer that is set to one if support for just-in-time
1407 compiling is available; otherwise it is set to zero.
1411 The output is a pointer to a zero-terminated "const char *" string. If
1412 JIT support is available, the string contains the name of the architec-
1413 ture for which the JIT compiler is configured, for example "x86 32bit
1414 (little endian + unaligned)". If JIT support is not available, the
1415 result is NULL.
1419 The output is an integer whose value specifies the default character
1420 sequence that is recognized as meaning "newline". The four values that
1421 are supported are: 10 for LF, 13 for CR, 3338 for CRLF, -2 for ANYCRLF,
1422 and -1 for ANY. Though they are derived from ASCII, the same values
1423 are returned in EBCDIC environments. The default should normally corre-
1424 spond to the standard sequence for your operating system.
1428 The output is an integer whose value indicates what character sequences
1429 the \R escape sequence matches by default. A value of 0 means that \R
1430 matches any Unicode line ending sequence; a value of 1 means that \R
1431 matches only CR, LF, or CRLF. The default can be overridden when a pat-
1432 tern is compiled or matched.
1436 The output is an integer that contains the number of bytes used for
1437 internal linkage in compiled regular expressions. For the 8-bit
1438 library, the value can be 2, 3, or 4. For the 16-bit library, the value
1439 is either 2 or 4 and is still a number of bytes. The default value of 2
1440 is sufficient for all but the most massive patterns, since it allows
1441 the compiled pattern to be up to 64K in size. Larger values allow
1442 larger regular expressions to be compiled, at the expense of slower
1443 matching.
1447 The output is an integer that contains the threshold above which the
1448 POSIX interface uses malloc() for output vectors. Further details are
1449 given in the pcreposix documentation.
1453 The output is a long integer that gives the default limit for the num-
1454 ber of internal matching function calls in a pcre_exec() execution.
1455 Further details are given with pcre_exec() below.
1459 The output is a long integer that gives the default limit for the depth
1460 of recursion when calling the internal matching function in a
1461 pcre_exec() execution. Further details are given with pcre_exec()
1462 below.
1466 The output is an integer that is set to one if internal recursion when
1467 running pcre_exec() is implemented by recursive function calls that use
1468 the stack to remember their state. This is the usual way that PCRE is
1469 compiled. The output is zero if PCRE was compiled to use blocks of data
1470 on the heap instead of recursive function calls. In this case,
1471 pcre_stack_malloc and pcre_stack_free are called to manage memory
1472 blocks on the heap, thus avoiding the use of the stack.
1477 pcre *pcre_compile(const char *pattern, int options,
1478 const char **errptr, int *erroffset,
1479 const unsigned char *tableptr);
1481 pcre *pcre_compile2(const char *pattern, int options,
1482 int *errorcodeptr,
1483 const char **errptr, int *erroffset,
1484 const unsigned char *tableptr);
1486 Either of the functions pcre_compile() or pcre_compile2() can be called
1487 to compile a pattern into an internal form. The only difference between
1488 the two interfaces is that pcre_compile2() has an additional argument,
1489 errorcodeptr, via which a numerical error code can be returned. To
1490 avoid too much repetition, we refer just to pcre_compile() below, but
1491 the information applies equally to pcre_compile2().
1493 The pattern is a C string terminated by a binary zero, and is passed in
1494 the pattern argument. A pointer to a single block of memory that is
1495 obtained via pcre_malloc is returned. This contains the compiled code
1496 and related data. The pcre type is defined for the returned block; this
1497 is a typedef for a structure whose contents are not externally defined.
1498 It is up to the caller to free the memory (via pcre_free) when it is no
1499 longer required.
1501 Although the compiled code of a PCRE regex is relocatable, that is, it
1502 does not depend on memory location, the complete pcre data block is not
1503 fully relocatable, because it may contain a copy of the tableptr argu-
1504 ment, which is an address (see below).
1506 The options argument contains various bit settings that affect the com-
1507 pilation. It should be zero if no options are required. The available
1508 options are described below. Some of them (in particular, those that
1509 are compatible with Perl, but some others as well) can also be set and
1510 unset from within the pattern (see the detailed description in the
1511 pcrepattern documentation). For those options that can be different in
1512 different parts of the pattern, the contents of the options argument
1513 specifies their settings at the start of compilation and execution. The
1515 PCRE_NO_START_OPTIMIZE options can be set at the time of matching as
1516 well as at compile time.
1518 If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
1519 if compilation of a pattern fails, pcre_compile() returns NULL, and
1520 sets the variable pointed to by errptr to point to a textual error mes-
1521 sage. This is a static string that is part of the library. You must not
1522 try to free it. Normally, the offset from the start of the pattern to
1523 the byte that was being processed when the error was discovered is
1524 placed in the variable pointed to by erroffset, which must not be NULL
1525 (if it is, an immediate error is given). However, for an invalid UTF-8
1526 string, the offset is that of the first byte of the failing character.
1528 Some errors are not detected until the whole pattern has been scanned;
1529 in these cases, the offset passed back is the length of the pattern.
1530 Note that the offset is in bytes, not characters, even in UTF-8 mode.
1531 It may sometimes point into the middle of a UTF-8 character.
1533 If pcre_compile2() is used instead of pcre_compile(), and the error-
1534 codeptr argument is not NULL, a non-zero error code number is returned
1535 via this argument in the event of an error. This is in addition to the
1536 textual error message. Error codes and messages are listed below.
1538 If the final argument, tableptr, is NULL, PCRE uses a default set of
1539 character tables that are built when PCRE is compiled, using the
1540 default C locale. Otherwise, tableptr must be an address that is the
1541 result of a call to pcre_maketables(). This value is stored with the
1542 compiled pattern, and used again by pcre_exec(), unless another table
1543 pointer is passed to it. For more discussion, see the section on locale
1544 support below.
1546 This code fragment shows a typical straightforward call to pcre_com-
1547 pile():
1549 pcre *re;
1550 const char *error;
1551 int erroffset;
1552 re = pcre_compile(
1553 "^A.*Z", /* the pattern */
1554 0, /* default options */
1555 &error, /* for error message */
1556 &erroffset, /* for error offset */
1557 NULL); /* use default character tables */
1559 The following names for option bits are defined in the pcre.h header
1560 file:
1564 If this bit is set, the pattern is forced to be "anchored", that is, it
1565 is constrained to match only at the first matching point in the string
1566 that is being searched (the "subject string"). This effect can also be
1567 achieved by appropriate constructs in the pattern itself, which is the
1568 only way to do it in Perl.
1572 If this bit is set, pcre_compile() automatically inserts callout items,
1573 all with number 255, before each pattern item. For discussion of the
1574 callout facility, see the pcrecallout documentation.
1579 These options (which are mutually exclusive) control what the \R escape
1580 sequence matches. The choice is either to match only CR, LF, or CRLF,
1581 or to match any Unicode newline sequence. The default is specified when
1582 PCRE is built. It can be overridden from within the pattern, or by set-
1583 ting an option when a compiled pattern is matched.
1587 If this bit is set, letters in the pattern match both upper and lower
1588 case letters. It is equivalent to Perl's /i option, and it can be
1589 changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
1590 always understands the concept of case for characters whose values are
1591 less than 128, so caseless matching is always possible. For characters
1592 with higher values, the concept of case is supported if PCRE is com-
1593 piled with Unicode property support, but not otherwise. If you want to
1594 use caseless matching for characters 128 and above, you must ensure
1595 that PCRE is compiled with Unicode property support as well as with
1596 UTF-8 support.
1600 If this bit is set, a dollar metacharacter in the pattern matches only
1601 at the end of the subject string. Without this option, a dollar also
1602 matches immediately before a newline at the end of the string (but not
1603 before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
1604 if PCRE_MULTILINE is set. There is no equivalent to this option in
1605 Perl, and no way to set it within a pattern.
1609 If this bit is set, a dot metacharacter in the pattern matches a char-
1610 acter of any value, including one that indicates a newline. However, it
1611 only ever matches one character, even if newlines are coded as CRLF.
1612 Without this option, a dot does not match when the current position is
1613 at a newline. This option is equivalent to Perl's /s option, and it can
1614 be changed within a pattern by a (?s) option setting. A negative class
1615 such as [^a] always matches newline characters, independent of the set-
1616 ting of this option.
1620 If this bit is set, names used to identify capturing subpatterns need
1621 not be unique. This can be helpful for certain types of pattern when it
1622 is known that only one instance of the named subpattern can ever be
1623 matched. There are more details of named subpatterns below; see also
1624 the pcrepattern documentation.
1628 If this bit is set, whitespace data characters in the pattern are
1629 totally ignored except when escaped or inside a character class. White-
1630 space does not include the VT character (code 11). In addition, charac-
1631 ters between an unescaped # outside a character class and the next new-
1632 line, inclusive, are also ignored. This is equivalent to Perl's /x
1633 option, and it can be changed within a pattern by a (?x) option set-
1634 ting.
1636 Which characters are interpreted as newlines is controlled by the
1637 options passed to pcre_compile() or by a special sequence at the start
1638 of the pattern, as described in the section entitled "Newline conven-
1639 tions" in the pcrepattern documentation. Note that the end of this type
1640 of comment is a literal newline sequence in the pattern; escape
1641 sequences that happen to represent a newline do not count.
1643 This option makes it possible to include comments inside complicated
1644 patterns. Note, however, that this applies only to data characters.
1645 Whitespace characters may never appear within special character
1646 sequences in a pattern, for example within the sequence (?( that intro-
1647 duces a conditional subpattern.
1651 This option was invented in order to turn on additional functionality
1652 of PCRE that is incompatible with Perl, but it is currently of very
1653 little use. When set, any backslash in a pattern that is followed by a
1654 letter that has no special meaning causes an error, thus reserving
1655 these combinations for future expansion. By default, as in Perl, a
1656 backslash followed by a letter with no special meaning is treated as a
1657 literal. (Perl can, however, be persuaded to give an error for this, by
1658 running it with the -w option.) There are at present no other features
1659 controlled by this option. It can also be set by a (?X) option setting
1660 within a pattern.
1664 If this option is set, an unanchored pattern is required to match
1665 before or at the first newline in the subject string, though the
1666 matched text may continue over the newline.
1670 If this option is set, PCRE's behaviour is changed in some ways so that
1671 it is compatible with JavaScript rather than Perl. The changes are as
1672 follows:
1674 (1) A lone closing square bracket in a pattern causes a compile-time
1675 error, because this is illegal in JavaScript (by default it is treated
1676 as a data character). Thus, the pattern AB]CD becomes illegal when this
1677 option is set.
1679 (2) At run time, a back reference to an unset subpattern group matches
1680 an empty string (by default this causes the current matching alterna-
1681 tive to fail). A pattern such as (\1)(a) succeeds when this option is
1682 set (assuming it can find an "a" in the subject), whereas it fails by
1683 default, for Perl compatibility.
1685 (3) \U matches an upper case "U" character; by default \U causes a com-
1686 pile time error (Perl uses \U to upper case subsequent characters).
1688 (4) \u matches a lower case "u" character unless it is followed by four
1689 hexadecimal digits, in which case the hexadecimal number defines the
1690 code point to match. By default, \u causes a compile time error (Perl
1691 uses it to upper case the following character).
1693 (5) \x matches a lower case "x" character unless it is followed by two
1694 hexadecimal digits, in which case the hexadecimal number defines the
1695 code point to match. By default, as in Perl, a hexadecimal number is
1696 always expected after \x, but it may have zero, one, or two digits (so,
1697 for example, \xz matches a binary zero character followed by z).
1701 By default, PCRE treats the subject string as consisting of a single
1702 line of characters (even if it actually contains newlines). The "start
1703 of line" metacharacter (^) matches only at the start of the string,
1704 while the "end of line" metacharacter ($) matches only at the end of
1705 the string, or before a terminating newline (unless PCRE_DOLLAR_ENDONLY
1706 is set). This is the same as Perl.
1708 When PCRE_MULTILINE it is set, the "start of line" and "end of line"
1709 constructs match immediately following or immediately before internal
1710 newlines in the subject string, respectively, as well as at the very
1711 start and end. This is equivalent to Perl's /m option, and it can be
1712 changed within a pattern by a (?m) option setting. If there are no new-
1713 lines in a subject string, or no occurrences of ^ or $ in a pattern,
1714 setting PCRE_MULTILINE has no effect.
1722 These options override the default newline definition that was chosen
1723 when PCRE was built. Setting the first or the second specifies that a
1724 newline is indicated by a single character (CR or LF, respectively).
1725 Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
1726 two-character CRLF sequence. Setting PCRE_NEWLINE_ANYCRLF specifies
1727 that any of the three preceding sequences should be recognized. Setting
1728 PCRE_NEWLINE_ANY specifies that any Unicode newline sequence should be
1729 recognized. The Unicode newline sequences are the three just mentioned,
1730 plus the single characters VT (vertical tab, U+000B), FF (formfeed,
1731 U+000C), NEL (next line, U+0085), LS (line separator, U+2028), and PS
1732 (paragraph separator, U+2029). For the 8-bit library, the last two are
1733 recognized only in UTF-8 mode.
1735 The newline setting in the options word uses three bits that are
1736 treated as a number, giving eight possibilities. Currently only six are
1737 used (default plus the five values above). This means that if you set
1738 more than one newline option, the combination may or may not be sensi-
1739 ble. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equivalent to
1740 PCRE_NEWLINE_CRLF, but other combinations may yield unused numbers and
1741 cause an error.
1743 The only time that a line break in a pattern is specially recognized
1744 when compiling is when PCRE_EXTENDED is set. CR and LF are whitespace
1745 characters, and so are ignored in this mode. Also, an unescaped # out-
1746 side a character class indicates a comment that lasts until after the
1747 next line break sequence. In other circumstances, line break sequences
1748 in patterns are treated as literal data.
1750 The newline option that is set at compile time becomes the default that
1751 is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
1755 If this option is set, it disables the use of numbered capturing paren-
1756 theses in the pattern. Any opening parenthesis that is not followed by
1757 ? behaves as if it were followed by ?: but named parentheses can still
1758 be used for capturing (and they acquire numbers in the usual way).
1759 There is no equivalent of this option in Perl.
1763 This is an option that acts at matching time; that is, it is really an
1764 option for pcre_exec() or pcre_dfa_exec(). If it is set at compile
1765 time, it is remembered with the compiled pattern and assumed at match-
1766 ing time. For details see the discussion of PCRE_NO_START_OPTIMIZE
1767 below.
1771 This option changes the way PCRE processes \B, \b, \D, \d, \S, \s, \W,
1772 \w, and some of the POSIX character classes. By default, only ASCII
1773 characters are recognized, but if PCRE_UCP is set, Unicode properties
1774 are used instead to classify characters. More details are given in the
1775 section on generic character types in the pcrepattern page. If you set
1776 PCRE_UCP, matching one of the items it affects takes much longer. The
1777 option is available only if PCRE has been compiled with Unicode prop-
1778 erty support.
1782 This option inverts the "greediness" of the quantifiers so that they
1783 are not greedy by default, but become greedy if followed by "?". It is
1784 not compatible with Perl. It can also be set by a (?U) option setting
1785 within the pattern.
1787 PCRE_UTF8
1789 This option causes PCRE to regard both the pattern and the subject as
1790 strings of UTF-8 characters instead of single-byte strings. However, it
1791 is available only when PCRE is built to include UTF support. If not,
1792 the use of this option provokes an error. Details of how this option
1793 changes the behaviour of PCRE are given in the pcreunicode page.
1797 When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
1798 automatically checked. There is a discussion about the validity of
1799 UTF-8 strings in the pcreunicode page. If an invalid UTF-8 sequence is
1800 found, pcre_compile() returns an error. If you already know that your
1801 pattern is valid, and you want to skip this check for performance rea-
1802 sons, you can set the PCRE_NO_UTF8_CHECK option. When it is set, the
1803 effect of passing an invalid UTF-8 string as a pattern is undefined. It
1804 may cause your program to crash. Note that this option can also be
1805 passed to pcre_exec() and pcre_dfa_exec(), to suppress the validity
1806 checking of subject strings.
1811 The following table lists the error codes than may be returned by
1812 pcre_compile2(), along with the error messages that may be returned by
1813 both compiling functions. Note that error messages are always 8-bit
1814 ASCII strings, even in 16-bit mode. As PCRE has developed, some error
1815 codes have fallen out of use. To avoid confusion, they have not been
1816 re-used.
1818 0 no error
1819 1 \ at end of pattern
1820 2 \c at end of pattern
1821 3 unrecognized character follows \
1822 4 numbers out of order in {} quantifier
1823 5 number too big in {} quantifier
1824 6 missing terminating ] for character class
1825 7 invalid escape sequence in character class
1826 8 range out of order in character class
1827 9 nothing to repeat
1828 10 [this code is not in use]
1829 11 internal error: unexpected repeat
1830 12 unrecognized character after (? or (?-
1831 13 POSIX named classes are supported only within a class
1832 14 missing )
1833 15 reference to non-existent subpattern
1834 16 erroffset passed as NULL
1835 17 unknown option bit(s) set
1836 18 missing ) after comment
1837 19 [this code is not in use]
1838 20 regular expression is too large
1839 21 failed to get memory
1840 22 unmatched parentheses
1841 23 internal error: code overflow
1842 24 unrecognized character after (?<
1843 25 lookbehind assertion is not fixed length
1844 26 malformed number or name after (?(
1845 27 conditional group contains more than two branches
1846 28 assertion expected after (?(
1847 29 (?R or (?[+-]digits must be followed by )
1848 30 unknown POSIX class name
1849 31 POSIX collating elements are not supported
1850 32 this version of PCRE is compiled without UTF support
1851 33 [this code is not in use]
1852 34 character value in \x{...} sequence is too large
1853 35 invalid condition (?(0)
1854 36 \C not allowed in lookbehind assertion
1855 37 PCRE does not support \L, \l, \N{name}, \U, or \u
1856 38 number after (?C is > 255
1857 39 closing ) for (?C expected
1858 40 recursive call could loop indefinitely
1859 41 unrecognized character after (?P
1860 42 syntax error in subpattern name (missing terminator)
1861 43 two named subpatterns have the same name
1862 44 invalid UTF-8 string (specifically UTF-8)
1863 45 support for \P, \p, and \X has not been compiled
1864 46 malformed \P or \p sequence
1865 47 unknown property name after \P or \p
1866 48 subpattern name is too long (maximum 32 characters)
1867 49 too many named subpatterns (maximum 10000)
1868 50 [this code is not in use]
1869 51 octal value is greater than \377 in 8-bit non-UTF-8 mode
1870 52 internal error: overran compiling workspace
1871 53 internal error: previously-checked referenced subpattern
1872 not found
1873 54 DEFINE group contains more than one branch
1874 55 repeating a DEFINE group is not allowed
1875 56 inconsistent NEWLINE options
1876 57 \g is not followed by a braced, angle-bracketed, or quoted
1877 name/number or by a plain number
1878 58 a numbered reference must not be zero
1879 59 an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)
1880 60 (*VERB) not recognized
1881 61 number is too big
1882 62 subpattern name expected
1883 63 digit expected after (?+
1884 64 ] is an invalid data character in JavaScript compatibility mode
1885 65 different names for subpatterns of the same number are
1886 not allowed
1887 66 (*MARK) must have an argument
1888 67 this version of PCRE is not compiled with Unicode property
1889 support
1890 68 \c must be followed by an ASCII character
1891 69 \k is not followed by a braced, angle-bracketed, or quoted name
1892 70 internal error: unknown opcode in find_fixedlength()
1893 71 \N is not supported in a class
1894 72 too many forward references
1895 73 disallowed Unicode code point (>= 0xd800 && <= 0xdfff)
1896 74 invalid UTF-16 string (specifically UTF-16)
1898 The numbers 32 and 10000 in errors 48 and 49 are defaults; different
1899 values may be used if the limits were changed when PCRE was built.
1904 pcre_extra *pcre_study(const pcre *code, int options
1905 const char **errptr);
1907 If a compiled pattern is going to be used several times, it is worth
1908 spending more time analyzing it in order to speed up the time taken for
1909 matching. The function pcre_study() takes a pointer to a compiled pat-
1910 tern as its first argument. If studying the pattern produces additional
1911 information that will help speed up matching, pcre_study() returns a
1912 pointer to a pcre_extra block, in which the study_data field points to
1913 the results of the study.
1915 The returned value from pcre_study() can be passed directly to
1916 pcre_exec() or pcre_dfa_exec(). However, a pcre_extra block also con-
1917 tains other fields that can be set by the caller before the block is
1918 passed; these are described below in the section on matching a pattern.
1920 If studying the pattern does not produce any useful information,
1921 pcre_study() returns NULL. In that circumstance, if the calling program
1922 wants to pass any of the other fields to pcre_exec() or
1923 pcre_dfa_exec(), it must set up its own pcre_extra block.
1925 The second argument of pcre_study() contains option bits. There are
1926 three options:
1932 If any of these are set, and the just-in-time compiler is available,
1933 the pattern is further compiled into machine code that executes much
1934 faster than the pcre_exec() interpretive matching function. If the
1935 just-in-time compiler is not available, these options are ignored. All
1936 other bits in the options argument must be zero.
1938 JIT compilation is a heavyweight optimization. It can take some time
1939 for patterns to be analyzed, and for one-off matches and simple pat-
1940 terns the benefit of faster execution might be offset by a much slower
1941 study time. Not all patterns can be optimized by the JIT compiler. For
1942 those that cannot be handled, matching automatically falls back to the
1943 pcre_exec() interpreter. For more details, see the pcrejit documenta-
1944 tion.
1946 The third argument for pcre_study() is a pointer for an error message.
1947 If studying succeeds (even if no data is returned), the variable it
1948 points to is set to NULL. Otherwise it is set to point to a textual
1949 error message. This is a static string that is part of the library. You
1950 must not try to free it. You should test the error pointer for NULL
1951 after calling pcre_study(), to be sure that it has run successfully.
1953 When you are finished with a pattern, you can free the memory used for
1954 the study data by calling pcre_free_study(). This function was added to
1955 the API for release 8.20. For earlier versions, the memory could be
1956 freed with pcre_free(), just like the pattern itself. This will still
1957 work in cases where JIT optimization is not used, but it is advisable
1958 to change to the new function when convenient.
1960 This is a typical way in which pcre_study() is used (except that in a
1961 real application there should be tests for errors):
1963 int rc;
1964 pcre *re;
1965 pcre_extra *sd;
1966 re = pcre_compile("pattern", 0, &error, &erroroffset, NULL);
1967 sd = pcre_study(
1968 re, /* result of pcre_compile() */
1969 0, /* no options */
1970 &error); /* set to NULL or points to a message */
1971 rc = pcre_exec( /* see below for details of pcre_exec() options */
1972 re, sd, "subject", 7, 0, 0, ovector, 30);
1973 ...
