1: -----------------------------------------------------------------------------
2: This file contains a concatenation of the PCRE man pages, converted to plain
3: text format for ease of searching with a text editor, or for use on systems
4: that do not have a man page processor. The small individual files that give
5: synopses of each function in the library have not been included. Neither has
6: the pcredemo program. There are separate text files for the pcregrep and
7: pcretest commands.
8: -----------------------------------------------------------------------------
9:
10:
11: PCRE(3) Library Functions Manual PCRE(3)
12:
13:
14:
15: NAME
16: PCRE - Perl-compatible regular expressions
17:
18: INTRODUCTION
19:
20: The PCRE library is a set of functions that implement regular expres-
21: sion pattern matching using the same syntax and semantics as Perl, with
22: just a few differences. Some features that appeared in Python and PCRE
23: before they appeared in Perl are also available using the Python syn-
24: tax, there is some support for one or two .NET and Oniguruma syntax
25: items, and there is an option for requesting some minor changes that
26: give better JavaScript compatibility.
27:
28: 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.
34:
35: Starting with release 8.32 it is possible to compile a third separate
36: PCRE library that supports 32-bit character strings (including UTF-32
37: strings). The build process allows any combination of the 8-, 16- and
38: 32-bit libraries. The work to make this possible was done by Christian
39: Persch.
40:
41: The three libraries contain identical sets of functions, except that
42: the names in the 16-bit library start with pcre16_ instead of pcre_,
43: and the names in the 32-bit library start with pcre32_ instead of
44: pcre_. To avoid over-complication and reduce the documentation mainte-
45: nance load, most of the documentation describes the 8-bit library, with
46: the differences for the 16-bit and 32-bit libraries described sepa-
47: rately in the pcre16 and pcre32 pages. References to functions or
48: structures of the form pcre[16|32]_xxx should be read as meaning
49: "pcre_xxx when using the 8-bit library, pcre16_xxx when using the
50: 16-bit library, or pcre32_xxx when using the 32-bit library".
51:
52: The current implementation of PCRE corresponds approximately with Perl
53: 5.12, including support for UTF-8/16/32 encoded strings and Unicode
54: general category properties. However, UTF-8/16/32 and Unicode support
55: has to be explicitly enabled; it is not the default. The Unicode tables
56: correspond to Unicode release 6.3.0.
57:
58: In addition to the Perl-compatible matching function, PCRE contains an
59: alternative function that matches the same compiled patterns in a dif-
60: ferent way. In certain circumstances, the alternative function has some
61: advantages. For a discussion of the two matching algorithms, see the
62: pcrematching page.
63:
64: PCRE is written in C and released as a C library. A number of people
65: have written wrappers and interfaces of various kinds. In particular,
66: Google Inc. have provided a comprehensive C++ wrapper for the 8-bit
67: library. This is now included as part of the PCRE distribution. The
68: pcrecpp page has details of this interface. Other people's contribu-
69: tions can be found in the Contrib directory at the primary FTP site,
70: which is:
71:
72: ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre
73:
74: Details of exactly which Perl regular expression features are and are
75: not supported by PCRE are given in separate documents. See the pcrepat-
76: tern and pcrecompat pages. There is a syntax summary in the pcresyntax
77: page.
78:
79: Some features of PCRE can be included, excluded, or changed when the
80: library is built. The pcre_config() function makes it possible for a
81: client to discover which features are available. The features them-
82: selves are described in the pcrebuild page. Documentation about build-
83: ing PCRE for various operating systems can be found in the README and
84: NON-AUTOTOOLS_BUILD files in the source distribution.
85:
86: The libraries contains a number of undocumented internal functions and
87: data tables that are used by more than one of the exported external
88: functions, but which are not intended for use by external callers.
89: Their names all begin with "_pcre_" or "_pcre16_" or "_pcre32_", which
90: hopefully will not provoke any name clashes. In some environments, it
91: is possible to control which external symbols are exported when a
92: shared library is built, and in these cases the undocumented symbols
93: are not exported.
94:
95:
96: SECURITY CONSIDERATIONS
97:
98: If you are using PCRE in a non-UTF application that permits users to
99: supply arbitrary patterns for compilation, you should be aware of a
100: feature that allows users to turn on UTF support from within a pattern,
101: provided that PCRE was built with UTF support. For example, an 8-bit
102: pattern that begins with "(*UTF8)" or "(*UTF)" turns on UTF-8 mode,
103: which interprets patterns and subjects as strings of UTF-8 characters
104: instead of individual 8-bit characters. This causes both the pattern
105: and any data against which it is matched to be checked for UTF-8 valid-
106: ity. If the data string is very long, such a check might use suffi-
107: ciently many resources as to cause your application to lose perfor-
108: mance.
109:
110: One way of guarding against this possibility is to use the
111: pcre_fullinfo() function to check the compiled pattern's options for
112: UTF. Alternatively, from release 8.33, you can set the PCRE_NEVER_UTF
113: option at compile time. This causes an compile time error if a pattern
114: contains a UTF-setting sequence.
115:
116: If your application is one that supports UTF, be aware that validity
117: checking can take time. If the same data string is to be matched many
118: times, you can use the PCRE_NO_UTF[8|16|32]_CHECK option for the second
119: and subsequent matches to save redundant checks.
120:
121: Another way that performance can be hit is by running a pattern that
122: has a very large search tree against a string that will never match.
123: Nested unlimited repeats in a pattern are a common example. PCRE pro-
124: vides some protection against this: see the PCRE_EXTRA_MATCH_LIMIT fea-
125: ture in the pcreapi page.
126:
127:
128: USER DOCUMENTATION
129:
130: The user documentation for PCRE comprises a number of different sec-
131: tions. In the "man" format, each of these is a separate "man page". In
132: the HTML format, each is a separate page, linked from the index page.
133: In the plain text format, all the sections, except the pcredemo sec-
134: tion, are concatenated, for ease of searching. The sections are as fol-
135: lows:
136:
137: pcre this document
138: pcre-config show PCRE installation configuration information
139: pcre16 details of the 16-bit library
140: pcre32 details of the 32-bit library
141: pcreapi details of PCRE's native C API
142: pcrebuild building PCRE
143: pcrecallout details of the callout feature
144: pcrecompat discussion of Perl compatibility
145: pcrecpp details of the C++ wrapper for the 8-bit library
146: pcredemo a demonstration C program that uses PCRE
147: pcregrep description of the pcregrep command (8-bit only)
148