1974 pcre_free_study(sd);
1975 pcre_free(re);
1977 Studying a pattern does two things: first, a lower bound for the length
1978 of subject string that is needed to match the pattern is computed. This
1979 does not mean that there are any strings of that length that match, but
1980 it does guarantee that no shorter strings match. The value is used by
1981 pcre_exec() and pcre_dfa_exec() to avoid wasting time by trying to
1982 match strings that are shorter than the lower bound. You can find out
1983 the value in a calling program via the pcre_fullinfo() function.
1985 Studying a pattern is also useful for non-anchored patterns that do not
1986 have a single fixed starting character. A bitmap of possible starting
1987 bytes is created. This speeds up finding a position in the subject at
1988 which to start matching. (In 16-bit mode, the bitmap is used for 16-bit
1989 values less than 256.)
1991 These two optimizations apply to both pcre_exec() and pcre_dfa_exec(),
1992 and the information is also used by the JIT compiler. The optimiza-
1993 tions can be disabled by setting the PCRE_NO_START_OPTIMIZE option when
1994 calling pcre_exec() or pcre_dfa_exec(), but if this is done, JIT execu-
1995 tion is also disabled. You might want to do this if your pattern con-
1996 tains callouts or (*MARK) and you want to make use of these facilities
1997 in cases where matching fails. See the discussion of
2003 PCRE handles caseless matching, and determines whether characters are
2004 letters, digits, or whatever, by reference to a set of tables, indexed
2005 by character value. When running in UTF-8 mode, this applies only to
2006 characters with codes less than 128. By default, higher-valued codes
2007 never match escapes such as \w or \d, but they can be tested with \p if
2008 PCRE is built with Unicode character property support. Alternatively,
2009 the PCRE_UCP option can be set at compile time; this causes \w and
2010 friends to use Unicode property support instead of built-in tables. The
2011 use of locales with Unicode is discouraged. If you are handling charac-
2012 ters with codes greater than 128, you should either use UTF-8 and Uni-
2013 code, or use locales, but not try to mix the two.
2015 PCRE contains an internal set of tables that are used when the final
2016 argument of pcre_compile() is NULL. These are sufficient for many
2017 applications. Normally, the internal tables recognize only ASCII char-
2018 acters. However, when PCRE is built, it is possible to cause the inter-
2019 nal tables to be rebuilt in the default "C" locale of the local system,
2020 which may cause them to be different.
2022 The internal tables can always be overridden by tables supplied by the
2023 application that calls PCRE. These may be created in a different locale
2024 from the default. As more and more applications change to using Uni-
2025 code, the need for this locale support is expected to die away.
2027 External tables are built by calling the pcre_maketables() function,
2028 which has no arguments, in the relevant locale. The result can then be
2029 passed to pcre_compile() or pcre_exec() as often as necessary. For
2030 example, to build and use tables that are appropriate for the French
2031 locale (where accented characters with values greater than 128 are
2032 treated as letters), the following code could be used:
2034 setlocale(LC_CTYPE, "fr_FR");
2035 tables = pcre_maketables();
2036 re = pcre_compile(..., tables);
2038 The locale name "fr_FR" is used on Linux and other Unix-like systems;
2039 if you are using Windows, the name for the French locale is "french".
2041 When pcre_maketables() runs, the tables are built in memory that is
2042 obtained via pcre_malloc. It is the caller's responsibility to ensure
2043 that the memory containing the tables remains available for as long as
2044 it is needed.
2046 The pointer that is passed to pcre_compile() is saved with the compiled
2047 pattern, and the same tables are used via this pointer by pcre_study()
2048 and normally also by pcre_exec(). Thus, by default, for any single pat-
2049 tern, compilation, studying and matching all happen in the same locale,
2050 but different patterns can be compiled in different locales.
2052 It is possible to pass a table pointer or NULL (indicating the use of
2053 the internal tables) to pcre_exec(). Although not intended for this
2054 purpose, this facility could be used to match a pattern in a different
2055 locale from the one in which it was compiled. Passing table pointers at
2056 run time is discussed below in the section on matching a pattern.
2061 int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
2062 int what, void *where);
2064 The pcre_fullinfo() function returns information about a compiled pat-
2065 tern. It replaces the pcre_info() function, which was removed from the
2066 library at version 8.30, after more than 10 years of obsolescence.
2068 The first argument for pcre_fullinfo() is a pointer to the compiled
2069 pattern. The second argument is the result of pcre_study(), or NULL if
2070 the pattern was not studied. The third argument specifies which piece
2071 of information is required, and the fourth argument is a pointer to a
2072 variable to receive the data. The yield of the function is zero for
2073 success, or one of the following negative numbers:
2075 PCRE_ERROR_NULL the argument code was NULL
2076 the argument where was NULL
2077 PCRE_ERROR_BADMAGIC the "magic number" was not found
2078 PCRE_ERROR_BADENDIANNESS the pattern was compiled with different
2079 endianness
2080 PCRE_ERROR_BADOPTION the value of what was invalid
2082 The "magic number" is placed at the start of each compiled pattern as
2083 an simple check against passing an arbitrary memory pointer. The endi-
2084 anness error can occur if a compiled pattern is saved and reloaded on a
2085 different host. Here is a typical call of pcre_fullinfo(), to obtain
2086 the length of the compiled pattern:
2088 int rc;
2089 size_t length;
2090 rc = pcre_fullinfo(
2091 re, /* result of pcre_compile() */
2092 sd, /* result of pcre_study(), or NULL */
2093 PCRE_INFO_SIZE, /* what is required */
2094 &length); /* where to put the data */
2096 The possible values for the third argument are defined in pcre.h, and
2097 are as follows:
2101 Return the number of the highest back reference in the pattern. The
2102 fourth argument should point to an int variable. Zero is returned if
2103 there are no back references.
2107 Return the number of capturing subpatterns in the pattern. The fourth
2108 argument should point to an int variable.
2112 Return a pointer to the internal default character tables within PCRE.
2113 The fourth argument should point to an unsigned char * variable. This
2114 information call is provided for internal use by the pcre_study() func-
2115 tion. External callers can cause PCRE to use its internal tables by
2116 passing a NULL table pointer.
2120 Return information about the first data unit of any matched string, for
2121 a non-anchored pattern. (The name of this option refers to the 8-bit
2122 library, where data units are bytes.) The fourth argument should point
2123 to an int variable.
2125 If there is a fixed first value, for example, the letter "c" from a
2126 pattern such as (cat|cow|coyote), its value is returned. In the 8-bit
2127 library, the value is always less than 256; in the 16-bit library the
2128 value can be up to 0xffff.
2130 If there is no fixed first value, and if either
2132 (a) the pattern was compiled with the PCRE_MULTILINE option, and every
2133 branch starts with "^", or
2135 (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
2136 set (if it were set, the pattern would be anchored),
2138 -1 is returned, indicating that the pattern matches only at the start
2139 of a subject string or after any newline within the string. Otherwise
2140 -2 is returned. For anchored patterns, -2 is returned.
2144 If the pattern was studied, and this resulted in the construction of a
2145 256-bit table indicating a fixed set of values for the first data unit
2146 in any matching string, a pointer to the table is returned. Otherwise
2147 NULL is returned. The fourth argument should point to an unsigned char
2148 * variable.
2152 Return 1 if the pattern contains any explicit matches for CR or LF
2153 characters, otherwise 0. The fourth argument should point to an int
2154 variable. An explicit match is either a literal CR or LF character, or
2155 \r or \n.
2159 Return 1 if the (?J) or (?-J) option setting is used in the pattern,
2160 otherwise 0. The fourth argument should point to an int variable. (?J)
2161 and (?-J) set and unset the local PCRE_DUPNAMES option, respectively.
2165 Return 1 if the pattern was studied with one of the JIT options, and
2166 just-in-time compiling was successful. The fourth argument should point
2167 to an int variable. A return value of 0 means that JIT support is not
2168 available in this version of PCRE, or that the pattern was not studied
2169 with a JIT option, or that the JIT compiler could not handle this par-
2170 ticular pattern. See the pcrejit documentation for details of what can
2171 and cannot be handled.
2175 If the pattern was successfully studied with a JIT option, return the
2176 size of the JIT compiled code, otherwise return zero. The fourth argu-
2177 ment should point to a size_t variable.
2181 Return the value of the rightmost literal data unit that must exist in
2182 any matched string, other than at its start, if such a value has been
2183 recorded. The fourth argument should point to an int variable. If there
2184 is no such value, -1 is returned. For anchored patterns, a last literal
2185 value is recorded only if it follows something of variable length. For
2186 example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
2187 /^a\dz\d/ the returned value is -1.
2191 Return the number of characters (NB not bytes) in the longest lookbe-
2192 hind assertion in the pattern. Note that the simple assertions \b and
2193 \B require a one-character lookbehind. This information is useful when
2194 doing multi-segment matching using the partial matching facilities.
2198 If the pattern was studied and a minimum length for matching subject
2199 strings was computed, its value is returned. Otherwise the returned
2200 value is -1. The value is a number of characters, which in UTF-8 mode
2201 may be different from the number of bytes. The fourth argument should
2202 point to an int variable. A non-negative value is a lower bound to the
2203 length of any matching string. There may not be any strings of that
2204 length that do actually match, but every string that does match is at
2205 least that long.
2211 PCRE supports the use of named as well as numbered capturing parenthe-
2212 ses. The names are just an additional way of identifying the parenthe-
2213 ses, which still acquire numbers. Several convenience functions such as
2214 pcre_get_named_substring() are provided for extracting captured sub-
2215 strings by name. It is also possible to extract the data directly, by
2216 first converting the name to a number in order to access the correct
2217 pointers in the output vector (described with pcre_exec() below). To do
2218 the conversion, you need to use the name-to-number map, which is
2219 described by these three values.
2221 The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
2222 gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
2223 of each entry; both of these return an int value. The entry size
2224 depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
2225 a pointer to the first entry of the table. This is a pointer to char in
2226 the 8-bit library, where the first two bytes of each entry are the num-
2227 ber of the capturing parenthesis, most significant byte first. In the
2228 16-bit library, the pointer points to 16-bit data units, the first of
2229 which contains the parenthesis number. The rest of the entry is the
2230 corresponding name, zero terminated.
2232 The names are in alphabetical order. Duplicate names may appear if (?|
2233 is used to create multiple groups with the same number, as described in
2234 the section on duplicate subpattern numbers in the pcrepattern page.
2235 Duplicate names for subpatterns with different numbers are permitted
2236 only if PCRE_DUPNAMES is set. In all cases of duplicate names, they
2237 appear in the table in the order in which they were found in the pat-
2238 tern. In the absence of (?| this is the order of increasing number;
2239 when (?| is used this is not necessarily the case because later subpat-
2240 terns may have lower numbers.
2242 As a simple example of the name/number table, consider the following
2243 pattern after compilation by the 8-bit library (assume PCRE_EXTENDED is
2244 set, so white space - including newlines - is ignored):
2246 (?<date> (?<year>(\d\d)?\d\d) -
2247 (?<month>\d\d) - (?<day>\d\d) )
2249 There are four named subpatterns, so the table has four entries, and
2250 each entry in the table is eight bytes long. The table is as follows,
2251 with non-printing bytes shows in hexadecimal, and undefined bytes shown
2252 as ??:
2254 00 01 d a t e 00 ??
2255 00 05 d a y 00 ?? ??
2256 00 04 m o n t h 00
2257 00 02 y e a r 00 ??
2259 When writing code to extract data from named subpatterns using the
2260 name-to-number map, remember that the length of the entries is likely
2261 to be different for each compiled pattern.
2265 Return 1 if the pattern can be used for partial matching with
2266 pcre_exec(), otherwise 0. The fourth argument should point to an int
2267 variable. From release 8.00, this always returns 1, because the
2268 restrictions that previously applied to partial matching have been
2269 lifted. The pcrepartial documentation gives details of partial match-
2270 ing.
2274 Return a copy of the options with which the pattern was compiled. The
2275 fourth argument should point to an unsigned long int variable. These
2276 option bits are those specified in the call to pcre_compile(), modified
2277 by any top-level option settings at the start of the pattern itself. In
2278 other words, they are the options that will be in force when matching
2279 starts. For example, if the pattern /(?im)abc(?-i)d/ is compiled with
2280 the PCRE_EXTENDED option, the result is PCRE_CASELESS, PCRE_MULTILINE,
2283 A pattern is automatically anchored by PCRE if all of its top-level
2284 alternatives begin with one of the following:
2286 ^ unless PCRE_MULTILINE is set
2287 \A always
2288 \G always
2289 .* if PCRE_DOTALL is set and there are no back
2290 references to the subpattern in which .* appears
2292 For such patterns, the PCRE_ANCHORED bit is set in the options returned
2293 by pcre_fullinfo().
2297 Return the size of the compiled pattern in bytes (for both libraries).
2298 The fourth argument should point to a size_t variable. This value does
2299 not include the size of the pcre structure that is returned by
2300 pcre_compile(). The value that is passed as the argument to pcre_mal-
2301 loc() when pcre_compile() is getting memory in which to place the com-
2302 piled data is the value returned by this option plus the size of the
2303 pcre structure. Studying a compiled pattern, with or without JIT, does
2304 not alter the value returned by this option.
2308 Return the size in bytes of the data block pointed to by the study_data
2309 field in a pcre_extra block. If pcre_extra is NULL, or there is no
2310 study data, zero is returned. The fourth argument should point to a
2311 size_t variable. The study_data field is set by pcre_study() to record
2312 information that will speed up matching (see the section entitled
2313 "Studying a pattern" above). The format of the study_data block is pri-
2314 vate, but its length is made available via this option so that it can
2315 be saved and restored (see the pcreprecompile documentation for
2316 details).
2321 int pcre_refcount(pcre *code, int adjust);
2323 The pcre_refcount() function is used to maintain a reference count in
2324 the data block that contains a compiled pattern. It is provided for the
2325 benefit of applications that operate in an object-oriented manner,
2326 where different parts of the application may be using the same compiled
2327 pattern, but you want to free the block when they are all done.
2329 When a pattern is compiled, the reference count field is initialized to
2330 zero. It is changed only by calling this function, whose action is to
2331 add the adjust value (which may be positive or negative) to it. The
2332 yield of the function is the new value. However, the value of the count
2333 is constrained to lie between 0 and 65535, inclusive. If the new value
2334 is outside these limits, it is forced to the appropriate limit value.
2336 Except when it is zero, the reference count is not correctly preserved
2337 if a pattern is compiled on one host and then transferred to a host
2338 whose byte-order is different. (This seems a highly unlikely scenario.)
2343 int pcre_exec(const pcre *code, const pcre_extra *extra,
2344 const char *subject, int length, int startoffset,
2345 int options, int *ovector, int ovecsize);
2347 The function pcre_exec() is called to match a subject string against a
2348 compiled pattern, which is passed in the code argument. If the pattern
2349 was studied, the result of the study should be passed in the extra
2350 argument. You can call pcre_exec() with the same code and extra argu-
2351 ments as many times as you like, in order to match different subject
2352 strings with the same pattern.
2354 This function is the main matching facility of the library, and it
2355 operates in a Perl-like manner. For specialist use there is also an
2356 alternative matching function, which is described below in the section
2357 about the pcre_dfa_exec() function.
2359 In most applications, the pattern will have been compiled (and option-
2360 ally studied) in the same process that calls pcre_exec(). However, it
2361 is possible to save compiled patterns and study data, and then use them
2362 later in different processes, possibly even on different hosts. For a
2363 discussion about this, see the pcreprecompile documentation.
2365 Here is an example of a simple call to pcre_exec():
2367 int rc;
2368 int ovector[30];
2369 rc = pcre_exec(
2370 re, /* result of pcre_compile() */
2371 NULL, /* we didn't study the pattern */
2372 "some string", /* the subject string */
2373 11, /* the length of the subject string */
2374 0, /* start at offset 0 in the subject */
2375 0, /* default options */
2376 ovector, /* vector of integers for substring information */
2377 30); /* number of elements (NOT size in bytes) */
2379 Extra data for pcre_exec()
2381 If the extra argument is not NULL, it must point to a pcre_extra data
2382 block. The pcre_study() function returns such a block (when it doesn't
2383 return NULL), but you can also create one for yourself, and pass addi-
2384 tional information in it. The pcre_extra block contains the following
2385 fields (not necessarily in this order):
2387 unsigned long int flags;
2388 void *study_data;
2389 void *executable_jit;
2390 unsigned long int match_limit;
2391 unsigned long int match_limit_recursion;
2392 void *callout_data;
2393 const unsigned char *tables;
2394 unsigned char **mark;
2396 In the 16-bit version of this structure, the mark field has type
2397 "PCRE_UCHAR16 **".
2399 The flags field is used to specify which of the other fields are set.
2400 The flag bits are:
2410 Other flag bits should be set to zero. The study_data field and some-
2411 times the executable_jit field are set in the pcre_extra block that is
2412 returned by pcre_study(), together with the appropriate flag bits. You
2413 should not set these yourself, but you may add to the block by setting
2414 other fields and their corresponding flag bits.
2416 The match_limit field provides a means of preventing PCRE from using up
2417 a vast amount of resources when running patterns that are not going to
2418 match, but which have a very large number of possibilities in their
2419 search trees. The classic example is a pattern that uses nested unlim-
2420 ited repeats.
2422 Internally, pcre_exec() uses a function called match(), which it calls
2423 repeatedly (sometimes recursively). The limit set by match_limit is
2424 imposed on the number of times this function is called during a match,
2425 which has the effect of limiting the amount of backtracking that can
2426 take place. For patterns that are not anchored, the count restarts from
2427 zero for each position in the subject string.
2429 When pcre_exec() is called with a pattern that was successfully studied
2430 with a JIT option, the way that the matching is executed is entirely
2431 different. However, there is still the possibility of runaway matching
2432 that goes on for a very long time, and so the match_limit value is also
2433 used in this case (but in a different way) to limit how long the match-
2434 ing can continue.
2436 The default value for the limit can be set when PCRE is built; the
2437 default default is 10 million, which handles all but the most extreme
2438 cases. You can override the default by suppling pcre_exec() with a
2439 pcre_extra block in which match_limit is set, and
2440 PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
2441 exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
2443 The match_limit_recursion field is similar to match_limit, but instead
2444 of limiting the total number of times that match() is called, it limits
2445 the depth of recursion. The recursion depth is a smaller number than
2446 the total number of calls, because not all calls to match() are recur-
2447 sive. This limit is of use only if it is set smaller than match_limit.
2449 Limiting the recursion depth limits the amount of machine stack that
2450 can be used, or, when PCRE has been compiled to use memory on the heap
2451 instead of the stack, the amount of heap memory that can be used. This
2452 limit is not relevant, and is ignored, when matching is done using JIT
2453 compiled code.
2455 The default value for match_limit_recursion can be set when PCRE is
2456 built; the default default is the same value as the default for
2457 match_limit. You can override the default by suppling pcre_exec() with
2458 a pcre_extra block in which match_limit_recursion is set, and
2459 PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
2460 limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
2462 The callout_data field is used in conjunction with the "callout" fea-
2463 ture, and is described in the pcrecallout documentation.
2465 The tables field is used to pass a character tables pointer to
2466 pcre_exec(); this overrides the value that is stored with the compiled
2467 pattern. A non-NULL value is stored with the compiled pattern only if
2468 custom tables were supplied to pcre_compile() via its tableptr argu-
2469 ment. If NULL is passed to pcre_exec() using this mechanism, it forces
2470 PCRE's internal tables to be used. This facility is helpful when re-
2471 using patterns that have been saved after compiling with an external
2472 set of tables, because the external tables might be at a different
2473 address when pcre_exec() is called. See the pcreprecompile documenta-
2474 tion for a discussion of saving compiled patterns for later use.
2476 If PCRE_EXTRA_MARK is set in the flags field, the mark field must be
2477 set to point to a suitable variable. If the pattern contains any back-
2478 tracking control verbs such as (*MARK:NAME), and the execution ends up
2479 with a name to pass back, a pointer to the name string (zero termi-
2480 nated) is placed in the variable pointed to by the mark field. The
2481 names are within the compiled pattern; if you wish to retain such a
2482 name you must copy it before freeing the memory of a compiled pattern.
2483 If there is no name to pass back, the variable pointed to by the mark
2484 field is set to NULL. For details of the backtracking control verbs,
2485 see the section entitled "Backtracking control" in the pcrepattern doc-
2486 umentation.
2488 Option bits for pcre_exec()
2490 The unused bits of the options argument for pcre_exec() must be zero.
2491 The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
2496 If the pattern was successfully studied with one of the just-in-time
2497 (JIT) compile options, the only supported options for JIT execution are
2500 unsupported option is used, JIT execution is disabled and the normal
2501 interpretive code in pcre_exec() is run.
2505 The PCRE_ANCHORED option limits pcre_exec() to matching at the first
2506 matching position. If a pattern was compiled with PCRE_ANCHORED, or
2507 turned out to be anchored by virtue of its contents, it cannot be made
2508 unachored at matching time.
2513 These options (which are mutually exclusive) control what the \R escape
2514 sequence matches. The choice is either to match only CR, LF, or CRLF,
2515 or to match any Unicode newline sequence. These options override the
2516 choice that was made or defaulted when the pattern was compiled.
2524 These options override the newline definition that was chosen or
2525 defaulted when the pattern was compiled. For details, see the descrip-
2526 tion of pcre_compile() above. During matching, the newline choice
2527 affects the behaviour of the dot, circumflex, and dollar metacharac-
2528 ters. It may also alter the way the match position is advanced after a
2529 match failure for an unanchored pattern.
2532 set, and a match attempt for an unanchored pattern fails when the cur-
2533 rent position is at a CRLF sequence, and the pattern contains no
2534 explicit matches for CR or LF characters, the match position is
2535 advanced by two characters instead of one, in other words, to after the
2536 CRLF.
2538 The above rule is a compromise that makes the most common cases work as
2539 expected. For example, if the pattern is .+A (and the PCRE_DOTALL
2540 option is not set), it does not match the string "\r\nA" because, after
2541 failing at the start, it skips both the CR and the LF before retrying.
2542 However, the pattern [\r\n]A does match that string, because it con-
2543 tains an explicit CR or LF reference, and so advances only by one char-
2544 acter after the first failure.
2546 An explicit match for CR of LF is either a literal appearance of one of
2547 those characters, or one of the \r or \n escape sequences. Implicit
2548 matches such as [^X] do not count, nor does \s (which includes CR and
2549 LF in the characters that it matches).
2551 Notwithstanding the above, anomalous effects may still occur when CRLF
2552 is a valid newline sequence and explicit \r or \n escapes appear in the
2553 pattern.
2557 This option specifies that first character of the subject string is not
2558 the beginning of a line, so the circumflex metacharacter should not
2559 match before it. Setting this without PCRE_MULTILINE (at compile time)
2560 causes circumflex never to match. This option affects only the behav-
2561 iour of the circumflex metacharacter. It does not affect \A.
2565 This option specifies that the end of the subject string is not the end
2566 of a line, so the dollar metacharacter should not match it nor (except
2567 in multiline mode) a newline immediately before it. Setting this with-
2568 out PCRE_MULTILINE (at compile time) causes dollar never to match. This
2569 option affects only the behaviour of the dollar metacharacter. It does
2570 not affect \Z or \z.
2574 An empty string is not considered to be a valid match if this option is
2575 set. If there are alternatives in the pattern, they are tried. If all
2576 the alternatives match the empty string, the entire match fails. For
2577 example, if the pattern
2579 a?b?
2581 is applied to a string not beginning with "a" or "b", it matches an
2582 empty string at the start of the subject. With PCRE_NOTEMPTY set, this
2583 match is not valid, so PCRE searches further into the string for occur-
2584 rences of "a" or "b".
2588 This is like PCRE_NOTEMPTY, except that an empty string match that is
2589 not at the start of the subject is permitted. If the pattern is
2590 anchored, such a match can occur only if the pattern contains \K.
2592 Perl has no direct equivalent of PCRE_NOTEMPTY or
2593 PCRE_NOTEMPTY_ATSTART, but it does make a special case of a pattern
2594 match of the empty string within its split() function, and when using
2595 the /g modifier. It is possible to emulate Perl's behaviour after
2596 matching a null string by first trying the match again at the same off-
2597 set with PCRE_NOTEMPTY_ATSTART and PCRE_ANCHORED, and then if that
2598 fails, by advancing the starting offset (see below) and trying an ordi-
2599 nary match again. There is some code that demonstrates how to do this
2600 in the pcredemo sample program. In the most general case, you have to
2601 check to see if the newline convention recognizes CRLF as a newline,
2602 and if so, and the current character is CR followed by LF, advance the
2603 starting offset by two characters instead of one.
2607 There are a number of optimizations that pcre_exec() uses at the start
2608 of a match, in order to speed up the process. For example, if it is
2609 known that an unanchored match must start with a specific character, it
2610 searches the subject for that character, and fails immediately if it
2611 cannot find it, without actually running the main matching function.
2612 This means that a special item such as (*COMMIT) at the start of a pat-
2613 tern is not considered until after a suitable starting point for the
2614 match has been found. When callouts or (*MARK) items are in use, these
2615 "start-up" optimizations can cause them to be skipped if the pattern is
2616 never actually used. The start-up optimizations are in effect a pre-
2617 scan of the subject that takes place before the pattern is run.
2619 The PCRE_NO_START_OPTIMIZE option disables the start-up optimizations,
2620 possibly causing performance to suffer, but ensuring that in cases
2621 where the result is "no match", the callouts do occur, and that items
2622 such as (*COMMIT) and (*MARK) are considered at every possible starting
2623 position in the subject string. If PCRE_NO_START_OPTIMIZE is set at
2624 compile time, it cannot be unset at matching time. The use of
2625 PCRE_NO_START_OPTIMIZE disables JIT execution; when it is set, matching
2626 is always done using interpretively.
2628 Setting PCRE_NO_START_OPTIMIZE can change the outcome of a matching
2629 operation. Consider the pattern
2633 When this is compiled, PCRE records the fact that a match must start
2634 with the character "A". Suppose the subject string is "DEFABC". The
2635 start-up optimization scans along the subject, finds "A" and runs the
2636 first match attempt from there. The (*COMMIT) item means that the pat-
2637 tern must match the current starting position, which in this case, it
2638 does. However, if the same match is run with PCRE_NO_START_OPTIMIZE
2639 set, the initial scan along the subject string does not happen. The
2640 first match attempt is run starting from "D" and when this fails,
2641 (*COMMIT) prevents any further matches being tried, so the overall
2642 result is "no match". If the pattern is studied, more start-up opti-
2643 mizations may be used. For example, a minimum length for the subject
2644 may be recorded. Consider the pattern
2646 (*MARK:A)(X|Y)
2648 The minimum length for a match is one character. If the subject is
2649 "ABC", there will be attempts to match "ABC", "BC", "C", and then
2650 finally an empty string. If the pattern is studied, the final attempt
2651 does not take place, because PCRE knows that the subject is too short,
2652 and so the (*MARK) is never encountered. In this case, studying the
2653 pattern does not affect the overall match result, which is still "no
2654 match", but it does affect the auxiliary information that is returned.
2658 When PCRE_UTF8 is set at compile time, the validity of the subject as a
2659 UTF-8 string is automatically checked when pcre_exec() is subsequently
2660 called. The entire string is checked before any other processing takes
2661 place. The value of startoffset is also checked to ensure that it
2662 points to the start of a UTF-8 character. There is a discussion about
2663 the validity of UTF-8 strings in the pcreunicode page. If an invalid
2664 sequence of bytes is found, pcre_exec() returns the error
2665 PCRE_ERROR_BADUTF8 or, if PCRE_PARTIAL_HARD is set and the problem is a
2666 truncated character at the end of the subject, PCRE_ERROR_SHORTUTF8. In
2667 both cases, information about the precise nature of the error may also
2668 be returned (see the descriptions of these errors in the section enti-
2669 tled Error return values from pcre_exec() below). If startoffset con-
2670 tains a value that does not point to the start of a UTF-8 character (or
2671 to the end of the subject), PCRE_ERROR_BADUTF8_OFFSET is returned.
2673 If you already know that your subject is valid, and you want to skip
2674 these checks for performance reasons, you can set the
2675 PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
2676 do this for the second and subsequent calls to pcre_exec() if you are
2677 making repeated calls to find all the matches in a single subject
2678 string. However, you should be sure that the value of startoffset
2679 points to the start of a character (or the end of the subject). When
2680 PCRE_NO_UTF8_CHECK is set, the effect of passing an invalid string as a
2681 subject or an invalid value of startoffset is undefined. Your program
2682 may crash.
2687 These options turn on the partial matching feature. For backwards com-
2688 patibility, PCRE_PARTIAL is a synonym for PCRE_PARTIAL_SOFT. A partial
2689 match occurs if the end of the subject string is reached successfully,
2690 but there are not enough subject characters to complete the match. If
2691 this happens when PCRE_PARTIAL_SOFT (but not PCRE_PARTIAL_HARD) is set,
2692 matching continues by testing any remaining alternatives. Only if no
2693 complete match can be found is PCRE_ERROR_PARTIAL returned instead of
2694 PCRE_ERROR_NOMATCH. In other words, PCRE_PARTIAL_SOFT says that the
2695 caller is prepared to handle a partial match, but only if no complete
2696 match can be found.
2698 If PCRE_PARTIAL_HARD is set, it overrides PCRE_PARTIAL_SOFT. In this
2699 case, if a partial match is found, pcre_exec() immediately returns
2700 PCRE_ERROR_PARTIAL, without considering any other alternatives. In
2701 other words, when PCRE_PARTIAL_HARD is set, a partial match is consid-
2702 ered to be more important that an alternative complete match.
2704 In both cases, the portion of the string that was inspected when the
2705 partial match was found is set as the first matching string. There is a
2706 more detailed discussion of partial and multi-segment matching, with
2707 examples, in the pcrepartial documentation.
2709 The string to be matched by pcre_exec()
2711 The subject string is passed to pcre_exec() as a pointer in subject, a
2712 length in bytes in length, and a starting byte offset in startoffset.
2713 If this is negative or greater than the length of the subject,
2714 pcre_exec() returns PCRE_ERROR_BADOFFSET. When the starting offset is
2715 zero, the search for a match starts at the beginning of the subject,
2716 and this is by far the most common case. In UTF-8 mode, the byte offset
2717 must point to the start of a UTF-8 character (or the end of the sub-
2718 ject). Unlike the pattern string, the subject may contain binary zero
2719 bytes.
2721 A non-zero starting offset is useful when searching for another match
2722 in the same subject by calling pcre_exec() again after a previous suc-
2723 cess. Setting startoffset differs from just passing over a shortened
2724 string and setting PCRE_NOTBOL in the case of a pattern that begins
2725 with any kind of lookbehind. For example, consider the pattern
2727 \Biss\B
2729 which finds occurrences of "iss" in the middle of words. (\B matches
2730 only if the current position in the subject is not a word boundary.)
2731 When applied to the string "Mississipi" the first call to pcre_exec()
2732 finds the first occurrence. If pcre_exec() is called again with just
2733 the remainder of the subject, namely "issipi", it does not match,
2734 because \B is always false at the start of the subject, which is deemed
2735 to be a word boundary. However, if pcre_exec() is passed the entire
2736 string again, but with startoffset set to 4, it finds the second occur-
2737 rence of "iss" because it is able to look behind the starting point to
2738 discover that it is preceded by a letter.
2740 Finding all the matches in a subject is tricky when the pattern can
2741 match an empty string. It is possible to emulate Perl's /g behaviour by
2742 first trying the match again at the same offset, with the
2743 PCRE_NOTEMPTY_ATSTART and PCRE_ANCHORED options, and then if that
2744 fails, advancing the starting offset and trying an ordinary match
2745 again. There is some code that demonstrates how to do this in the pcre-
2746 demo sample program. In the most general case, you have to check to see
2747 if the newline convention recognizes CRLF as a newline, and if so, and
2748 the current character is CR followed by LF, advance the starting offset
2749 by two characters instead of one.
2751 If a non-zero starting offset is passed when the pattern is anchored,
2752 one attempt to match at the given offset is made. This can only succeed
2753 if the pattern does not require the match to be at the start of the
2754 subject.
2756 How pcre_exec() returns captured substrings
2758 In general, a pattern matches a certain portion of the subject, and in
2759 addition, further substrings from the subject may be picked out by
2760 parts of the pattern. Following the usage in Jeffrey Friedl's book,
2761 this is called "capturing" in what follows, and the phrase "capturing
2762 subpattern" is used for a fragment of a pattern that picks out a sub-
2763 string. PCRE supports several other kinds of parenthesized subpattern
2764 that do not cause substrings to be captured.
2766 Captured substrings are returned to the caller via a vector of integers
2767 whose address is passed in ovector. The number of elements in the vec-
2768 tor is passed in ovecsize, which must be a non-negative number. Note:
2769 this argument is NOT the size of ovector in bytes.
2771 The first two-thirds of the vector is used to pass back captured sub-
2772 strings, each substring using a pair of integers. The remaining third
2773 of the vector is used as workspace by pcre_exec() while matching cap-
2774 turing subpatterns, and is not available for passing back information.
2775 The number passed in ovecsize should always be a multiple of three. If
2776 it is not, it is rounded down.
2778 When a match is successful, information about captured substrings is
2779 returned in pairs of integers, starting at the beginning of ovector,
2780 and continuing up to two-thirds of its length at the most. The first
2781 element of each pair is set to the byte offset of the first character
2782 in a substring, and the second is set to the byte offset of the first
2783 character after the end of a substring. Note: these values are always
2784 byte offsets, even in UTF-8 mode. They are not character counts.
2786 The first pair of integers, ovector[0] and ovector[1], identify the
2787 portion of the subject string matched by the entire pattern. The next
2788 pair is used for the first capturing subpattern, and so on. The value
2789 returned by pcre_exec() is one more than the highest numbered pair that
2790 has been set. For example, if two substrings have been captured, the
2791 returned value is 3. If there are no capturing subpatterns, the return
2792 value from a successful match is 1, indicating that just the first pair
2793 of offsets has been set.
2795 If a capturing subpattern is matched repeatedly, it is the last portion
2796 of the string that it matched that is returned.
2798 If the vector is too small to hold all the captured substring offsets,
2799 it is used as far as possible (up to two-thirds of its length), and the
2800 function returns a value of zero. If neither the actual string matched
2801 nor any captured substrings are of interest, pcre_exec() may be called
2802 with ovector passed as NULL and ovecsize as zero. However, if the pat-
2803 tern contains back references and the ovector is not big enough to
2804 remember the related substrings, PCRE has to get additional memory for
2805 use during matching. Thus it is usually advisable to supply an ovector
2806 of reasonable size.
2808 There are some cases where zero is returned (indicating vector over-
2809 flow) when in fact the vector is exactly the right size for the final
2810 match. For example, consider the pattern
2812 (a)(?:(b)c|bd)
2814 If a vector of 6 elements (allowing for only 1 captured substring) is
2815 given with subject string "abd", pcre_exec() will try to set the second
2816 captured string, thereby recording a vector overflow, before failing to
2817 match "c" and backing up to try the second alternative. The zero
2818 return, however, does correctly indicate that the maximum number of
2819 slots (namely 2) have been filled. In similar cases where there is tem-
2820 porary overflow, but the final number of used slots is actually less
2821 than the maximum, a non-zero value is returned.
2823 The pcre_fullinfo() function can be used to find out how many capturing
2824 subpatterns there are in a compiled pattern. The smallest size for
2825 ovector that will allow for n captured substrings, in addition to the
2826 offsets of the substring matched by the whole pattern, is (n+1)*3.
2828 It is possible for capturing subpattern number n+1 to match some part
2829 of the subject when subpattern n has not been used at all. For example,
2830 if the string "abc" is matched against the pattern (a|(z))(bc) the
2831 return from the function is 4, and subpatterns 1 and 3 are matched, but
2832 2 is not. When this happens, both values in the offset pairs corre-
2833 sponding to unused subpatterns are set to -1.
2835 Offset values that correspond to unused subpatterns at the end of the
2836 expression are also set to -1. For example, if the string "abc" is
2837 matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
2838 matched. The return from the function is 2, because the highest used
2839 capturing subpattern number is 1, and the offsets for for the second
2840 and third capturing subpatterns (assuming the vector is large enough,
2841 of course) are set to -1.
2843 Note: Elements in the first two-thirds of ovector that do not corre-
2844 spond to capturing parentheses in the pattern are never changed. That
2845 is, if a pattern contains n capturing parentheses, no more than ovec-
2846 tor[0] to ovector[2n+1] are set by pcre_exec(). The other elements (in
2847 the first two-thirds) retain whatever values they previously had.
2849 Some convenience functions are provided for extracting the captured
2850 substrings as separate strings. These are described below.
2852 Error return values from pcre_exec()
2854 If pcre_exec() fails, it returns a negative number. The following are
2855 defined in the header file:
2859 The subject string did not match the pattern.
2863 Either code or subject was passed as NULL, or ovector was NULL and
2864 ovecsize was not zero.
2868 An unrecognized bit was set in the options argument.
2872 PCRE stores a 4-byte "magic number" at the start of the compiled code,
2873 to catch the case when it is passed a junk pointer and to detect when a
2874 pattern that was compiled in an environment of one endianness is run in
2875 an environment with the other endianness. This is the error that PCRE
2876 gives when the magic number is not present.
2880 While running the pattern match, an unknown item was encountered in the
2881 compiled pattern. This error could be caused by a bug in PCRE or by
2882 overwriting of the compiled pattern.
2886 If a pattern contains back references, but the ovector that is passed
2887 to pcre_exec() is not big enough to remember the referenced substrings,
2888 PCRE gets a block of memory at the start of matching to use for this
2889 purpose. If the call via pcre_malloc() fails, this error is given. The
2890 memory is automatically freed at the end of matching.
2892 This error is also given if pcre_stack_malloc() fails in pcre_exec().
2893 This can happen only when PCRE has been compiled with --disable-stack-
2894 for-recursion.
2898 This error is used by the pcre_copy_substring(), pcre_get_substring(),
2899 and pcre_get_substring_list() functions (see below). It is never
2900 returned by pcre_exec().
2904 The backtracking limit, as specified by the match_limit field in a
2905 pcre_extra structure (or defaulted) was reached. See the description
2906 above.
2910 This error is never generated by pcre_exec() itself. It is provided for
2911 use by callout functions that want to yield a distinctive error code.
2912 See the pcrecallout documentation for details.
2916 A string that contains an invalid UTF-8 byte sequence was passed as a
2917 subject, and the PCRE_NO_UTF8_CHECK option was not set. If the size of
2918 the output vector (ovecsize) is at least 2, the byte offset to the
2919 start of the the invalid UTF-8 character is placed in the first ele-
2920 ment, and a reason code is placed in the second element. The reason
2921 codes are listed in the following section. For backward compatibility,
2922 if PCRE_PARTIAL_HARD is set and the problem is a truncated UTF-8 char-
2923 acter at the end of the subject (reason codes 1 to 5),
2924 PCRE_ERROR_SHORTUTF8 is returned instead of PCRE_ERROR_BADUTF8.
2928 The UTF-8 byte sequence that was passed as a subject was checked and
2929 found to be valid (the PCRE_NO_UTF8_CHECK option was not set), but the
2930 value of startoffset did not point to the beginning of a UTF-8 charac-
2931 ter or the end of the subject.
2935 The subject string did not match, but it did match partially. See the
2936 pcrepartial documentation for details of partial matching.
2940 This code is no longer in use. It was formerly returned when the
2941 PCRE_PARTIAL option was used with a compiled pattern containing items
2942 that were not supported for partial matching. From release 8.00
2943 onwards, there are no restrictions on partial matching.
2947 An unexpected internal error has occurred. This error could be caused
2948 by a bug in PCRE or by overwriting of the compiled pattern.
2952 This error is given if the value of the ovecsize argument is negative.
2956 The internal recursion limit, as specified by the match_limit_recursion
2957 field in a pcre_extra structure (or defaulted) was reached. See the
2958 description above.
2962 An invalid combination of PCRE_NEWLINE_xxx options was given.
2966 The value of startoffset was negative or greater than the length of the
2967 subject, that is, the value in length.
2971 This error is returned instead of PCRE_ERROR_BADUTF8 when the subject
2972 string ends with a truncated UTF-8 character and the PCRE_PARTIAL_HARD
2973 option is set. Information about the failure is returned as for
2974 PCRE_ERROR_BADUTF8. It is in fact sufficient to detect this case, but
2975 this special error code for PCRE_PARTIAL_HARD precedes the implementa-
2976 tion of returned information; it is retained for backwards compatibil-
2977 ity.
2981 This error is returned when pcre_exec() detects a recursion loop within
2982 the pattern. Specifically, it means that either the whole pattern or a
2983 subpattern has been called recursively for the second time at the same
2984 position in the subject string. Some simple patterns that might do this
2985 are detected and faulted at compile time, but more complicated cases,
2986 in particular mutual recursions between two different subpatterns, can-
2987 not be detected until run time.
2991 This error is returned when a pattern that was successfully studied
2992 using a JIT compile option is being matched, but the memory available
2993 for the just-in-time processing stack is not large enough. See the
2994 pcrejit documentation for more details.
2998 This error is given if a pattern that was compiled by the 8-bit library
2999 is passed to a 16-bit library function, or vice versa.
3003 This error is given if a pattern that was compiled and saved is
3004 reloaded on a host with different endianness. The utility function
3005 pcre_pattern_to_host_byte_order() can be used to convert such a pattern
3006 so that it runs on the new host.
3008 Error numbers -16 to -20 and -22 are not used by pcre_exec().
3010 Reason codes for invalid UTF-8 strings
3012 This section applies only to the 8-bit library. The corresponding
3013 information for the 16-bit library is given in the pcre16 page.
3015 When pcre_exec() returns either PCRE_ERROR_BADUTF8 or PCRE_ERROR_SHORT-
3016 UTF8, and the size of the output vector (ovecsize) is at least 2, the
3017 offset of the start of the invalid UTF-8 character is placed in the
3018 first output vector element (ovector[0]) and a reason code is placed in
3019 the second element (ovector[1]). The reason codes are given names in
3020 the pcre.h header file:
3028 The string ends with a truncated UTF-8 character; the code specifies
3029 how many bytes are missing (1 to 5). Although RFC 3629 restricts UTF-8
3030 characters to be no longer than 4 bytes, the encoding scheme (origi-
3031 nally defined by RFC 2279) allows for up to 6 bytes, and this is
3032 checked first; hence the possibility of 4 or 5 missing bytes.
3038 PCRE_UTF8_ERR10
3040 The two most significant bits of the 2nd, 3rd, 4th, 5th, or 6th byte of
3041 the character do not have the binary value 0b10 (that is, either the
3042 most significant bit is 0, or the next bit is 1).
3044 PCRE_UTF8_ERR11
3045 PCRE_UTF8_ERR12
3047 A character that is valid by the RFC 2279 rules is either 5 or 6 bytes
3048 long; these code points are excluded by RFC 3629.
3050 PCRE_UTF8_ERR13
3052 A 4-byte character has a value greater than 0x10fff; these code points
3053 are excluded by RFC 3629.
3055 PCRE_UTF8_ERR14
3057 A 3-byte character has a value in the range 0xd800 to 0xdfff; this
3058 range of code points are reserved by RFC 3629 for use with UTF-16, and
3059 so are excluded from UTF-8.
3061 PCRE_UTF8_ERR15
3062 PCRE_UTF8_ERR16
3063 PCRE_UTF8_ERR17
3064 PCRE_UTF8_ERR18
3065 PCRE_UTF8_ERR19
3067 A 2-, 3-, 4-, 5-, or 6-byte character is "overlong", that is, it codes
3068 for a value that can be represented by fewer bytes, which is invalid.
3069 For example, the two bytes 0xc0, 0xae give the value 0x2e, whose cor-
3070 rect coding uses just one byte.
3072 PCRE_UTF8_ERR20
3074 The two most significant bits of the first byte of a character have the
3075 binary value 0b10 (that is, the most significant bit is 1 and the sec-
3076 ond is 0). Such a byte can only validly occur as the second or subse-
3077 quent byte of a multi-byte character.
3079 PCRE_UTF8_ERR21
3081 The first byte of a character has the value 0xfe or 0xff. These values
3082 can never occur in a valid UTF-8 string.
3087 int pcre_copy_substring(const char *subject, int *ovector,
3088 int stringcount, int stringnumber, char *buffer,
3089 int buffersize);
3091 int pcre_get_substring(const char *subject, int *ovector,
3092 int stringcount, int stringnumber,
3093 const char **stringptr);
3095 int pcre_get_substring_list(const char *subject,
3096 int *ovector, int stringcount, const char ***listptr);
3098 Captured substrings can be accessed directly by using the offsets
3099 returned by pcre_exec() in ovector. For convenience, the functions
3100 pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
3101 string_list() are provided for extracting captured substrings as new,
3102 separate, zero-terminated strings. These functions identify substrings
3103 by number. The next section describes functions for extracting named
3104 substrings.
3106 A substring that contains a binary zero is correctly extracted and has
3107 a further zero added on the end, but the result is not, of course, a C
3108 string. However, you can process such a string by referring to the
3109 length that is returned by pcre_copy_substring() and pcre_get_sub-
3110 string(). Unfortunately, the interface to pcre_get_substring_list() is
3111 not adequate for handling strings containing binary zeros, because the
3112 end of the final string is not independently indicated.
3114 The first three arguments are the same for all three of these func-
3115 tions: subject is the subject string that has just been successfully
3116 matched, ovector is a pointer to the vector of integer offsets that was
3117 passed to pcre_exec(), and stringcount is the number of substrings that
3118 were captured by the match, including the substring that matched the
3119 entire regular expression. This is the value returned by pcre_exec() if
3120 it is greater than zero. If pcre_exec() returned zero, indicating that
3121 it ran out of space in ovector, the value passed as stringcount should
3122 be the number of elements in the vector divided by three.
3124 The functions pcre_copy_substring() and pcre_get_substring() extract a
3125 single substring, whose number is given as stringnumber. A value of
3126 zero extracts the substring that matched the entire pattern, whereas
3127 higher values extract the captured substrings. For pcre_copy_sub-
3128 string(), the string is placed in buffer, whose length is given by
3129 buffersize, while for pcre_get_substring() a new block of memory is
3130 obtained via pcre_malloc, and its address is returned via stringptr.
3131 The yield of the function is the length of the string, not including
3132 the terminating zero, or one of these error codes:
3136 The buffer was too small for pcre_copy_substring(), or the attempt to
3137 get memory failed for pcre_get_substring().
3141 There is no substring whose number is stringnumber.
3143 The pcre_get_substring_list() function extracts all available sub-
3144 strings and builds a list of pointers to them. All this is done in a
3145 single block of memory that is obtained via pcre_malloc. The address of
3146 the memory block is returned via listptr, which is also the start of
3147 the list of string pointers. The end of the list is marked by a NULL
3148 pointer. The yield of the function is zero if all went well, or the
3149 error code
3153 if the attempt to get the memory block failed.
3155 When any of these functions encounter a substring that is unset, which
3156 can happen when capturing subpattern number n+1 matches some part of
3157 the subject, but subpattern n has not been used at all, they return an
3158 empty string. This can be distinguished from a genuine zero-length sub-
3159 string by inspecting the appropriate offset in ovector, which is nega-
3160 tive for unset substrings.
3162 The two convenience functions pcre_free_substring() and pcre_free_sub-
3163 string_list() can be used to free the memory returned by a previous
3164 call of pcre_get_substring() or pcre_get_substring_list(), respec-
3165 tively. They do nothing more than call the function pointed to by
3166 pcre_free, which of course could be called directly from a C program.
3167 However, PCRE is used in some situations where it is linked via a spe-
3168 cial interface to another programming language that cannot use
3169 pcre_free directly; it is for these cases that the functions are pro-
3170 vided.
3175 int pcre_get_stringnumber(const pcre *code,
3176 const char *name);
3178 int pcre_copy_named_substring(const pcre *code,
3179 const char *subject, int *ovector,
3180 int stringcount, const char *stringname,
3181 char *buffer, int buffersize);
3183 int pcre_get_named_substring(const pcre *code,
3184 const char *subject, int *ovector,
3185 int stringcount, const char *stringname,
3186 const char **stringptr);
3188 To extract a substring by name, you first have to find associated num-
3189 ber. For example, for this pattern
3191 (a+)b(?<xxx>\d+)...
3193 the number of the subpattern called "xxx" is 2. If the name is known to
3194 be unique (PCRE_DUPNAMES was not set), you can find the number from the
3195 name by calling pcre_get_stringnumber(). The first argument is the com-
3196 piled pattern, and the second is the name. The yield of the function is
3197 the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
3198 subpattern of that name.
3200 Given the number, you can extract the substring directly, or use one of
3201 the functions described in the previous section. For convenience, there
3202 are also two functions that do the whole job.
3204 Most of the arguments of pcre_copy_named_substring() and
3205 pcre_get_named_substring() are the same as those for the similarly
3206 named functions that extract by number. As these are described in the
3207 previous section, they are not re-described here. There are just two
3208 differences:
3210 First, instead of a substring number, a substring name is given. Sec-
3211 ond, there is an extra argument, given at the start, which is a pointer
3212 to the compiled pattern. This is needed in order to gain access to the
3213 name-to-number translation table.
3215 These functions call pcre_get_stringnumber(), and if it succeeds, they
3216 then call pcre_copy_substring() or pcre_get_substring(), as appropri-
3217 ate. NOTE: If PCRE_DUPNAMES is set and there are duplicate names, the
3218 behaviour may not be what you want (see the next section).
3220 Warning: If the pattern uses the (?| feature to set up multiple subpat-
3221 terns with the same number, as described in the section on duplicate
3222 subpattern numbers in the pcrepattern page, you cannot use names to
3223 distinguish the different subpatterns, because names are not included
3224 in the compiled code. The matching process uses only numbers. For this
3225 reason, the use of different names for subpatterns of the same number
3226 causes an error at compile time.
3231 int pcre_get_stringtable_entries(const pcre *code,
3232 const char *name, char **first, char **last);
3234 When a pattern is compiled with the PCRE_DUPNAMES option, names for
3235 subpatterns are not required to be unique. (Duplicate names are always
3236 allowed for subpatterns with the same number, created by using the (?|
3237 feature. Indeed, if such subpatterns are named, they are required to
3238 use the same names.)
3240 Normally, patterns with duplicate names are such that in any one match,
3241 only one of the named subpatterns participates. An example is shown in
3242 the pcrepattern documentation.
3244 When duplicates are present, pcre_copy_named_substring() and
3245 pcre_get_named_substring() return the first substring corresponding to
3246 the given name that is set. If none are set, PCRE_ERROR_NOSUBSTRING
3247 (-7) is returned; no data is returned. The pcre_get_stringnumber()
3248 function returns one of the numbers that are associated with the name,
3249 but it is not defined which it is.
3251 If you want to get full details of all captured substrings for a given
3252 name, you must use the pcre_get_stringtable_entries() function. The
3253 first argument is the compiled pattern, and the second is the name. The
3254 third and fourth are pointers to variables which are updated by the
3255 function. After it has run, they point to the first and last entries in
3256 the name-to-number table for the given name. The function itself
3257 returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
3258 there are none. The format of the table is described above in the sec-
3259 tion entitled Information about a pattern above. Given all the rele-
3260 vant entries for the name, you can extract each of their numbers, and
3261 hence the captured data, if any.
3266 The traditional matching function uses a similar algorithm to Perl,
3267 which stops when it finds the first match, starting at a given point in
3268 the subject. If you want to find all possible matches, or the longest
3269 possible match, consider using the alternative matching function (see
3270 below) instead. If you cannot use the alternative function, but still
3271 need to find all possible matches, you can kludge it up by making use
3272 of the callout facility, which is described in the pcrecallout documen-
3273 tation.
3275 What you have to do is to insert a callout right at the end of the pat-
3276 tern. When your callout function is called, extract and save the cur-
3277 rent matched substring. Then return 1, which forces pcre_exec() to
3278 backtrack and try other alternatives. Ultimately, when it runs out of
3279 matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
3284 Matching certain patterns using pcre_exec() can use a lot of process
3285 stack, which in certain environments can be rather limited in size.
3286 Some users find it helpful to have an estimate of the amount of stack
3287 that is used by pcre_exec(), to help them set recursion limits, as
3288 described in the pcrestack documentation. The estimate that is output
3289 by pcretest when called with the -m and -C options is obtained by call-
3290 ing pcre_exec with the values NULL, NULL, NULL, -999, and -999 for its
3291 first five arguments.
3293 Normally, if its first argument is NULL, pcre_exec() immediately
3294 returns the negative error code PCRE_ERROR_NULL, but with this special
3295 combination of arguments, it returns instead a negative number whose
3296 absolute value is the approximate stack frame size in bytes. (A nega-
3297 tive number is used so that it is clear that no match has happened.)
3298 The value is approximate because in some cases, recursive calls to
3299 pcre_exec() occur when there are one or two additional variables on the
3300 stack.
3302 If PCRE has been compiled to use the heap instead of the stack for
3303 recursion, the value returned is the size of each block that is
3304 obtained from the heap.
3309 int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
3310 const char *subject, int length, int startoffset,
3311 int options, int *ovector, int ovecsize,
3312 int *workspace, int wscount);
3314 The function pcre_dfa_exec() is called to match a subject string
3315 against a compiled pattern, using a matching algorithm that scans the
3316 subject string just once, and does not backtrack. This has different
3317 characteristics to the normal algorithm, and is not compatible with
3318 Perl. Some of the features of PCRE patterns are not supported. Never-
3319 theless, there are times when this kind of matching can be useful. For
3320 a discussion of the two matching algorithms, and a list of features
3321 that pcre_dfa_exec() does not support, see the pcrematching documenta-
3322 tion.
3324 The arguments for the pcre_dfa_exec() function are the same as for
3325 pcre_exec(), plus two extras. The ovector argument is used in a differ-
3326 ent way, and this is described below. The other common arguments are
3327 used in the same way as for pcre_exec(), so their description is not
3328 repeated here.
3330 The two additional arguments provide workspace for the function. The
3331 workspace vector should contain at least 20 elements. It is used for
3332 keeping track of multiple paths through the pattern tree. More
3333 workspace will be needed for patterns and subjects where there are a
3334 lot of potential matches.
3336 Here is an example of a simple call to pcre_dfa_exec():
3338 int rc;
3339 int ovector[10];
3340 int wspace[20];
3341 rc = pcre_dfa_exec(
3342 re, /* result of pcre_compile() */
3343 NULL, /* we didn't study the pattern */
3344 "some string", /* the subject string */
3345 11, /* the length of the subject string */
3346 0, /* start at offset 0 in the subject */
3347 0, /* default options */
3348 ovector, /* vector of integers for substring information */
3349 10, /* number of elements (NOT size in bytes) */
3350 wspace, /* working space vector */
3351 20); /* number of elements (NOT size in bytes) */
3353 Option bits for pcre_dfa_exec()
3355 The unused bits of the options argument for pcre_dfa_exec() must be
3356 zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
3361 four of these are exactly the same as for pcre_exec(), so their
3362 description is not repeated here.
3367 These have the same general effect as they do for pcre_exec(), but the
3368 details are slightly different. When PCRE_PARTIAL_HARD is set for
3369 pcre_dfa_exec(), it returns PCRE_ERROR_PARTIAL if the end of the sub-
3370 ject is reached and there is still at least one matching possibility
3371 that requires additional characters. This happens even if some complete
3372 matches have also been found. When PCRE_PARTIAL_SOFT is set, the return
3373 code PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end
3374 of the subject is reached, there have been no complete matches, but
3375 there is still at least one matching possibility. The portion of the
3376 string that was inspected when the longest partial match was found is
3377 set as the first matching string in both cases. There is a more
3378 detailed discussion of partial and multi-segment matching, with exam-
3379 ples, in the pcrepartial documentation.
3383 Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
3384 stop as soon as it has found one match. Because of the way the alterna-
3385 tive algorithm works, this is necessarily the shortest possible match
3386 at the first possible matching point in the subject string.
3390 When pcre_dfa_exec() returns a partial match, it is possible to call it
3391 again, with additional subject characters, and have it continue with
3392 the same match. The PCRE_DFA_RESTART option requests this action; when
3393 it is set, the workspace and wscount options must reference the same
3394 vector as before because data about the match so far is left in them
3395 after a partial match. There is more discussion of this facility in the
3396 pcrepartial documentation.
3398 Successful returns from pcre_dfa_exec()
3400 When pcre_dfa_exec() succeeds, it may have matched more than one sub-
3401 string in the subject. Note, however, that all the matches from one run
3402 of the function start at the same point in the subject. The shorter
3403 matches are all initial substrings of the longer matches. For example,
3404 if the pattern
3406 <.*>
3408 is matched against the string
3410 This is <something> <something else> <something further> no more
3412 the three matched strings are
3414 <something>
3415 <something> <something else>
3416 <something> <something else> <something further>
3418 On success, the yield of the function is a number greater than zero,
3419 which is the number of matched substrings. The substrings themselves
3420 are returned in ovector. Each string uses two elements; the first is
3421 the offset to the start, and the second is the offset to the end. In
3422 fact, all the strings have the same start offset. (Space could have
3423 been saved by giving this only once, but it was decided to retain some
3424 compatibility with the way pcre_exec() returns data, even though the
3425 meaning of the strings is different.)
3427 The strings are returned in reverse order of length; that is, the long-
3428 est matching string is given first. If there were too many matches to
3429 fit into ovector, the yield of the function is zero, and the vector is
3430 filled with the longest matches. Unlike pcre_exec(), pcre_dfa_exec()
3431 can use the entire ovector for returning matched strings.
3433 Error returns from pcre_dfa_exec()
3435 The pcre_dfa_exec() function returns a negative number when it fails.
3436 Many of the errors are the same as for pcre_exec(), and these are
3437 described above. There are in addition the following errors that are
3438 specific to pcre_dfa_exec():
3442 This return is given if pcre_dfa_exec() encounters an item in the pat-
3443 tern that it does not support, for instance, the use of \C or a back
3444 reference.
3448 This return is given if pcre_dfa_exec() encounters a condition item
3449 that uses a back reference for the condition, or a test for recursion
3450 in a specific group. These are not supported.
3454 This return is given if pcre_dfa_exec() is called with an extra block
3455 that contains a setting of the match_limit or match_limit_recursion
3456 fields. This is not supported (these fields are meaningless for DFA
3457 matching).
3461 This return is given if pcre_dfa_exec() runs out of space in the
3462 workspace vector.
3466 When a recursive subpattern is processed, the matching function calls
3467 itself recursively, using private vectors for ovector and workspace.
3468 This error is given if the output vector is not large enough. This
3469 should be extremely rare, as a vector of size 1000 is used.
3474 pcre16(3), pcrebuild(3), pcrecallout(3), pcrecpp(3)(3), pcrematch-
3475 ing(3), pcrepartial(3), pcreposix(3), pcreprecompile(3), pcresample(3),
3476 pcrestack(3).
3481 Philip Hazel
3482 University Computing Service
3483 Cambridge CB2 3QH, England.
3488 Last updated: 14 April 2012
3489 Copyright (c) 1997-2012 University of Cambridge.
3490 ------------------------------------------------------------------------------
3496 NAME
3497 PCRE - Perl-compatible regular expressions
3502 int (*pcre_callout)(pcre_callout_block *);
3504 int (*pcre16_callout)(pcre16_callout_block *);
3506 PCRE provides a feature called "callout", which is a means of temporar-
3507 ily passing control to the caller of PCRE in the middle of pattern
3508 matching. The caller of PCRE provides an external function by putting
3509 its entry point in the global variable pcre_callout (pcre16_callout for
3510 the 16-bit library). By default, this variable contains NULL, which
3511 disables all calling out.
3513 Within a regular expression, (?C) indicates the points at which the
3514 external function is to be called. Different callout points can be
3515 identified by putting a number less than 256 after the letter C. The
3516 default value is zero. For example, this pattern has two callout
3517 points:
3519 (?C1)abc(?C2)def
3521 If the PCRE_AUTO_CALLOUT option bit is set when a pattern is compiled,
3522 PCRE automatically inserts callouts, all with number 255, before each
3523 item in the pattern. For example, if PCRE_AUTO_CALLOUT is used with the
3524 pattern
3526 A(\d{2}|--)
3528 it is processed as if it were
3530 (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
3532 Notice that there is a callout before and after each parenthesis and
3533 alternation bar. Automatic callouts can be used for tracking the
3534 progress of pattern matching. The pcretest command has an option that
3535 sets automatic callouts; when it is used, the output indicates how the
3536 pattern is matched. This is useful information when you are trying to
3537 optimize the performance of a particular pattern.
3539 The use of callouts in a pattern makes it ineligible for optimization
3540 by the just-in-time compiler. Studying such a pattern with the
3541 PCRE_STUDY_JIT_COMPILE option always fails.
3546 You should be aware that, because of optimizations in the way PCRE
3547 matches patterns by default, callouts sometimes do not happen. For
3548 example, if the pattern is
3550 ab(?C4)cd
3552 PCRE knows that any matching string must contain the letter "d". If the
3553 subject string is "abyz", the lack of "d" means that matching doesn't
3554 ever start, and the callout is never reached. However, with "abyd",
3555 though the result is still no match, the callout is obeyed.
3557 If the pattern is studied, PCRE knows the minimum length of a matching
3558 string, and will immediately give a "no match" return without actually
3559 running a match if the subject is not long enough, or, for unanchored
3560 patterns, if it has been scanned far enough.
3562 You can disable these optimizations by passing the PCRE_NO_START_OPTI-
3563 MIZE option to the matching function, or by starting the pattern with
3564 (*NO_START_OPT). This slows down the matching process, but does ensure
3565 that callouts such as the example above are obeyed.
3570 During matching, when PCRE reaches a callout point, the external func-
3571 tion defined by pcre_callout or pcre16_callout is called (if it is
3572 set). This applies to both normal and DFA matching. The only argument
3573 to the callout function is a pointer to a pcre_callout or pcre16_call-
3574 out block. These structures contains the following fields:
3576 int version;
3577 int callout_number;
3578 int *offset_vector;
3579 const char *subject; (8-bit version)
3580 PCRE_SPTR16 subject; (16-bit version)
3581 int subject_length;
3582 int start_match;
3583 int current_position;
3584 int capture_top;
3585 int capture_last;
3586 void *callout_data;
3587 int pattern_position;
3588 int next_item_length;
3589 const unsigned char *mark; (8-bit version)
3590 const PCRE_UCHAR16 *mark; (16-bit version)
3592 The version field is an integer containing the version number of the
3593 block format. The initial version was 0; the current version is 2. The
3594 version number will change again in future if additional fields are
3595 added, but the intention is never to remove any of the existing fields.
3597 The callout_number field contains the number of the callout, as com-
3598 piled into the pattern (that is, the number after ?C for manual call-
3599 outs, and 255 for automatically generated callouts).
3601 The offset_vector field is a pointer to the vector of offsets that was
3602 passed by the caller to the matching function. When pcre_exec() or
3603 pcre16_exec() is used, the contents can be inspected, in order to
3604 extract substrings that have been matched so far, in the same way as
3605 for extracting substrings after a match has completed. For the DFA
3606 matching functions, this field is not useful.
3608 The subject and subject_length fields contain copies of the values that
3609 were passed to the matching function.
3611 The start_match field normally contains the offset within the subject
3612 at which the current match attempt started. However, if the escape
3613 sequence \K has been encountered, this value is changed to reflect the
3614 modified starting point. If the pattern is not anchored, the callout
3615 function may be called several times from the same point in the pattern
3616 for different starting points in the subject.
3618 The current_position field contains the offset within the subject of
3619 the current match pointer.
3621 When the pcre_exec() or pcre16_exec() is used, the capture_top field
3622 contains one more than the number of the highest numbered captured sub-
3623 string so far. If no substrings have been captured, the value of cap-
3624 ture_top is one. This is always the case when the DFA functions are
3625 used, because they do not support captured substrings.
3627 The capture_last field contains the number of the most recently cap-
3628 tured substring. If no substrings have been captured, its value is -1.
3629 This is always the case for the DFA matching functions.
3631 The callout_data field contains a value that is passed to a matching
3632 function specifically so that it can be passed back in callouts. It is
3633 passed in the callout_data field of a pcre_extra or pcre16_extra data
3634 structure. If no such data was passed, the value of callout_data in a
3635 callout block is NULL. There is a description of the pcre_extra struc-
3636 ture in the pcreapi documentation.
3638 The pattern_position field is present from version 1 of the callout
3639 structure. It contains the offset to the next item to be matched in the
3640 pattern string.
3642 The next_item_length field is present from version 1 of the callout
3643 structure. It contains the length of the next item to be matched in the
3644 pattern string. When the callout immediately precedes an alternation
3645 bar, a closing parenthesis, or the end of the pattern, the length is
3646 zero. When the callout precedes an opening parenthesis, the length is
3647 that of the entire subpattern.
3649 The pattern_position and next_item_length fields are intended to help
3650 in distinguishing between different automatic callouts, which all have
3651 the same callout number. However, they are set for all callouts.
3653 The mark field is present from version 2 of the callout structure. In
3654 callouts from pcre_exec() or pcre16_exec() it contains a pointer to the
3655 zero-terminated name of the most recently passed (*MARK), (*PRUNE), or
3656 (*THEN) item in the match, or NULL if no such items have been passed.
3657 Instances of (*PRUNE) or (*THEN) without a name do not obliterate a
3658 previous (*MARK). In callouts from the DFA matching functions this
3659 field always contains NULL.
3664 The external callout function returns an integer to PCRE. If the value
3665 is zero, matching proceeds as normal. If the value is greater than
3666 zero, matching fails at the current point, but the testing of other
3667 matching possibilities goes ahead, just as if a lookahead assertion had
3668 failed. If the value is less than zero, the match is abandoned, the
3669 matching function returns the negative value.
3671 Negative values should normally be chosen from the set of
3672 PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
3673 dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
3674 reserved for use by callout functions; it will never be used by PCRE
3675 itself.
3680 Philip Hazel
3681 University Computing Service
3682 Cambridge CB2 3QH, England.
3687 Last updated: 08 Janurary 2012
3688 Copyright (c) 1997-2012 University of Cambridge.
3689 ------------------------------------------------------------------------------
3695 NAME
3696 PCRE - Perl-compatible regular expressions
3701 This document describes the differences in the ways that PCRE and Perl
3702 handle regular expressions. The differences described here are with
3703 respect to Perl versions 5.10 and above.
3705 1. PCRE has only a subset of Perl's Unicode support. Details of what it
3706 does have are given in the pcreunicode page.
3708 2. PCRE allows repeat quantifiers only on parenthesized assertions, but
3709 they do not mean what you might think. For example, (?!a){3} does not
3710 assert that the next three characters are not "a". It just asserts that
3711 the next character is not "a" three times (in principle: PCRE optimizes
3712 this to run the assertion just once). Perl allows repeat quantifiers on
3713 other assertions such as \b, but these do not seem to have any use.
3715 3. Capturing subpatterns that occur inside negative lookahead asser-
3716 tions are counted, but their entries in the offsets vector are never
3717 set. Perl sets its numerical variables from any such patterns that are
3718 matched before the assertion fails to match something (thereby succeed-
3719 ing), but only if the negative lookahead assertion contains just one
3720 branch.
3722 4. Though binary zero characters are supported in the subject string,
3723 they are not allowed in a pattern string because it is passed as a nor-
3724 mal C string, terminated by zero. The escape sequence \0 can be used in
3725 the pattern to represent a binary zero.
3727 5. The following Perl escape sequences are not supported: \l, \u, \L,
3728 \U, and \N when followed by a character name or Unicode value. (\N on
3729 its own, matching a non-newline character, is supported.) In fact these
3730 are implemented by Perl's general string-handling and are not part of
3731 its pattern matching engine. If any of these are encountered by PCRE,
3732 an error is generated by default. However, if the PCRE_JAVASCRIPT_COM-
3733 PAT option is set, \U and \u are interpreted as JavaScript interprets
3734 them.
3736 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
3737 is built with Unicode character property support. The properties that
3738 can be tested with \p and \P are limited to the general category prop-
3739 erties such as Lu and Nd, script names such as Greek or Han, and the
3740 derived properties Any and L&. PCRE does support the Cs (surrogate)
3741 property, which Perl does not; the Perl documentation says "Because
3742 Perl hides the need for the user to understand the internal representa-
3743 tion of Unicode characters, there is no need to implement the somewhat
3744 messy concept of surrogates."
3746 7. PCRE implements a simpler version of \X than Perl, which changed to
3747 make \X match what Unicode calls an "extended grapheme cluster". This
3748 is more complicated than an extended Unicode sequence, which is what
3749 PCRE matches.
3751 8. PCRE does support the \Q...\E escape for quoting substrings. Charac-
3752 ters in between are treated as literals. This is slightly different
3753 from Perl in that $ and @ are also handled as literals inside the
3754 quotes. In Perl, they cause variable interpolation (but of course PCRE
3755 does not have variables). Note the following examples:
3757 Pattern PCRE matches Perl matches
3759 \Qabc$xyz\E abc$xyz abc followed by the
3760 contents of $xyz
3761 \Qabc\$xyz\E abc\$xyz abc\$xyz
3762 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3764 The \Q...\E sequence is recognized both inside and outside character
3765 classes.
3767 9. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
3768 constructions. However, there is support for recursive patterns. This
3769 is not available in Perl 5.8, but it is in Perl 5.10. Also, the PCRE
3770 "callout" feature allows an external function to be called during pat-
3771 tern matching. See the pcrecallout documentation for details.
3773 10. Subpatterns that are called as subroutines (whether or not recur-
3774 sively) are always treated as atomic groups in PCRE. This is like
3775 Python, but unlike Perl. Captured values that are set outside a sub-
3776 routine call can be reference from inside in PCRE, but not in Perl.
3777 There is a discussion that explains these differences in more detail in
3778 the section on recursion differences from Perl in the pcrepattern page.
3780 11. If (*THEN) is present in a group that is called as a subroutine,
3781 its action is limited to that group, even if the group does not contain
3782 any | characters.
3784 12. There are some differences that are concerned with the settings of
3785 captured strings when part of a pattern is repeated. For example,
3786 matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
3787 unset, but in PCRE it is set to "b".
3789 13. PCRE's handling of duplicate subpattern numbers and duplicate sub-
3790 pattern names is not as general as Perl's. This is a consequence of the
3791 fact the PCRE works internally just with numbers, using an external ta-
3792 ble to translate between numbers and names. In particular, a pattern
3793 such as (?|(?<a>A)|(?<b)B), where the two capturing parentheses have
3794 the same number but different names, is not supported, and causes an
3795 error at compile time. If it were allowed, it would not be possible to
3796 distinguish which parentheses matched, because both names map to cap-
3797 turing subpattern number 1. To avoid this confusing situation, an error
3798 is given at compile time.
3800 14. Perl recognizes comments in some places that PCRE does not, for
3801 example, between the ( and ? at the start of a subpattern. If the /x
3802 modifier is set, Perl allows whitespace between ( and ? but PCRE never
3803 does, even if the PCRE_EXTENDED option is set.
3805 15. PCRE provides some extensions to the Perl regular expression facil-
3806 ities. Perl 5.10 includes new features that are not in earlier ver-
3807 sions of Perl, some of which (such as named parentheses) have been in
3808 PCRE for some time. This list is with respect to Perl 5.10:
3810 (a) Although lookbehind assertions in PCRE must match fixed length
3811 strings, each alternative branch of a lookbehind assertion can match a
3812 different length of string. Perl requires them all to have the same
3813 length.
3815 (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
3816 meta-character matches only at the very end of the string.
3818 (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
3819 cial meaning is faulted. Otherwise, like Perl, the backslash is quietly
3820 ignored. (Perl can be made to issue a warning.)
3822 (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
3823 fiers is inverted, that is, by default they are not greedy, but if fol-
3824 lowed by a question mark they are.
3826 (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
3827 tried only at the first matching position in the subject string.
3830 and PCRE_NO_AUTO_CAPTURE options for pcre_exec() have no Perl equiva-
3831 lents.
3833 (g) The \R escape sequence can be restricted to match only CR, LF, or
3834 CRLF by the PCRE_BSR_ANYCRLF option.
3836 (h) The callout facility is PCRE-specific.
3838 (i) The partial matching facility is PCRE-specific.
3840 (j) Patterns compiled by PCRE can be saved and re-used at a later time,
3841 even on different hosts that have the other endianness. However, this
3842 does not apply to optimized data created by the just-in-time compiler.
3844 (k) The alternative matching functions (pcre_dfa_exec() and
3845 pcre16_dfa_exec()) match in a different way and are not Perl-compati-
3846 ble.
3848 (l) PCRE recognizes some special sequences such as (*CR) at the start
3849 of a pattern that set overall options that cannot be changed within the
3850 pattern.
3855 Philip Hazel
3856 University Computing Service
3857 Cambridge CB2 3QH, England.
3862 Last updated: 08 Januray 2012
3863 Copyright (c) 1997-2012 University of Cambridge.
3864 ------------------------------------------------------------------------------
3870 NAME
3871 PCRE - Perl-compatible regular expressions
3876 The syntax and semantics of the regular expressions that are supported
3877 by PCRE are described in detail below. There is a quick-reference syn-
3878 tax summary in the pcresyntax page. PCRE tries to match Perl syntax and
3879 semantics as closely as it can. PCRE also supports some alternative
3880 regular expression syntax (which does not conflict with the Perl syn-
3881 tax) in order to provide some compatibility with regular expressions in
3882 Python, .NET, and Oniguruma.
3884 Perl's regular expressions are described in its own documentation, and
3885 regular expressions in general are covered in a number of books, some
3886 of which have copious examples. Jeffrey Friedl's "Mastering Regular
3887 Expressions", published by O'Reilly, covers regular expressions in
3888 great detail. This description of PCRE's regular expressions is
3889 intended as reference material.
3891 The original operation of PCRE was on strings of one-byte characters.
3892 However, there is now also support for UTF-8 strings in the original
3893 library, and a second library that supports 16-bit and UTF-16 character
3894 strings. To use these features, PCRE must be built to include appropri-
3895 ate support. When using UTF strings you must either call the compiling
3896 function with the PCRE_UTF8 or PCRE_UTF16 option, or the pattern must
3897 start with one of these special sequences:
3899 (*UTF8)
3900 (*UTF16)
3902 Starting a pattern with such a sequence is equivalent to setting the
3903 relevant option. This feature is not Perl-compatible. How setting a UTF
3904 mode affects pattern matching is mentioned in several places below.
3905 There is also a summary of features in the pcreunicode page.
3907 Another special sequence that may appear at the start of a pattern or
3908 in combination with (*UTF8) or (*UTF16) is:
3910 (*UCP)
3912 This has the same effect as setting the PCRE_UCP option: it causes
3913 sequences such as \d and \w to use Unicode properties to determine
3914 character types, instead of recognizing only characters with codes less
3915 than 128 via a lookup table.
3917 If a pattern starts with (*NO_START_OPT), it has the same effect as
3918 setting the PCRE_NO_START_OPTIMIZE option either at compile or matching
3919 time. There are also some more of these special sequences that are con-
3920 cerned with the handling of newlines; they are described below.
3922 The remainder of this document discusses the patterns that are sup-
3923 ported by PCRE when one its main matching functions, pcre_exec()
3924 (8-bit) or pcre16_exec() (16-bit), is used. PCRE also has alternative
3925 matching functions, pcre_dfa_exec() and pcre16_dfa_exec(), which match
3926 using a different algorithm that is not Perl-compatible. Some of the
3927 features discussed below are not available when DFA matching is used.
3928 The advantages and disadvantages of the alternative functions, and how
3929 they differ from the normal functions, are discussed in the pcrematch-
3930 ing page.
3935 PCRE supports five different conventions for indicating line breaks in
3936 strings: a single CR (carriage return) character, a single LF (line-
3937 feed) character, the two-character sequence CRLF, any of the three pre-
3938 ceding, or any Unicode newline sequence. The pcreapi page has further
3939 discussion about newlines, and shows how to set the newline convention
3940 in the options arguments for the compiling and matching functions.
3942 It is also possible to specify a newline convention by starting a pat-
3943 tern string with one of the following five sequences:
3945 (*CR) carriage return
3946 (*LF) linefeed
3947 (*CRLF) carriage return, followed by linefeed
3948 (*ANYCRLF) any of the three above
3949 (*ANY) all Unicode newline sequences
3951 These override the default and the options given to the compiling func-
3952 tion. For example, on a Unix system where LF is the default newline
3953 sequence, the pattern
3955 (*CR)a.b
3957 changes the convention to CR. That pattern matches "a\nb" because LF is
3958 no longer a newline. Note that these special settings, which are not
3959 Perl-compatible, are recognized only at the very start of a pattern,
3960 and that they must be in upper case. If more than one of them is
3961 present, the last one is used.
3963 The newline convention affects the interpretation of the dot metachar-
3964 acter when PCRE_DOTALL is not set, and also the behaviour of \N. How-
3965 ever, it does not affect what the \R escape sequence matches. By
3966 default, this is any Unicode newline sequence, for Perl compatibility.
3967 However, this can be changed; see the description of \R in the section
3968 entitled "Newline sequences" below. A change of \R setting can be com-
3969 bined with a change of newline convention.
3974 A regular expression is a pattern that is matched against a subject
3975 string from left to right. Most characters stand for themselves in a
3976 pattern, and match the corresponding characters in the subject. As a
3977 trivial example, the pattern
3979 The quick brown fox
3981 matches a portion of a subject string that is identical to itself. When
3982 caseless matching is specified (the PCRE_CASELESS option), letters are
3983 matched independently of case. In a UTF mode, PCRE always understands
3984 the concept of case for characters whose values are less than 128, so
3985 caseless matching is always possible. For characters with higher val-
3986 ues, the concept of case is supported if PCRE is compiled with Unicode
3987 property support, but not otherwise. If you want to use caseless
3988 matching for characters 128 and above, you must ensure that PCRE is
3989 compiled with Unicode property support as well as with UTF support.
3991 The power of regular expressions comes from the ability to include
3992 alternatives and repetitions in the pattern. These are encoded in the
3993 pattern by the use of metacharacters, which do not stand for themselves
3994 but instead are interpreted in some special way.
3996 There are two different sets of metacharacters: those that are recog-
3997 nized anywhere in the pattern except within square brackets, and those
3998 that are recognized within square brackets. Outside square brackets,
3999 the metacharacters are as follows:
4001 \ general escape character with several uses
4002 ^ assert start of string (or line, in multiline mode)
4003 $ assert end of string (or line, in multiline mode)
4004 . match any character except newline (by default)
4005 [ start character class definition
4006 | start of alternative branch
4007 ( start subpattern
4008 ) end subpattern
4009 ? extends the meaning of (
4010 also 0 or 1 quantifier
4011 also quantifier minimizer
4012 * 0 or more quantifier
4013 + 1 or more quantifier
4014 also "possessive quantifier"
4015 { start min/max quantifier
4017 Part of a pattern that is in square brackets is called a "character
4018 class". In a character class the only metacharacters are:
4020 \ general escape character
4021 ^ negate the class, but only if the first character
4022 - indicates character range
4023 [ POSIX character class (only if followed by POSIX
4024 syntax)
4025 ] terminates the character class
4027 The following sections describe the use of each of the metacharacters.
4032 The backslash character has several uses. Firstly, if it is followed by
4033 a character that is not a number or a letter, it takes away any special
4034 meaning that character may have. This use of backslash as an escape
4035 character applies both inside and outside character classes.
4037 For example, if you want to match a * character, you write \* in the
4038 pattern. This escaping action applies whether or not the following
4039 character would otherwise be interpreted as a metacharacter, so it is
4040 always safe to precede a non-alphanumeric with backslash to specify
4041 that it stands for itself. In particular, if you want to match a back-
4042 slash, you write \\.
4044 In a UTF mode, only ASCII numbers and letters have any special meaning
4045 after a backslash. All other characters (in particular, those whose
4046 codepoints are greater than 127) are treated as literals.
4048 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in
4049 the pattern (other than in a character class) and characters between a
4050 # outside a character class and the next newline are ignored. An escap-
4051 ing backslash can be used to include a whitespace or # character as
4052 part of the pattern.
4054 If you want to remove the special meaning from a sequence of charac-
4055 ters, you can do so by putting them between \Q and \E. This is differ-
4056 ent from Perl in that $ and @ are handled as literals in \Q...\E
4057 sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
4058 tion. Note the following examples:
4060 Pattern PCRE matches Perl matches
4062 \Qabc$xyz\E abc$xyz abc followed by the
4063 contents of $xyz
4064 \Qabc\$xyz\E abc\$xyz abc\$xyz
4065 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4067 The \Q...\E sequence is recognized both inside and outside character
4068 classes. An isolated \E that is not preceded by \Q is ignored. If \Q
4069 is not followed by \E later in the pattern, the literal interpretation
4070 continues to the end of the pattern (that is, \E is assumed at the
4071 end). If the isolated \Q is inside a character class, this causes an
4072 error, because the character class is not terminated.
4074 Non-printing characters
4076 A second use of backslash provides a way of encoding non-printing char-
4077 acters in patterns in a visible manner. There is no restriction on the
4078 appearance of non-printing characters, apart from the binary zero that
4079 terminates a pattern, but when a pattern is being prepared by text
4080 editing, it is often easier to use one of the following escape
4081 sequences than the binary character it represents:
4083 \a alarm, that is, the BEL character (hex 07)
4084 \cx "control-x", where x is any ASCII character
4085 \e escape (hex 1B)
4086 \f formfeed (hex 0C)
4087 \n linefeed (hex 0A)
4088 \r carriage return (hex 0D)
4089 \t tab (hex 09)
4090 \ddd character with octal code ddd, or back reference
4091 \xhh character with hex code hh
4092 \x{hhh..} character with hex code hhh.. (non-JavaScript mode)
4093 \uhhhh character with hex code hhhh (JavaScript mode only)
4095 The precise effect of \cx is as follows: if x is a lower case letter,
4096 it is converted to upper case. Then bit 6 of the character (hex 40) is
4097 inverted. Thus \cz becomes hex 1A (z is 7A), but \c{ becomes hex 3B ({
4098 is 7B), while \c; becomes hex 7B (; is 3B). If the byte following \c
4099 has a value greater than 127, a compile-time error occurs. This locks
4100 out non-ASCII characters in all modes. (When PCRE is compiled in EBCDIC
4101 mode, all byte values are valid. A lower case letter is converted to
4102 upper case, and then the 0xc0 bits are flipped.)
4104 By default, after \x, from zero to two hexadecimal digits are read
4105 (letters can be in upper or lower case). Any number of hexadecimal dig-
4106 its may appear between \x{ and }, but the character code is constrained
4107 as follows:
4109 8-bit non-UTF mode less than 0x100
4110 8-bit UTF-8 mode less than 0x10ffff and a valid codepoint
4111 16-bit non-UTF mode less than 0x10000
4112 16-bit UTF-16 mode less than 0x10ffff and a valid codepoint
4114 Invalid Unicode codepoints are the range 0xd800 to 0xdfff (the so-
4115 called "surrogate" codepoints).
4117 If characters other than hexadecimal digits appear between \x{ and },
4118 or if there is no terminating }, this form of escape is not recognized.
4119 Instead, the initial \x will be interpreted as a basic hexadecimal
4120 escape, with no following digits, giving a character whose value is
4121 zero.
4123 If the PCRE_JAVASCRIPT_COMPAT option is set, the interpretation of \x
4124 is as just described only when it is followed by two hexadecimal dig-
4125 its. Otherwise, it matches a literal "x" character. In JavaScript
4126 mode, support for code points greater than 256 is provided by \u, which
4127 must be followed by four hexadecimal digits; otherwise it matches a
4128 literal "u" character.
4130 Characters whose value is less than 256 can be defined by either of the
4131 two syntaxes for \x (or by \u in JavaScript mode). There is no differ-
4132 ence in the way they are handled. For example, \xdc is exactly the same
4133 as \x{dc} (or \u00dc in JavaScript mode).
4135 After \0 up to two further octal digits are read. If there are fewer
4136 than two digits, just those that are present are used. Thus the
4137 sequence \0\x\07 specifies two binary zeros followed by a BEL character
4138 (code value 7). Make sure you supply two digits after the initial zero
4139 if the pattern character that follows is itself an octal digit.
4141 The handling of a backslash followed by a digit other than 0 is compli-
4142 cated. Outside a character class, PCRE reads it and any following dig-
4143 its as a decimal number. If the number is less than 10, or if there
4144 have been at least that many previous capturing left parentheses in the
4145 expression, the entire sequence is taken as a back reference. A
4146 description of how this works is given later, following the discussion
4147 of parenthesized subpatterns.
4149 Inside a character class, or if the decimal number is greater than 9
4150 and there have not been that many capturing subpatterns, PCRE re-reads
4151 up to three octal digits following the backslash, and uses them to gen-
4152 erate a data character. Any subsequent digits stand for themselves. The
4153 value of the character is constrained in the same way as characters
4154 specified in hexadecimal. For example:
4156 \040 is another way of writing a space
4157 \40 is the same, provided there are fewer than 40
4158 previous capturing subpatterns
4159 \7 is always a back reference
4160 \11 might be a back reference, or another way of
4161 writing a tab
4162 \011 is always a tab
4163 \0113 is a tab followed by the character "3"
4164 \113 might be a back reference, otherwise the
4165 character with octal code 113
4166 \377 might be a back reference, otherwise
4167 the value 255 (decimal)
4168 \81 is either a back reference, or a binary zero
4169 followed by the two characters "8" and "1"
4171 Note that octal values of 100 or greater must not be introduced by a
4172 leading zero, because no more than three octal digits are ever read.
4174 All the sequences that define a single character value can be used both
4175 inside and outside character classes. In addition, inside a character
4176 class, \b is interpreted as the backspace character (hex 08).
4178 \N is not allowed in a character class. \B, \R, and \X are not special
4179 inside a character class. Like other unrecognized escape sequences,
4180 they are treated as the literal characters "B", "R", and "X" by
4181 default, but cause an error if the PCRE_EXTRA option is set. Outside a
4182 character class, these sequences have different meanings.
4184 Unsupported escape sequences
4186 In Perl, the sequences \l, \L, \u, and \U are recognized by its string
4187 handler and used to modify the case of following characters. By
4188 default, PCRE does not support these escape sequences. However, if the
4189 PCRE_JAVASCRIPT_COMPAT option is set, \U matches a "U" character, and
4190 \u can be used to define a character by code point, as described in the
4191 previous section.
4193 Absolute and relative back references
4195 The sequence \g followed by an unsigned or a negative number, option-
4196 ally enclosed in braces, is an absolute or relative back reference. A
4197 named back reference can be coded as \g{name}. Back references are dis-
4198 cussed later, following the discussion of parenthesized subpatterns.
4200 Absolute and relative subroutine calls
4202 For compatibility with Oniguruma, the non-Perl syntax \g followed by a
4203 name or a number enclosed either in angle brackets or single quotes, is
4204 an alternative syntax for referencing a subpattern as a "subroutine".
4205 Details are discussed later. Note that \g{...} (Perl syntax) and
4206 \g<...> (Oniguruma syntax) are not synonymous. The former is a back
4207 reference; the latter is a subroutine call.
4209 Generic character types
4211 Another use of backslash is for specifying generic character types:
4213 \d any decimal digit
4214 \D any character that is not a decimal digit
4215 \h any horizontal whitespace character
4216 \H any character that is not a horizontal whitespace character
4217 \s any whitespace character
4218 \S any character that is not a whitespace character
4219 \v any vertical whitespace character
4220 \V any character that is not a vertical whitespace character
4221 \w any "word" character
4222 \W any "non-word" character
4224 There is also the single sequence \N, which matches a non-newline char-
4225 acter. This is the same as the "." metacharacter when PCRE_DOTALL is
4226 not set. Perl also uses \N to match characters by name; PCRE does not
4227 support this.
4229 Each pair of lower and upper case escape sequences partitions the com-
4230 plete set of characters into two disjoint sets. Any given character
4231 matches one, and only one, of each pair. The sequences can appear both
4232 inside and outside character classes. They each match one character of
4233 the appropriate type. If the current matching point is at the end of
4234 the subject string, all of them fail, because there is no character to
4235 match.
4237 For compatibility with Perl, \s does not match the VT character (code
4238 11). This makes it different from the the POSIX "space" class. The \s
4239 characters are HT (9), LF (10), FF (12), CR (13), and space (32). If
4240 "use locale;" is included in a Perl script, \s may match the VT charac-
4241 ter. In PCRE, it never does.
4243 A "word" character is an underscore or any character that is a letter
4244 or digit. By default, the definition of letters and digits is con-
4245 trolled by PCRE's low-valued character tables, and may vary if locale-
4246 specific matching is taking place (see "Locale support" in the pcreapi
4247 page). For example, in a French locale such as "fr_FR" in Unix-like
4248 systems, or "french" in Windows, some character codes greater than 128
4249 are used for accented letters, and these are then matched by \w. The
4250 use of locales with Unicode is discouraged.
4252 By default, in a UTF mode, characters with values greater than 128
4253 never match \d, \s, or \w, and always match \D, \S, and \W. These
4254 sequences retain their original meanings from before UTF support was
4255 available, mainly for efficiency reasons. However, if PCRE is compiled
4256 with Unicode property support, and the PCRE_UCP option is set, the be-
4257 haviour is changed so that Unicode properties are used to determine
4258 character types, as follows:
4260 \d any character that \p{Nd} matches (decimal digit)
4261 \s any character that \p{Z} matches, plus HT, LF, FF, CR
4262 \w any character that \p{L} or \p{N} matches, plus underscore
4264 The upper case escapes match the inverse sets of characters. Note that
4265 \d matches only decimal digits, whereas \w matches any Unicode digit,
4266 as well as any Unicode letter, and underscore. Note also that PCRE_UCP
4267 affects \b, and \B because they are defined in terms of \w and \W.
4268 Matching these sequences is noticeably slower when PCRE_UCP is set.
4270 The sequences \h, \H, \v, and \V are features that were added to Perl
4271 at release 5.10. In contrast to the other sequences, which match only
4272 ASCII characters by default, these always match certain high-valued
4273 codepoints, whether or not PCRE_UCP is set. The horizontal space char-
4274 acters are:
4276 U+0009 Horizontal tab
4277 U+0020 Space
4278 U+00A0 Non-break space
4279 U+1680 Ogham space mark
4280 U+180E Mongolian vowel separator
4281 U+2000 En quad
4282 U+2001 Em quad
4283 U+2002 En space
4284 U+2003 Em space
4285 U+2004 Three-per-em space
4286 U+2005 Four-per-em space
4287 U+2006 Six-per-em space
4288 U+2007 Figure space
4289 U+2008 Punctuation space
4290 U+2009 Thin space
4291 U+200A Hair space
4292 U+202F Narrow no-break space
4293 U+205F Medium mathematical space
4294 U+3000 Ideographic space
4296 The vertical space characters are:
4298 U+000A Linefeed
4299 U+000B Vertical tab
4300 U+000C Formfeed
4301 U+000D Carriage return
4302 U+0085 Next line
4303 U+2028 Line separator
4304 U+2029 Paragraph separator
4306 In 8-bit, non-UTF-8 mode, only the characters with codepoints less than
4307 256 are relevant.
4309 Newline sequences
4311 Outside a character class, by default, the escape sequence \R matches
4312 any Unicode newline sequence. In 8-bit non-UTF-8 mode \R is equivalent
4313 to the following:
4315 (?>\r\n|\n|\x0b|\f|\r|\x85)
4317 This is an example of an "atomic group", details of which are given
4318 below. This particular group matches either the two-character sequence
4319 CR followed by LF, or one of the single characters LF (linefeed,
4320 U+000A), VT (vertical tab, U+000B), FF (formfeed, U+000C), CR (carriage
4321 return, U+000D), or NEL (next line, U+0085). The two-character sequence
4322 is treated as a single unit that cannot be split.
4324 In other modes, two additional characters whose codepoints are greater
4325 than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
4326 rator, U+2029). Unicode character property support is not needed for
4327 these characters to be recognized.
4329 It is possible to restrict \R to match only CR, LF, or CRLF (instead of
4330 the complete set of Unicode line endings) by setting the option
4331 PCRE_BSR_ANYCRLF either at compile time or when the pattern is matched.
4332 (BSR is an abbrevation for "backslash R".) This can be made the default
4333 when PCRE is built; if this is the case, the other behaviour can be
4334 requested via the PCRE_BSR_UNICODE option. It is also possible to
4335 specify these settings by starting a pattern string with one of the
4336 following sequences:
4338 (*BSR_ANYCRLF) CR, LF, or CRLF only
4339 (*BSR_UNICODE) any Unicode newline sequence
4341 These override the default and the options given to the compiling func-
4342 tion, but they can themselves be overridden by options given to a
4343 matching function. Note that these special settings, which are not
4344 Perl-compatible, are recognized only at the very start of a pattern,
4345 and that they must be in upper case. If more than one of them is
4346 present, the last one is used. They can be combined with a change of
4347 newline convention; for example, a pattern can start with:
4351 They can also be combined with the (*UTF8), (*UTF16), or (*UCP) special
4352 sequences. Inside a character class, \R is treated as an unrecognized
4353 escape sequence, and so matches the letter "R" by default, but causes
4354 an error if PCRE_EXTRA is set.
4356 Unicode character properties
4358 When PCRE is built with Unicode character property support, three addi-
4359 tional escape sequences that match characters with specific properties
4360 are available. When in 8-bit non-UTF-8 mode, these sequences are of
4361 course limited to testing characters whose codepoints are less than
4362 256, but they do work in this mode. The extra escape sequences are:
4364 \p{xx} a character with the xx property
4365 \P{xx} a character without the xx property
4366 \X an extended Unicode sequence
4368 The property names represented by xx above are limited to the Unicode
4369 script names, the general category properties, "Any", which matches any
4370 character (including newline), and some special PCRE properties
4371 (described in the next section). Other Perl properties such as "InMu-
4372 sicalSymbols" are not currently supported by PCRE. Note that \P{Any}
4373 does not match any characters, so always causes a match failure.
4375 Sets of Unicode characters are defined as belonging to certain scripts.
4376 A character from one of these sets can be matched using a script name.
4377 For example:
4379 \p{Greek}
4380 \P{Han}
4382 Those that are not part of an identified script are lumped together as
4383 "Common". The current list of scripts is:
4385 Arabic, Armenian, Avestan, Balinese, Bamum, Batak, Bengali, Bopomofo,
4386 Brahmi, Braille, Buginese, Buhid, Canadian_Aboriginal, Carian, Chakma,
4387 Cham, Cherokee, Common, Coptic, Cuneiform, Cypriot, Cyrillic, Deseret,
4388 Devanagari, Egyptian_Hieroglyphs, Ethiopic, Georgian, Glagolitic,
4389 Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
4390 gana, Imperial_Aramaic, Inherited, Inscriptional_Pahlavi, Inscrip-
4391 tional_Parthian, Javanese, Kaithi, Kannada, Katakana, Kayah_Li,
4392 Kharoshthi, Khmer, Lao, Latin, Lepcha, Limbu, Linear_B, Lisu, Lycian,
4393 Lydian, Malayalam, Mandaic, Meetei_Mayek, Meroitic_Cursive,
4394 Meroitic_Hieroglyphs, Miao, Mongolian, Myanmar, New_Tai_Lue, Nko,
4395 Ogham, Old_Italic, Old_Persian, Old_South_Arabian, Old_Turkic,
4396 Ol_Chiki, Oriya, Osmanya, Phags_Pa, Phoenician, Rejang, Runic, Samari-
4397 tan, Saurashtra, Sharada, Shavian, Sinhala, Sora_Sompeng, Sundanese,
4398 Syloti_Nagri, Syriac, Tagalog, Tagbanwa, Tai_Le, Tai_Tham, Tai_Viet,
4399 Takri, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Vai,
4400 Yi.
4402 Each character has exactly one Unicode general category property, spec-
4403 ified by a two-letter abbreviation. For compatibility with Perl, nega-
4404 tion can be specified by including a circumflex between the opening
4405 brace and the property name. For example, \p{^Lu} is the same as
4406 \P{Lu}.
4408 If only one letter is specified with \p or \P, it includes all the gen-
4409 eral category properties that start with that letter. In this case, in
4410 the absence of negation, the curly brackets in the escape sequence are
4411 optional; these two examples have the same effect:
4413 \p{L}
4414 \pL
4416 The following general category property codes are supported:
4418 C Other
4419 Cc Control
4420 Cf Format
4421 Cn Unassigned
4422 Co Private use
4423 Cs Surrogate
4425 L Letter
4426 Ll Lower case letter
4427 Lm Modifier letter
4428 Lo Other letter
4429 Lt Title case letter
4430 Lu Upper case letter
4432 M Mark
4433 Mc Spacing mark
4434 Me Enclosing mark
4435 Mn Non-spacing mark
4437 N Number
4438 Nd Decimal number
4439 Nl Letter number
4440 No Other number
4442 P Punctuation
4443 Pc Connector punctuation
4444 Pd Dash punctuation
4445 Pe Close punctuation
4446 Pf Final punctuation
4447 Pi Initial punctuation
4448 Po Other punctuation
4449 Ps Open punctuation
4451 S Symbol
4452 Sc Currency symbol
4453 Sk Modifier symbol
4454 Sm Mathematical symbol
4455 So Other symbol
4457 Z Separator
4458 Zl Line separator
4459 Zp Paragraph separator
4460 Zs Space separator
4462 The special property L& is also supported: it matches a character that
4463 has the Lu, Ll, or Lt property, in other words, a letter that is not
4464 classified as a modifier or "other".
4466 The Cs (Surrogate) property applies only to characters in the range
4467 U+D800 to U+DFFF. Such characters are not valid in Unicode strings and
4468 so cannot be tested by PCRE, unless UTF validity checking has been
4469 turned off (see the discussion of PCRE_NO_UTF8_CHECK and
4470 PCRE_NO_UTF16_CHECK in the pcreapi page). Perl does not support the Cs
4471 property.
4473 The long synonyms for property names that Perl supports (such as
4474 \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
4475 any of these properties with "Is".
4477 No character that is in the Unicode table has the Cn (unassigned) prop-
4478 erty. Instead, this property is assumed for any code point that is not
4479 in the Unicode table.
4481 Specifying caseless matching does not affect these escape sequences.
4482 For example, \p{Lu} always matches only upper case letters.
4484 The \X escape matches any number of Unicode characters that form an
4485 extended Unicode sequence. \X is equivalent to
4487 (?>\PM\pM*)
4489 That is, it matches a character without the "mark" property, followed
4490 by zero or more characters with the "mark" property, and treats the
4491 sequence as an atomic group (see below). Characters with the "mark"
4492 property are typically accents that affect the preceding character.
4493 None of them have codepoints less than 256, so in 8-bit non-UTF-8 mode
4494 \X matches any one character.
4496 Note that recent versions of Perl have changed \X to match what Unicode
4497 calls an "extended grapheme cluster", which has a more complicated def-
4498 inition.
4500 Matching characters by Unicode property is not fast, because PCRE has
4501 to search a structure that contains data for over fifteen thousand
4502 characters. That is why the traditional escape sequences such as \d and
4503 \w do not use Unicode properties in PCRE by default, though you can
4504 make them do so by setting the PCRE_UCP option or by starting the pat-
4505 tern with (*UCP).
4507 PCRE's additional properties
4509 As well as the standard Unicode properties described in the previous
4510 section, PCRE supports four more that make it possible to convert tra-
4511 ditional escape sequences such as \w and \s and POSIX character classes
4512 to use Unicode properties. PCRE uses these non-standard, non-Perl prop-
4513 erties internally when PCRE_UCP is set. They are:
4515 Xan Any alphanumeric character
4516 Xps Any POSIX space character
4517 Xsp Any Perl space character
4518 Xwd Any Perl "word" character
4520 Xan matches characters that have either the L (letter) or the N (num-
4521 ber) property. Xps matches the characters tab, linefeed, vertical tab,
4522 formfeed, or carriage return, and any other character that has the Z
4523 (separator) property. Xsp is the same as Xps, except that vertical tab
4524 is excluded. Xwd matches the same characters as Xan, plus underscore.
4526 Resetting the match start
4528 The escape sequence \K causes any previously matched characters not to
4529 be included in the final matched sequence. For example, the pattern:
4531 foo\Kbar
4533 matches "foobar", but reports that it has matched "bar". This feature
4534 is similar to a lookbehind assertion (described below). However, in
4535 this case, the part of the subject before the real match does not have
4536 to be of fixed length, as lookbehind assertions do. The use of \K does
4537 not interfere with the setting of captured substrings. For example,
4538 when the pattern
4540 (foo)\Kbar
4542 matches "foobar", the first substring is still set to "foo".
4544 Perl documents that the use of \K within assertions is "not well
4545 defined". In PCRE, \K is acted upon when it occurs inside positive
4546 assertions, but is ignored in negative assertions.
4548 Simple assertions
4550 The final use of backslash is for certain simple assertions. An asser-
4551 tion specifies a condition that has to be met at a particular point in
4552 a match, without consuming any characters from the subject string. The
4553 use of subpatterns for more complicated assertions is described below.
4554 The backslashed assertions are:
4556 \b matches at a word boundary
4557 \B matches when not at a word boundary
4558 \A matches at the start of the subject
4559 \Z matches at the end of the subject
4560 also matches before a newline at the end of the subject
4561 \z matches only at the end of the subject
4562 \G matches at the first matching position in the subject
4564 Inside a character class, \b has a different meaning; it matches the
4565 backspace character. If any other of these assertions appears in a
4566 character class, by default it matches the corresponding literal char-
4567 acter (for example, \B matches the letter B). However, if the
4568 PCRE_EXTRA option is set, an "invalid escape sequence" error is gener-
4569 ated instead.
4571 A word boundary is a position in the subject string where the current
4572 character and the previous character do not both match \w or \W (i.e.
4573 one matches \w and the other matches \W), or the start or end of the
4574 string if the first or last character matches \w, respectively. In a
4575 UTF mode, the meanings of \w and \W can be changed by setting the
4576 PCRE_UCP option. When this is done, it also affects \b and \B. Neither
4577 PCRE nor Perl has a separate "start of word" or "end of word" metase-
4578 quence. However, whatever follows \b normally determines which it is.
4579 For example, the fragment \ba matches "a" at the start of a word.
4581 The \A, \Z, and \z assertions differ from the traditional circumflex
4582 and dollar (described in the next section) in that they only ever match
4583 at the very start and end of the subject string, whatever options are
4584 set. Thus, they are independent of multiline mode. These three asser-
4585 tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
4586 affect only the behaviour of the circumflex and dollar metacharacters.
4587 However, if the startoffset argument of pcre_exec() is non-zero, indi-
4588 cating that matching is to start at a point other than the beginning of
4589 the subject, \A can never match. The difference between \Z and \z is
4590 that \Z matches before a newline at the end of the string as well as at
4591 the very end, whereas \z matches only at the end.
4593 The \G assertion is true only when the current matching position is at
4594 the start point of the match, as specified by the startoffset argument
4595 of pcre_exec(). It differs from \A when the value of startoffset is
4596 non-zero. By calling pcre_exec() multiple times with appropriate argu-
4597 ments, you can mimic Perl's /g option, and it is in this kind of imple-
4598 mentation where \G can be useful.
4600 Note, however, that PCRE's interpretation of \G, as the start of the
4601 current match, is subtly different from Perl's, which defines it as the
4602 end of the previous match. In Perl, these can be different when the
4603 previously matched string was empty. Because PCRE does just one match
4604 at a time, it cannot reproduce this behaviour.
4606 If all the alternatives of a pattern begin with \G, the expression is
4607 anchored to the starting match position, and the "anchored" flag is set
4608 in the compiled regular expression.
4613 Outside a character class, in the default matching mode, the circumflex
4614 character is an assertion that is true only if the current matching
4615 point is at the start of the subject string. If the startoffset argu-
4616 ment of pcre_exec() is non-zero, circumflex can never match if the
4617 PCRE_MULTILINE option is unset. Inside a character class, circumflex
4618 has an entirely different meaning (see below).
4620 Circumflex need not be the first character of the pattern if a number
4621 of alternatives are involved, but it should be the first thing in each
4622 alternative in which it appears if the pattern is ever to match that
4623 branch. If all possible alternatives start with a circumflex, that is,
4624 if the pattern is constrained to match only at the start of the sub-
4625 ject, it is said to be an "anchored" pattern. (There are also other
4626 constructs that can cause a pattern to be anchored.)
4628 A dollar character is an assertion that is true only if the current
4629 matching point is at the end of the subject string, or immediately
4630 before a newline at the end of the string (by default). Dollar need not
4631 be the last character of the pattern if a number of alternatives are
4632 involved, but it should be the last item in any branch in which it
4633 appears. Dollar has no special meaning in a character class.
4635 The meaning of dollar can be changed so that it matches only at the
4636 very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
4637 compile time. This does not affect the \Z assertion.
4639 The meanings of the circumflex and dollar characters are changed if the
4640 PCRE_MULTILINE option is set. When this is the case, a circumflex
4641 matches immediately after internal newlines as well as at the start of
4642 the subject string. It does not match after a newline that ends the
4643 string. A dollar matches before any newlines in the string, as well as
4644 at the very end, when PCRE_MULTILINE is set. When newline is specified
4645 as the two-character sequence CRLF, isolated CR and LF characters do
4646 not indicate newlines.
4648 For example, the pattern /^abc$/ matches the subject string "def\nabc"
4649 (where \n represents a newline) in multiline mode, but not otherwise.
4650 Consequently, patterns that are anchored in single line mode because
4651 all branches start with ^ are not anchored in multiline mode, and a
4652 match for circumflex is possible when the startoffset argument of
4653 pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
4654 PCRE_MULTILINE is set.
4656 Note that the sequences \A, \Z, and \z can be used to match the start
4657 and end of the subject in both modes, and if all branches of a pattern
4658 start with \A it is always anchored, whether or not PCRE_MULTILINE is
4659 set.
4664 Outside a character class, a dot in the pattern matches any one charac-
4665 ter in the subject string except (by default) a character that signi-
4666 fies the end of a line.
4668 When a line ending is defined as a single character, dot never matches
4669 that character; when the two-character sequence CRLF is used, dot does
4670 not match CR if it is immediately followed by LF, but otherwise it
4671 matches all characters (including isolated CRs and LFs). When any Uni-
4672 code line endings are being recognized, dot does not match CR or LF or
4673 any of the other line ending characters.
4675 The behaviour of dot with regard to newlines can be changed. If the
4676 PCRE_DOTALL option is set, a dot matches any one character, without
4677 exception. If the two-character sequence CRLF is present in the subject
4678 string, it takes two dots to match it.
4680 The handling of dot is entirely independent of the handling of circum-
4681 flex and dollar, the only relationship being that they both involve
4682 newlines. Dot has no special meaning in a character class.
4684 The escape sequence \N behaves like a dot, except that it is not
4685 affected by the PCRE_DOTALL option. In other words, it matches any
4686 character except one that signifies the end of a line. Perl also uses
4687 \N to match characters by name; PCRE does not support this.
4692 Outside a character class, the escape sequence \C matches any one data
4693 unit, whether or not a UTF mode is set. In the 8-bit library, one data
4694 unit is one byte; in the 16-bit library it is a 16-bit unit. Unlike a
4695 dot, \C always matches line-ending characters. The feature is provided
4696 in Perl in order to match individual bytes in UTF-8 mode, but it is
4697 unclear how it can usefully be used. Because \C breaks up characters
4698 into individual data units, matching one unit with \C in a UTF mode
4699 means that the rest of the string may start with a malformed UTF char-
4700 acter. This has undefined results, because PCRE assumes that it is
4701 dealing with valid UTF strings (and by default it checks this at the
4702 start of processing unless the PCRE_NO_UTF8_CHECK or
4703 PCRE_NO_UTF16_CHECK option is used).
4705 PCRE does not allow \C to appear in lookbehind assertions (described
4706 below) in a UTF mode, because this would make it impossible to calcu-
4707 late the length of the lookbehind.
4709 In general, the \C escape sequence is best avoided. However, one way of
4710 using it that avoids the problem of malformed UTF characters is to use
4711 a lookahead to check the length of the next character, as in this pat-
4712 tern, which could be used with a UTF-8 string (ignore white space and
4713 line breaks):
4715 (?| (?=[\x00-\x7f])(\C) |
4716 (?=[\x80-\x{7ff}])(\C)(\C) |
4717 (?=[\x{800}-\x{ffff}])(\C)(\C)(\C) |
4718 (?=[\x{10000}-\x{1fffff}])(\C)(\C)(\C)(\C))
4720 A group that starts with (?| resets the capturing parentheses numbers
4721 in each alternative (see "Duplicate Subpattern Numbers" below). The
4722 assertions at the start of each branch check the next UTF-8 character
4723 for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The
4724 character's individual bytes are then captured by the appropriate num-
4725 ber of groups.
4730 An opening square bracket introduces a character class, terminated by a
4731 closing square bracket. A closing square bracket on its own is not spe-
4732 cial by default. However, if the PCRE_JAVASCRIPT_COMPAT option is set,
4733 a lone closing square bracket causes a compile-time error. If a closing
4734 square bracket is required as a member of the class, it should be the
4735 first data character in the class (after an initial circumflex, if
4736 present) or escaped with a backslash.
4738 A character class matches a single character in the subject. In a UTF
4739 mode, the character may be more than one data unit long. A matched
4740 character must be in the set of characters defined by the class, unless
4741 the first character in the class definition is a circumflex, in which
4742 case the subject character must not be in the set defined by the class.
4743 If a circumflex is actually required as a member of the class, ensure
4744 it is not the first character, or escape it with a backslash.
4746 For example, the character class [aeiou] matches any lower case vowel,
4747 while [^aeiou] matches any character that is not a lower case vowel.
4748 Note that a circumflex is just a convenient notation for specifying the
4749 characters that are in the class by enumerating those that are not. A
4750 class that starts with a circumflex is not an assertion; it still con-
4751 sumes a character from the subject string, and therefore it fails if
4752 the current pointer is at the end of the string.
4754 In UTF-8 (UTF-16) mode, characters with values greater than 255
4755 (0xffff) can be included in a class as a literal string of data units,
4756 or by using the \x{ escaping mechanism.
4758 When caseless matching is set, any letters in a class represent both
4759 their upper case and lower case versions, so for example, a caseless
4760 [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
4761 match "A", whereas a caseful version would. In a UTF mode, PCRE always
4762 understands the concept of case for characters whose values are less
4763 than 128, so caseless matching is always possible. For characters with
4764 higher values, the concept of case is supported if PCRE is compiled
4765 with Unicode property support, but not otherwise. If you want to use
4766 caseless matching in a UTF mode for characters 128 and above, you must
4767 ensure that PCRE is compiled with Unicode property support as well as
4768 with UTF support.
4770 Characters that might indicate line breaks are never treated in any
4771 special way when matching character classes, whatever line-ending
4772 sequence is in use, and whatever setting of the PCRE_DOTALL and
4773 PCRE_MULTILINE options is used. A class such as [^a] always matches one
4774 of these characters.
4776 The minus (hyphen) character can be used to specify a range of charac-
4777 ters in a character class. For example, [d-m] matches any letter
4778 between d and m, inclusive. If a minus character is required in a
4779 class, it must be escaped with a backslash or appear in a position
4780 where it cannot be interpreted as indicating a range, typically as the
4781 first or last character in the class.
4783 It is not possible to have the literal character "]" as the end charac-
4784 ter of a range. A pattern such as [W-]46] is interpreted as a class of
4785 two characters ("W" and "-") followed by a literal string "46]", so it
4786 would match "W46]" or "-46]". However, if the "]" is escaped with a
4787 backslash it is interpreted as the end of range, so [W-\]46] is inter-
4788 preted as a class containing a range followed by two other characters.
4789 The octal or hexadecimal representation of "]" can also be used to end
4790 a range.
4792 Ranges operate in the collating sequence of character values. They can
4793 also be used for characters specified numerically, for example
4794 [\000-\037]. Ranges can include any characters that are valid for the
4795 current mode.
4797 If a range that includes letters is used when caseless matching is set,
4798 it matches the letters in either case. For example, [W-c] is equivalent
4799 to [][\\^_`wxyzabc], matched caselessly, and in a non-UTF mode, if
4800 character tables for a French locale are in use, [\xc8-\xcb] matches
4801 accented E characters in both cases. In UTF modes, PCRE supports the
4802 concept of case for characters with values greater than 128 only when
4803 it is compiled with Unicode property support.
4805 The character escape sequences \d, \D, \h, \H, \p, \P, \s, \S, \v, \V,
4806 \w, and \W may appear in a character class, and add the characters that
4807 they match to the class. For example, [\dABCDEF] matches any hexadeci-
4808 mal digit. In UTF modes, the PCRE_UCP option affects the meanings of
4809 \d, \s, \w and their upper case partners, just as it does when they
4810 appear outside a character class, as described in the section entitled
4811 "Generic character types" above. The escape sequence \b has a different
4812 meaning inside a character class; it matches the backspace character.
4813 The sequences \B, \N, \R, and \X are not special inside a character
4814 class. Like any other unrecognized escape sequences, they are treated
4815 as the literal characters "B", "N", "R", and "X" by default, but cause
4816 an error if the PCRE_EXTRA option is set.
4818 A circumflex can conveniently be used with the upper case character
4819 types to specify a more restricted set of characters than the matching
4820 lower case type. For example, the class [^\W_] matches any letter or
4821 digit, but not underscore, whereas [\w] includes underscore. A positive
4822 character class should be read as "something OR something OR ..." and a
4823 negative class as "NOT something AND NOT something AND NOT ...".
4825 The only metacharacters that are recognized in character classes are
4826 backslash, hyphen (only where it can be interpreted as specifying a
4827 range), circumflex (only at the start), opening square bracket (only
4828 when it can be interpreted as introducing a POSIX class name - see the
4829 next section), and the terminating closing square bracket. However,
4830 escaping other non-alphanumeric characters does no harm.
4835 Perl supports the POSIX notation for character classes. This uses names
4836 enclosed by [: and :] within the enclosing square brackets. PCRE also
4837 supports this notation. For example,
4839 [01[:alpha:]%]
4841 matches "0", "1", any alphabetic character, or "%". The supported class
4842 names are:
4844 alnum letters and digits
4845 alpha letters
4846 ascii character codes 0 - 127
4847 blank space or tab only
4848 cntrl control characters
4849 digit decimal digits (same as \d)
4850 graph printing characters, excluding space
4851 lower lower case letters
4852 print printing characters, including space
4853 punct printing characters, excluding letters and digits and space
4854 space white space (not quite the same as \s)
4855 upper upper case letters
4856 word "word" characters (same as \w)
4857 xdigit hexadecimal digits
4859 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
4860 and space (32). Notice that this list includes the VT character (code
4861 11). This makes "space" different to \s, which does not include VT (for
4862 Perl compatibility).
4864 The name "word" is a Perl extension, and "blank" is a GNU extension
4865 from Perl 5.8. Another Perl extension is negation, which is indicated
4866 by a ^ character after the colon. For example,
4868 [12[:^digit:]]
4870 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
4871 POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
4872 these are not supported, and an error is given if they are encountered.
4874 By default, in UTF modes, characters with values greater than 128 do
4875 not match any of the POSIX character classes. However, if the PCRE_UCP
4876 option is passed to pcre_compile(), some of the classes are changed so
4877 that Unicode character properties are used. This is achieved by replac-
4878 ing the POSIX classes by other sequences, as follows:
4880 [:alnum:] becomes \p{Xan}
4881 [:alpha:] becomes \p{L}
4882 [:blank:] becomes \h
4883 [:digit:] becomes \p{Nd}
4884 [:lower:] becomes \p{Ll}
4885 [:space:] becomes \p{Xps}
4886 [:upper:] becomes \p{Lu}
4887 [:word:] becomes \p{Xwd}
4889 Negated versions, such as [:^alpha:] use \P instead of \p. The other
4890 POSIX classes are unchanged, and match only characters with code points
4891 less than 128.
4896 Vertical bar characters are used to separate alternative patterns. For
4897 example, the pattern
4899 gilbert|sullivan
4901 matches either "gilbert" or "sullivan". Any number of alternatives may
4902 appear, and an empty alternative is permitted (matching the empty
4903 string). The matching process tries each alternative in turn, from left
4904 to right, and the first one that succeeds is used. If the alternatives
4905 are within a subpattern (defined below), "succeeds" means matching the
4906 rest of the main pattern as well as the alternative in the subpattern.
4911 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
4912 PCRE_EXTENDED options (which are Perl-compatible) can be changed from
4913 within the pattern by a sequence of Perl option letters enclosed
4914 between "(?" and ")". The option letters are
4916 i for PCRE_CASELESS
4918 s for PCRE_DOTALL
4919 x for PCRE_EXTENDED
4921 For example, (?im) sets caseless, multiline matching. It is also possi-
4922 ble to unset these options by preceding the letter with a hyphen, and a
4923 combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
4925 is also permitted. If a letter appears both before and after the
4926 hyphen, the option is unset.
4929 can be changed in the same way as the Perl-compatible options by using
4930 the characters J, U and X respectively.
4932 When one of these option changes occurs at top level (that is, not
4933 inside subpattern parentheses), the change applies to the remainder of
4934 the pattern that follows. If the change is placed right at the start of
4935 a pattern, PCRE extracts it into the global options (and it will there-
4936 fore show up in data extracted by the pcre_fullinfo() function).
4938 An option change within a subpattern (see below for a description of
4939 subpatterns) affects only that part of the subpattern that follows it,
4940 so
4942 (a(?i)b)c
4944 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
4945 used). By this means, options can be made to have different settings
4946 in different parts of the pattern. Any changes made in one alternative
4947 do carry on into subsequent branches within the same subpattern. For
4948 example,
4950 (a(?i)b|c)
4952 matches "ab", "aB", "c", and "C", even though when matching "C" the
4953 first branch is abandoned before the option setting. This is because
4954 the effects of option settings happen at compile time. There would be
4955 some very weird behaviour otherwise.
4957 Note: There are other PCRE-specific options that can be set by the
4958 application when the compiling or matching functions are called. In
4959 some cases the pattern can contain special leading sequences such as
4960 (*CRLF) to override what the application has set or what has been
4961 defaulted. Details are given in the section entitled "Newline
4962 sequences" above. There are also the (*UTF8), (*UTF16), and (*UCP)
4963 leading sequences that can be used to set UTF and Unicode property
4964 modes; they are equivalent to setting the PCRE_UTF8, PCRE_UTF16, and
4965 the PCRE_UCP options, respectively.
4970 Subpatterns are delimited by parentheses (round brackets), which can be
4971 nested. Turning part of a pattern into a subpattern does two things:
4973 1. It localizes a set of alternatives. For example, the pattern
4975 cat(aract|erpillar|)
4977 matches "cataract", "caterpillar", or "cat". Without the parentheses,
4978 it would match "cataract", "erpillar" or an empty string.
4980 2. It sets up the subpattern as a capturing subpattern. This means
4981 that, when the whole pattern matches, that portion of the subject
4982 string that matched the subpattern is passed back to the caller via the
4983 ovector argument of the matching function. (This applies only to the
4984 traditional matching functions; the DFA matching functions do not sup-
4985 port capturing.)
4987 Opening parentheses are counted from left to right (starting from 1) to
4988 obtain numbers for the capturing subpatterns. For example, if the
4989 string "the red king" is matched against the pattern
4991 the ((red|white) (king|queen))
4993 the captured substrings are "red king", "red", and "king", and are num-
4994 bered 1, 2, and 3, respectively.
4996 The fact that plain parentheses fulfil two functions is not always
4997 helpful. There are often times when a grouping subpattern is required
4998 without a capturing requirement. If an opening parenthesis is followed
4999 by a question mark and a colon, the subpattern does not do any captur-
5000 ing, and is not counted when computing the number of any subsequent
5001 capturing subpatterns. For example, if the string "the white queen" is
5002 matched against the pattern
5004 the ((?:red|white) (king|queen))
5006 the captured substrings are "white queen" and "queen", and are numbered
5007 1 and 2. The maximum number of capturing subpatterns is 65535.
5009 As a convenient shorthand, if any option settings are required at the
5010 start of a non-capturing subpattern, the option letters may appear
5011 between the "?" and the ":". Thus the two patterns
5013 (?i:saturday|sunday)
5014 (?:(?i)saturday|sunday)
5016 match exactly the same set of strings. Because alternative branches are
5017 tried from left to right, and options are not reset until the end of
5018 the subpattern is reached, an option setting in one branch does affect
5019 subsequent branches, so the above patterns match "SUNDAY" as well as
5020 "Saturday".
5025 Perl 5.10 introduced a feature whereby each alternative in a subpattern
5026 uses the same numbers for its capturing parentheses. Such a subpattern
5027 starts with (?| and is itself a non-capturing subpattern. For example,
5028 consider this pattern:
5030 (?|(Sat)ur|(Sun))day
5032 Because the two alternatives are inside a (?| group, both sets of cap-
5033 turing parentheses are numbered one. Thus, when the pattern matches,
5034 you can look at captured substring number one, whichever alternative
5035 matched. This construct is useful when you want to capture part, but
5036 not all, of one of a number of alternatives. Inside a (?| group, paren-
5037 theses are numbered as usual, but the number is reset at the start of
5038 each branch. The numbers of any capturing parentheses that follow the
5039 subpattern start after the highest number used in any branch. The fol-
5040 lowing example is taken from the Perl documentation. The numbers under-
5041 neath show in which buffer the captured content will be stored.
5043 # before ---------------branch-reset----------- after
5044 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
5045 # 1 2 2 3 2 3 4
5047 A back reference to a numbered subpattern uses the most recent value
5048 that is set for that number by any subpattern. The following pattern
5049 matches "abcabc" or "defdef":
5051 /(?|(abc)|(def))\1/
5053 In contrast, a subroutine call to a numbered subpattern always refers
5054 to the first one in the pattern with the given number. The following
5055 pattern matches "abcabc" or "defabc":
5057 /(?|(abc)|(def))(?1)/
5059 If a condition test for a subpattern's having matched refers to a non-
5060 unique number, the test is true if any of the subpatterns of that num-
5061 ber have matched.
5063 An alternative approach to using this "branch reset" feature is to use
5064 duplicate named subpatterns, as described in the next section.
5069 Identifying capturing parentheses by number is simple, but it can be
5070 very hard to keep track of the numbers in complicated regular expres-
5071 sions. Furthermore, if an expression is modified, the numbers may
5072 change. To help with this difficulty, PCRE supports the naming of sub-
5073 patterns. This feature was not added to Perl until release 5.10. Python
5074 had the feature earlier, and PCRE introduced it at release 4.0, using
5075 the Python syntax. PCRE now supports both the Perl and the Python syn-
5076 tax. Perl allows identically numbered subpatterns to have different
5077 names, but PCRE does not.
5079 In PCRE, a subpattern can be named in one of three ways: (?<name>...)
5080 or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
5081 to capturing parentheses from other parts of the pattern, such as back
5082 references, recursion, and conditions, can be made by name as well as
5083 by number.
5085 Names consist of up to 32 alphanumeric characters and underscores.
5086 Named capturing parentheses are still allocated numbers as well as
5087 names, exactly as if the names were not present. The PCRE API provides
5088 function calls for extracting the name-to-number translation table from
5089 a compiled pattern. There is also a convenience function for extracting
5090 a captured substring by name.
5092 By default, a name must be unique within a pattern, but it is possible
5093 to relax this constraint by setting the PCRE_DUPNAMES option at compile
5094 time. (Duplicate names are also always permitted for subpatterns with
5095 the same number, set up as described in the previous section.) Dupli-
5096 cate names can be useful for patterns where only one instance of the
5097 named parentheses can match. Suppose you want to match the name of a
5098 weekday, either as a 3-letter abbreviation or as the full name, and in
5099 both cases you want to extract the abbreviation. This pattern (ignoring
5100 the line breaks) does the job:
5102 (?<DN>Mon|Fri|Sun)(?:day)?|
5103 (?<DN>Tue)(?:sday)?|
5104 (?<DN>Wed)(?:nesday)?|
5105 (?<DN>Thu)(?:rsday)?|
5106 (?<DN>Sat)(?:urday)?
5108 There are five capturing substrings, but only one is ever set after a
5109 match. (An alternative way of solving this problem is to use a "branch
5110 reset" subpattern, as described in the previous section.)
5112 The convenience function for extracting the data by name returns the
5113 substring for the first (and in this example, the only) subpattern of
5114 that name that matched. This saves searching to find which numbered
5115 subpattern it was.
5117 If you make a back reference to a non-unique named subpattern from
5118 elsewhere in the pattern, the one that corresponds to the first occur-
5119 rence of the name is used. In the absence of duplicate numbers (see the
5120 previous section) this is the one with the lowest number. If you use a
5121 named reference in a condition test (see the section about conditions
5122 below), either to check whether a subpattern has matched, or to check
5123 for recursion, all subpatterns with the same name are tested. If the
5124 condition is true for any one of them, the overall condition is true.
5125 This is the same behaviour as testing by number. For further details of
5126 the interfaces for handling named subpatterns, see the pcreapi documen-
5127 tation.
5129 Warning: You cannot use different names to distinguish between two sub-
5130 patterns with the same number because PCRE uses only the numbers when
5131 matching. For this reason, an error is given at compile time if differ-
5132 ent names are given to subpatterns with the same number. However, you
5133 can give the same name to subpatterns with the same number, even when
5134 PCRE_DUPNAMES is not set.
5139 Repetition is specified by quantifiers, which can follow any of the
5140 following items:
5142 a literal data character
5143 the dot metacharacter
5144 the \C escape sequence
5145 the \X escape sequence
5146 the \R escape sequence
5147 an escape such as \d or \pL that matches a single character
5148 a character class
5149 a back reference (see next section)
5150 a parenthesized subpattern (including assertions)
5151 a subroutine call to a subpattern (recursive or otherwise)
5153 The general repetition quantifier specifies a minimum and maximum num-
5154 ber of permitted matches, by giving the two numbers in curly brackets
5155 (braces), separated by a comma. The numbers must be less than 65536,
5156 and the first must be less than or equal to the second. For example:
5158 z{2,4}
5160 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
5161 special character. If the second number is omitted, but the comma is
5162 present, there is no upper limit; if the second number and the comma
5163 are both omitted, the quantifier specifies an exact number of required
5164 matches. Thus
5166 [aeiou]{3,}
5168 matches at least 3 successive vowels, but may match many more, while
5170 \d{8}
5172 matches exactly 8 digits. An opening curly bracket that appears in a
5173 position where a quantifier is not allowed, or one that does not match
5174 the syntax of a quantifier, is taken as a literal character. For exam-
5175 ple, {,6} is not a quantifier, but a literal string of four characters.
5177 In UTF modes, quantifiers apply to characters rather than to individual
5178 data units. Thus, for example, \x{100}{2} matches two characters, each
5179 of which is represented by a two-byte sequence in a UTF-8 string. Simi-
5180 larly, \X{3} matches three Unicode extended sequences, each of which
5181 may be several data units long (and they may be of different lengths).
5183 The quantifier {0} is permitted, causing the expression to behave as if
5184 the previous item and the quantifier were not present. This may be use-
5185 ful for subpatterns that are referenced as subroutines from elsewhere
5186 in the pattern (but see also the section entitled "Defining subpatterns
5187 for use by reference only" below). Items other than subpatterns that
5188 have a {0} quantifier are omitted from the compiled pattern.
5190 For convenience, the three most common quantifiers have single-charac-
5191 ter abbreviations:
5193 * is equivalent to {0,}
5194 + is equivalent to {1,}
5195 ? is equivalent to {0,1}
5197 It is possible to construct infinite loops by following a subpattern
5198 that can match no characters with a quantifier that has no upper limit,
5199 for example:
5201 (a?)*
5203 Earlier versions of Perl and PCRE used to give an error at compile time
5204 for such patterns. However, because there are cases where this can be
5205 useful, such patterns are now accepted, but if any repetition of the
5206 subpattern does in fact match no characters, the loop is forcibly bro-
5207 ken.
5209 By default, the quantifiers are "greedy", that is, they match as much
5210 as possible (up to the maximum number of permitted times), without
5211 causing the rest of the pattern to fail. The classic example of where
5212 this gives problems is in trying to match comments in C programs. These
5213 appear between /* and */ and within the comment, individual * and /
5214 characters may appear. An attempt to match C comments by applying the
5215 pattern
5217 /\*.*\*/
5219 to the string
5221 /* first comment */ not comment /* second comment */
5223 fails, because it matches the entire string owing to the greediness of
5224 the .* item.
5226 However, if a quantifier is followed by a question mark, it ceases to
5227 be greedy, and instead matches the minimum number of times possible, so
5228 the pattern
5230 /\*.*?\*/
5232 does the right thing with the C comments. The meaning of the various
5233 quantifiers is not otherwise changed, just the preferred number of
5234 matches. Do not confuse this use of question mark with its use as a
5235 quantifier in its own right. Because it has two uses, it can sometimes
5236 appear doubled, as in
5238 \d??\d
5240 which matches one digit by preference, but can match two if that is the
5241 only way the rest of the pattern matches.
5243 If the PCRE_UNGREEDY option is set (an option that is not available in
5244 Perl), the quantifiers are not greedy by default, but individual ones
5245 can be made greedy by following them with a question mark. In other
5246 words, it inverts the default behaviour.
5248 When a parenthesized subpattern is quantified with a minimum repeat
5249 count that is greater than 1 or with a limited maximum, more memory is
5250 required for the compiled pattern, in proportion to the size of the
5251 minimum or maximum.
5253 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
5254 alent to Perl's /s) is set, thus allowing the dot to match newlines,
5255 the pattern is implicitly anchored, because whatever follows will be
5256 tried against every character position in the subject string, so there
5257 is no point in retrying the overall match at any position after the
5258 first. PCRE normally treats such a pattern as though it were preceded
5259 by \A.
5261 In cases where it is known that the subject string contains no new-
5262 lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
5263 mization, or alternatively using ^ to indicate anchoring explicitly.
5265 However, there is one situation where the optimization cannot be used.
5266 When .* is inside capturing parentheses that are the subject of a back
5267 reference elsewhere in the pattern, a match at the start may fail where
5268 a later one succeeds. Consider, for example:
5270 (.*)abc\1
5272 If the subject is "xyz123abc123" the match point is the fourth charac-
5273 ter. For this reason, such a pattern is not implicitly anchored.
5275 When a capturing subpattern is repeated, the value captured is the sub-
5276 string that matched the final iteration. For example, after
5278 (tweedle[dume]{3}\s*)+
5280 has matched "tweedledum tweedledee" the value of the captured substring
5281 is "tweedledee". However, if there are nested capturing subpatterns,
5282 the corresponding captured values may have been set in previous itera-
5283 tions. For example, after
5285 /(a|(b))+/
5287 matches "aba" the value of the second captured substring is "b".
5292 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
5293 repetition, failure of what follows normally causes the repeated item
5294 to be re-evaluated to see if a different number of repeats allows the
5295 rest of the pattern to match. Sometimes it is useful to prevent this,
5296 either to change the nature of the match, or to cause it fail earlier
5297 than it otherwise might, when the author of the pattern knows there is
5298 no point in carrying on.
5300 Consider, for example, the pattern \d+foo when applied to the subject
5301 line
5303 123456bar
5305 After matching all 6 digits and then failing to match "foo", the normal
5306 action of the matcher is to try again with only 5 digits matching the
5307 \d+ item, and then with 4, and so on, before ultimately failing.
5308 "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
5309 the means for specifying that once a subpattern has matched, it is not
5310 to be re-evaluated in this way.
5312 If we use atomic grouping for the previous example, the matcher gives
5313 up immediately on failing to match "foo" the first time. The notation
5314 is a kind of special parenthesis, starting with (?> as in this example:
5316 (?>\d+)foo
5318 This kind of parenthesis "locks up" the part of the pattern it con-
5319 tains once it has matched, and a failure further into the pattern is
5320 prevented from backtracking into it. Backtracking past it to previous
5321 items, however, works as normal.
5323 An alternative description is that a subpattern of this type matches
5324 the string of characters that an identical standalone pattern would
5325 match, if anchored at the current point in the subject string.
5327 Atomic grouping subpatterns are not capturing subpatterns. Simple cases
5328 such as the above example can be thought of as a maximizing repeat that
5329 must swallow everything it can. So, while both \d+ and \d+? are pre-
5330 pared to adjust the number of digits they match in order to make the
5331 rest of the pattern match, (?>\d+) can only match an entire sequence of
5332 digits.
5334 Atomic groups in general can of course contain arbitrarily complicated
5335 subpatterns, and can be nested. However, when the subpattern for an
5336 atomic group is just a single repeated item, as in the example above, a
5337 simpler notation, called a "possessive quantifier" can be used. This
5338 consists of an additional + character following a quantifier. Using
5339 this notation, the previous example can be rewritten as
5341 \d++foo
5343 Note that a possessive quantifier can be used with an entire group, for
5344 example:
5346 (abc|xyz){2,3}+
5348 Possessive quantifiers are always greedy; the setting of the
5349 PCRE_UNGREEDY option is ignored. They are a convenient notation for the
5350 simpler forms of atomic group. However, there is no difference in the
5351 meaning of a possessive quantifier and the equivalent atomic group,
5352 though there may be a performance difference; possessive quantifiers
5353 should be slightly faster.
5355 The possessive quantifier syntax is an extension to the Perl 5.8 syn-
5356 tax. Jeffrey Friedl originated the idea (and the name) in the first
5357 edition of his book. Mike McCloskey liked it, so implemented it when he
5358 built Sun's Java package, and PCRE copied it from there. It ultimately
5359 found its way into Perl at release 5.10.
5361 PCRE has an optimization that automatically "possessifies" certain sim-
5362 ple pattern constructs. For example, the sequence A+B is treated as
5363 A++B because there is no point in backtracking into a sequence of A's
5364 when B must follow.
5366 When a pattern contains an unlimited repeat inside a subpattern that
5367 can itself be repeated an unlimited number of times, the use of an
5368 atomic group is the only way to avoid some failing matches taking a
5369 very long time indeed. The pattern
5371 (\D+|<\d+>)*[!?]
5373 matches an unlimited number of substrings that either consist of non-
5374 digits, or digits enclosed in <>, followed by either ! or ?. When it
5375 matches, it runs quickly. However, if it is applied to
5377 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
5379 it takes a long time before reporting failure. This is because the
5380 string can be divided between the internal \D+ repeat and the external
5381 * repeat in a large number of ways, and all have to be tried. (The
5382 example uses [!?] rather than a single character at the end, because
5383 both PCRE and Perl have an optimization that allows for fast failure
5384 when a single character is used. They remember the last single charac-
5385 ter that is required for a match, and fail early if it is not present
5386 in the string.) If the pattern is changed so that it uses an atomic
5387 group, like this:
5389 ((?>\D+)|<\d+>)*[!?]
5391 sequences of non-digits cannot be broken, and failure happens quickly.
5396 Outside a character class, a backslash followed by a digit greater than
5397 0 (and possibly further digits) is a back reference to a capturing sub-
5398 pattern earlier (that is, to its left) in the pattern, provided there
5399 have been that many previous capturing left parentheses.
5401 However, if the decimal number following the backslash is less than 10,
5402 it is always taken as a back reference, and causes an error only if
5403 there are not that many capturing left parentheses in the entire pat-
5404 tern. In other words, the parentheses that are referenced need not be
5405 to the left of the reference for numbers less than 10. A "forward back
5406 reference" of this type can make sense when a repetition is involved
5407 and the subpattern to the right has participated in an earlier itera-
5408 tion.
5410 It is not possible to have a numerical "forward back reference" to a
5411 subpattern whose number is 10 or more using this syntax because a
5412 sequence such as \50 is interpreted as a character defined in octal.
5413 See the subsection entitled "Non-printing characters" above for further
5414 details of the handling of digits following a backslash. There is no
5415 such problem when named parentheses are used. A back reference to any
5416 subpattern is possible using named parentheses (see below).
5418 Another way of avoiding the ambiguity inherent in the use of digits
5419 following a backslash is to use the \g escape sequence. This escape
5420 must be followed by an unsigned number or a negative number, optionally
5421 enclosed in braces. These examples are all identical:
5423 (ring), \1
5424 (ring), \g1
5425 (ring), \g{1}
5427 An unsigned number specifies an absolute reference without the ambigu-
5428 ity that is present in the older syntax. It is also useful when literal
5429 digits follow the reference. A negative number is a relative reference.
5430 Consider this example:
5432 (abc(def)ghi)\g{-1}
5434 The sequence \g{-1} is a reference to the most recently started captur-
5435 ing subpattern before \g, that is, is it equivalent to \2 in this exam-
5436 ple. Similarly, \g{-2} would be equivalent to \1. The use of relative
5437 references can be helpful in long patterns, and also in patterns that
5438 are created by joining together fragments that contain references
5439 within themselves.
5441 A back reference matches whatever actually matched the capturing sub-
5442 pattern in the current subject string, rather than anything matching
5443 the subpattern itself (see "Subpatterns as subroutines" below for a way
5444 of doing that). So the pattern
5446 (sens|respons)e and \1ibility
5448 matches "sense and sensibility" and "response and responsibility", but
5449 not "sense and responsibility". If caseful matching is in force at the
5450 time of the back reference, the case of letters is relevant. For exam-
5451 ple,
5453 ((?i)rah)\s+\1
5455 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
5456 original capturing subpattern is matched caselessly.
5458 There are several different ways of writing back references to named
5459 subpatterns. The .NET syntax \k{name} and the Perl syntax \k<name> or
5460 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's
5461 unified back reference syntax, in which \g can be used for both numeric
5462 and named references, is also supported. We could rewrite the above
5463 example in any of the following ways:
5465 (?<p1>(?i)rah)\s+\k<p1>
5466 (?'p1'(?i)rah)\s+\k{p1}
5467 (?P<p1>(?i)rah)\s+(?P=p1)
5468 (?<p1>(?i)rah)\s+\g{p1}
5470 A subpattern that is referenced by name may appear in the pattern
5471 before or after the reference.
5473 There may be more than one back reference to the same subpattern. If a
5474 subpattern has not actually been used in a particular match, any back
5475 references to it always fail by default. For example, the pattern
5477 (a|(bc))\2
5479 always fails if it starts to match "a" rather than "bc". However, if
5480 the PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back refer-
5481 ence to an unset value matches an empty string.
5483 Because there may be many capturing parentheses in a pattern, all dig-
5484 its following a backslash are taken as part of a potential back refer-
5485 ence number. If the pattern continues with a digit character, some
5486 delimiter must be used to terminate the back reference. If the
5487 PCRE_EXTENDED option is set, this can be whitespace. Otherwise, the \g{
5488 syntax or an empty comment (see "Comments" below) can be used.
5490 Recursive back references
5492 A back reference that occurs inside the parentheses to which it refers
5493 fails when the subpattern is first used, so, for example, (a\1) never
5494 matches. However, such references can be useful inside repeated sub-
5495 patterns. For example, the pattern
5497 (a|b\1)+
5499 matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
5500 ation of the subpattern, the back reference matches the character
5501 string corresponding to the previous iteration. In order for this to
5502 work, the pattern must be such that the first iteration does not need
5503 to match the back reference. This can be done using alternation, as in
5504 the example above, or by a quantifier with a minimum of zero.
5506 Back references of this type cause the group that they reference to be
5507 treated as an atomic group. Once the whole group has been matched, a
5508 subsequent matching failure cannot cause backtracking into the middle
5509 of the group.
5514 An assertion is a test on the characters following or preceding the
5515 current matching point that does not actually consume any characters.
5516 The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
5517 described above.
5519 More complicated assertions are coded as subpatterns. There are two
5520 kinds: those that look ahead of the current position in the subject
5521 string, and those that look behind it. An assertion subpattern is
5522 matched in the normal way, except that it does not cause the current
5523 matching position to be changed.
5525 Assertion subpatterns are not capturing subpatterns. If such an asser-
5526 tion contains capturing subpatterns within it, these are counted for
5527 the purposes of numbering the capturing subpatterns in the whole pat-
5528 tern. However, substring capturing is carried out only for positive
5529 assertions, because it does not make sense for negative assertions.
5531 For compatibility with Perl, assertion subpatterns may be repeated;
5532 though it makes no sense to assert the same thing several times, the
5533 side effect of capturing parentheses may occasionally be useful. In
5534 practice, there only three cases:
5536 (1) If the quantifier is {0}, the assertion is never obeyed during
5537 matching. However, it may contain internal capturing parenthesized
5538 groups that are called from elsewhere via the subroutine mechanism.
5540 (2) If quantifier is {0,n} where n is greater than zero, it is treated
5541 as if it were {0,1}. At run time, the rest of the pattern match is
5542 tried with and without the assertion, the order depending on the greed-
5543 iness of the quantifier.
5545 (3) If the minimum repetition is greater than zero, the quantifier is
5546 ignored. The assertion is obeyed just once when encountered during
5547 matching.
5549 Lookahead assertions
5551 Lookahead assertions start with (?= for positive assertions and (?! for
5552 negative assertions. For example,
5554 \w+(?=;)
5556 matches a word followed by a semicolon, but does not include the semi-
5557 colon in the match, and
5559 foo(?!bar)
5561 matches any occurrence of "foo" that is not followed by "bar". Note
5562 that the apparently similar pattern
5564 (?!foo)bar
5566 does not find an occurrence of "bar" that is preceded by something
5567 other than "foo"; it finds any occurrence of "bar" whatsoever, because
5568 the assertion (?!foo) is always true when the next three characters are
5569 "bar". A lookbehind assertion is needed to achieve the other effect.
5571 If you want to force a matching failure at some point in a pattern, the
5572 most convenient way to do it is with (?!) because an empty string
5573 always matches, so an assertion that requires there not to be an empty
5574 string must always fail. The backtracking control verb (*FAIL) or (*F)
5575 is a synonym for (?!).
5577 Lookbehind assertions
5579 Lookbehind assertions start with (?<= for positive assertions and (?<!
5580 for negative assertions. For example,
5582 (?<!foo)bar
5584 does find an occurrence of "bar" that is not preceded by "foo". The
5585 contents of a lookbehind assertion are restricted such that all the
5586 strings it matches must have a fixed length. However, if there are sev-
5587 eral top-level alternatives, they do not all have to have the same
5588 fixed length. Thus
5590 (?<=bullock|donkey)
5592 is permitted, but
5594 (?<!dogs?|cats?)
5596 causes an error at compile time. Branches that match different length
5597 strings are permitted only at the top level of a lookbehind assertion.
5598 This is an extension compared with Perl, which requires all branches to
5599 match the same length of string. An assertion such as
5601 (?<=ab(c|de))
5603 is not permitted, because its single top-level branch can match two
5604 different lengths, but it is acceptable to PCRE if rewritten to use two
5605 top-level branches:
5607 (?<=abc|abde)
5609 In some cases, the escape sequence \K (see above) can be used instead
5610 of a lookbehind assertion to get round the fixed-length restriction.
5612 The implementation of lookbehind assertions is, for each alternative,
5613 to temporarily move the current position back by the fixed length and
5614 then try to match. If there are insufficient characters before the cur-
5615 rent position, the assertion fails.
5617 In a UTF mode, PCRE does not allow the \C escape (which matches a sin-
5618 gle data unit even in a UTF mode) to appear in lookbehind assertions,
5619 because it makes it impossible to calculate the length of the lookbe-
5620 hind. The \X and \R escapes, which can match different numbers of data
5621 units, are also not permitted.
5623 "Subroutine" calls (see below) such as (?2) or (?&X) are permitted in
5624 lookbehinds, as long as the subpattern matches a fixed-length string.
5625 Recursion, however, is not supported.
5627 Possessive quantifiers can be used in conjunction with lookbehind
5628 assertions to specify efficient matching of fixed-length strings at the
5629 end of subject strings. Consider a simple pattern such as
5631 abcd$
5633 when applied to a long string that does not match. Because matching
5634 proceeds from left to right, PCRE will look for each "a" in the subject
5635 and then see if what follows matches the rest of the pattern. If the
5636 pattern is specified as
5638 ^.*abcd$
5640 the initial .* matches the entire string at first, but when this fails
5641 (because there is no following "a"), it backtracks to match all but the
5642 last character, then all but the last two characters, and so on. Once
5643 again the search for "a" covers the entire string, from right to left,
5644 so we are no better off. However, if the pattern is written as
5646 ^.*+(?<=abcd)
5648 there can be no backtracking for the .*+ item; it can match only the
5649 entire string. The subsequent lookbehind assertion does a single test
5650 on the last four characters. If it fails, the match fails immediately.
5651 For long strings, this approach makes a significant difference to the
5652 processing time.
5654 Using multiple assertions
5656 Several assertions (of any sort) may occur in succession. For example,
5658 (?<=\d{3})(?<!999)foo
5660 matches "foo" preceded by three digits that are not "999". Notice that
5661 each of the assertions is applied independently at the same point in
5662 the subject string. First there is a check that the previous three
5663 characters are all digits, and then there is a check that the same
5664 three characters are not "999". This pattern does not match "foo" pre-
5665 ceded by six characters, the first of which are digits and the last
5666 three of which are not "999". For example, it doesn't match "123abc-
5667 foo". A pattern to do that is
5669 (?<=\d{3}...)(?<!999)foo
5671 This time the first assertion looks at the preceding six characters,
5672 checking that the first three are digits, and then the second assertion
5673 checks that the preceding three characters are not "999".
5675 Assertions can be nested in any combination. For example,
5677 (?<=(?<!foo)bar)baz
5679 matches an occurrence of "baz" that is preceded by "bar" which in turn
5680 is not preceded by "foo", while
5682 (?<=\d{3}(?!999)...)foo
5684 is another pattern that matches "foo" preceded by three digits and any
5685 three characters that are not "999".
5690 It is possible to cause the matching process to obey a subpattern con-
5691 ditionally or to choose between two alternative subpatterns, depending
5692 on the result of an assertion, or whether a specific capturing subpat-
5693 tern has already been matched. The two possible forms of conditional
5694 subpattern are:
5696 (?(condition)yes-pattern)
5697 (?(condition)yes-pattern|no-pattern)
5699 If the condition is satisfied, the yes-pattern is used; otherwise the
5700 no-pattern (if present) is used. If there are more than two alterna-
5701 tives in the subpattern, a compile-time error occurs. Each of the two
5702 alternatives may itself contain nested subpatterns of any form, includ-
5703 ing conditional subpatterns; the restriction to two alternatives
5704 applies only at the level of the condition. This pattern fragment is an
5705 example where the alternatives are complex:
5707 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
5710 There are four kinds of condition: references to subpatterns, refer-
5711 ences to recursion, a pseudo-condition called DEFINE, and assertions.
5713 Checking for a used subpattern by number
5715 If the text between the parentheses consists of a sequence of digits,
5716 the condition is true if a capturing subpattern of that number has pre-
5717 viously matched. If there is more than one capturing subpattern with
5718 the same number (see the earlier section about duplicate subpattern
5719 numbers), the condition is true if any of them have matched. An alter-
5720 native notation is to precede the digits with a plus or minus sign. In
5721 this case, the subpattern number is relative rather than absolute. The
5722 most recently opened parentheses can be referenced by (?(-1), the next
5723 most recent by (?(-2), and so on. Inside loops it can also make sense
5724 to refer to subsequent groups. The next parentheses to be opened can be
5725 referenced as (?(+1), and so on. (The value zero in any of these forms
5726 is not used; it provokes a compile-time error.)
5728 Consider the following pattern, which contains non-significant white
5729 space to make it more readable (assume the PCRE_EXTENDED option) and to
5730 divide it into three parts for ease of discussion:
5732 ( \( )? [^()]+ (?(1) \) )
5734 The first part matches an optional opening parenthesis, and if that
5735 character is present, sets it as the first captured substring. The sec-
5736 ond part matches one or more characters that are not parentheses. The
5737 third part is a conditional subpattern that tests whether or not the
5738 first set of parentheses matched. If they did, that is, if subject
5739 started with an opening parenthesis, the condition is true, and so the
5740 yes-pattern is executed and a closing parenthesis is required. Other-
5741 wise, since no-pattern is not present, the subpattern matches nothing.
5742 In other words, this pattern matches a sequence of non-parentheses,
5743 optionally enclosed in parentheses.
5745 If you were embedding this pattern in a larger one, you could use a
5746 relative reference:
5748 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
5750 This makes the fragment independent of the parentheses in the larger
5751 pattern.
5753 Checking for a used subpattern by name
5755 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
5756 used subpattern by name. For compatibility with earlier versions of
5757 PCRE, which had this facility before Perl, the syntax (?(name)...) is
5758 also recognized. However, there is a possible ambiguity with this syn-
5759 tax, because subpattern names may consist entirely of digits. PCRE
5760 looks first for a named subpattern; if it cannot find one and the name
5761 consists entirely of digits, PCRE looks for a subpattern of that num-
5762 ber, which must be greater than zero. Using subpattern names that con-
5763 sist entirely of digits is not recommended.
5765 Rewriting the above example to use a named subpattern gives this:
5767 (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
5769 If the name used in a condition of this kind is a duplicate, the test
5770 is applied to all subpatterns of the same name, and is true if any one
5771 of them has matched.
5773 Checking for pattern recursion
5775 If the condition is the string (R), and there is no subpattern with the
5776 name R, the condition is true if a recursive call to the whole pattern
5777 or any subpattern has been made. If digits or a name preceded by amper-
5778 sand follow the letter R, for example:
5780 (?(R3)...) or (?(R&name)...)
5782 the condition is true if the most recent recursion is into a subpattern
5783 whose number or name is given. This condition does not check the entire
5784 recursion stack. If the name used in a condition of this kind is a
5785 duplicate, the test is applied to all subpatterns of the same name, and
5786 is true if any one of them is the most recent recursion.
5788 At "top level", all these recursion test conditions are false. The
5789 syntax for recursive patterns is described below.
5791 Defining subpatterns for use by reference only
5793 If the condition is the string (DEFINE), and there is no subpattern
5794 with the name DEFINE, the condition is always false. In this case,
5795 there may be only one alternative in the subpattern. It is always
5796 skipped if control reaches this point in the pattern; the idea of
5797 DEFINE is that it can be used to define subroutines that can be refer-
5798 enced from elsewhere. (The use of subroutines is described below.) For
5799 example, a pattern to match an IPv4 address such as ""
5800 could be written like this (ignore whitespace and line breaks):
5802 (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
5803 \b (?&byte) (\.(?&byte)){3} \b
5805 The first part of the pattern is a DEFINE group inside which a another
5806 group named "byte" is defined. This matches an individual component of
5807 an IPv4 address (a number less than 256). When matching takes place,
5808 this part of the pattern is skipped because DEFINE acts like a false
5809 condition. The rest of the pattern uses references to the named group
5810 to match the four dot-separated components of an IPv4 address, insist-
5811 ing on a word boundary at each end.
5813 Assertion conditions
5815 If the condition is not in any of the above formats, it must be an
5816 assertion. This may be a positive or negative lookahead or lookbehind
5817 assertion. Consider this pattern, again containing non-significant
5818 white space, and with the two alternatives on the second line:
5820 (?(?=[^a-z]*[a-z])
5821 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
5823 The condition is a positive lookahead assertion that matches an
5824 optional sequence of non-letters followed by a letter. In other words,
5825 it tests for the presence of at least one letter in the subject. If a
5826 letter is found, the subject is matched against the first alternative;
5827 otherwise it is matched against the second. This pattern matches
5828 strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
5829 letters and dd are digits.
5834 There are two ways of including comments in patterns that are processed
5835 by PCRE. In both cases, the start of the comment must not be in a char-
5836 acter class, nor in the middle of any other sequence of related charac-
5837 ters such as (?: or a subpattern name or number. The characters that
5838 make up a comment play no part in the pattern matching.
5840 The sequence (?# marks the start of a comment that continues up to the
5841 next closing parenthesis. Nested parentheses are not permitted. If the
5842 PCRE_EXTENDED option is set, an unescaped # character also introduces a
5843 comment, which in this case continues to immediately after the next
5844 newline character or character sequence in the pattern. Which charac-
5845 ters are interpreted as newlines is controlled by the options passed to
5846 a compiling function or by a special sequence at the start of the pat-
5847 tern, as described in the section entitled "Newline conventions" above.
5848 Note that the end of this type of comment is a literal newline sequence
5849 in the pattern; escape sequences that happen to represent a newline do
5850 not count. For example, consider this pattern when PCRE_EXTENDED is
5851 set, and the default newline convention is in force:
5853 abc #comment \n still comment
5855 On encountering the # character, pcre_compile() skips along, looking
5856 for a newline in the pattern. The sequence \n is still literal at this
5857 stage, so it does not terminate the comment. Only an actual character
5858 with the code value 0x0a (the default newline) does so.
5863 Consider the problem of matching a string in parentheses, allowing for
5864 unlimited nested parentheses. Without the use of recursion, the best
5865 that can be done is to use a pattern that matches up to some fixed
5866 depth of nesting. It is not possible to handle an arbitrary nesting
5867 depth.
5869 For some time, Perl has provided a facility that allows regular expres-
5870 sions to recurse (amongst other things). It does this by interpolating
5871 Perl code in the expression at run time, and the code can refer to the
5872 expression itself. A Perl pattern using code interpolation to solve the
5873 parentheses problem can be created like this:
5875 $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
5877 The (?p{...}) item interpolates Perl code at run time, and in this case
5878 refers recursively to the pattern in which it appears.
5880 Obviously, PCRE cannot support the interpolation of Perl code. Instead,
5881 it supports special syntax for recursion of the entire pattern, and
5882 also for individual subpattern recursion. After its introduction in
5883 PCRE and Python, this kind of recursion was subsequently introduced
5884 into Perl at release 5.10.
5886 A special item that consists of (? followed by a number greater than
5887 zero and a closing parenthesis is a recursive subroutine call of the
5888 subpattern of the given number, provided that it occurs inside that
5889 subpattern. (If not, it is a non-recursive subroutine call, which is
5890 described in the next section.) The special item (?R) or (?0) is a
5891 recursive call of the entire regular expression.
5893 This PCRE pattern solves the nested parentheses problem (assume the
5894 PCRE_EXTENDED option is set so that white space is ignored):
5896 \( ( [^()]++ | (?R) )* \)
5898 First it matches an opening parenthesis. Then it matches any number of
5899 substrings which can either be a sequence of non-parentheses, or a
5900 recursive match of the pattern itself (that is, a correctly parenthe-
5901 sized substring). Finally there is a closing parenthesis. Note the use
5902 of a possessive quantifier to avoid backtracking into sequences of non-
5903 parentheses.
5905 If this were part of a larger pattern, you would not want to recurse
5906 the entire pattern, so instead you could use this:
5908 ( \( ( [^()]++ | (?1) )* \) )
5910 We have put the pattern into parentheses, and caused the recursion to
5911 refer to them instead of the whole pattern.
5913 In a larger pattern, keeping track of parenthesis numbers can be
5914 tricky. This is made easier by the use of relative references. Instead
5915 of (?1) in the pattern above you can write (?-2) to refer to the second
5916 most recently opened parentheses preceding the recursion. In other
5917 words, a negative number counts capturing parentheses leftwards from
5918 the point at which it is encountered.
5920 It is also possible to refer to subsequently opened parentheses, by
5921 writing references such as (?+2). However, these cannot be recursive
5922 because the reference is not inside the parentheses that are refer-
5923 enced. They are always non-recursive subroutine calls, as described in
5924 the next section.
5926 An alternative approach is to use named parentheses instead. The Perl
5927 syntax for this is (?&name); PCRE's earlier syntax (?P>name) is also
5928 supported. We could rewrite the above example as follows:
5930 (?<pn> \( ( [^()]++ | (?&pn) )* \) )
5932 If there is more than one subpattern with the same name, the earliest
5933 one is used.
5935 This particular example pattern that we have been looking at contains
5936 nested unlimited repeats, and so the use of a possessive quantifier for
5937 matching strings of non-parentheses is important when applying the pat-
5938 tern to strings that do not match. For example, when this pattern is
5939 applied to
5941 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
5943 it yields "no match" quickly. However, if a possessive quantifier is
5944 not used, the match runs for a very long time indeed because there are
5945 so many different ways the + and * repeats can carve up the subject,
5946 and all have to be tested before failure can be reported.
5948 At the end of a match, the values of capturing parentheses are those
5949 from the outermost level. If you want to obtain intermediate values, a
5950 callout function can be used (see below and the pcrecallout documenta-
5951 tion). If the pattern above is matched against
5953 (ab(cd)ef)
5955 the value for the inner capturing parentheses (numbered 2) is "ef",
5956 which is the last value taken on at the top level. If a capturing sub-
5957 pattern is not matched at the top level, its final captured value is
5958 unset, even if it was (temporarily) set at a deeper level during the
5959 matching process.
5961 If there are more than 15 capturing parentheses in a pattern, PCRE has
5962 to obtain extra memory to store data during a recursion, which it does
5963 by using pcre_malloc, freeing it via pcre_free afterwards. If no memory
5964 can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
5966 Do not confuse the (?R) item with the condition (R), which tests for
5967 recursion. Consider this pattern, which matches text in angle brack-
5968 ets, allowing for arbitrary nesting. Only digits are allowed in nested
5969 brackets (that is, when recursing), whereas any characters are permit-
5970 ted at the outer level.
5972 < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
5974 In this pattern, (?(R) is the start of a conditional subpattern, with
5975 two different alternatives for the recursive and non-recursive cases.
5976 The (?R) item is the actual recursive call.
5978 Differences in recursion processing between PCRE and Perl
5980 Recursion processing in PCRE differs from Perl in two important ways.
5981 In PCRE (like Python, but unlike Perl), a recursive subpattern call is
5982 always treated as an atomic group. That is, once it has matched some of
5983 the subject string, it is never re-entered, even if it contains untried
5984 alternatives and there is a subsequent matching failure. This can be
5985 illustrated by the following pattern, which purports to match a palin-
5986 dromic string that contains an odd number of characters (for example,
5987 "a", "aba", "abcba", "abcdcba"):
5989 ^(.|(.)(?1)\2)$
5991 The idea is that it either matches a single character, or two identical
5992 characters surrounding a sub-palindrome. In Perl, this pattern works;
5993 in PCRE it does not if the pattern is longer than three characters.
5994 Consider the subject string "abcba":
5996 At the top level, the first character is matched, but as it is not at
5997 the end of the string, the first alternative fails; the second alterna-
5998 tive is taken and the recursion kicks in. The recursive call to subpat-
5999 tern 1 successfully matches the next character ("b"). (Note that the
6000 beginning and end of line tests are not part of the recursion).
6002 Back at the top level, the next character ("c") is compared with what
6003 subpattern 2 matched, which was "a". This fails. Because the recursion
6004 is treated as an atomic group, there are now no backtracking points,
6005 and so the entire match fails. (Perl is able, at this point, to re-
6006 enter the recursion and try the second alternative.) However, if the
6007 pattern is written with the alternatives in the other order, things are
6008 different:
6010 ^((.)(?1)\2|.)$
6012 This time, the recursing alternative is tried first, and continues to
6013 recurse until it runs out of characters, at which point the recursion
6014 fails. But this time we do have another alternative to try at the
6015 higher level. That is the big difference: in the previous case the
6016 remaining alternative is at a deeper recursion level, which PCRE cannot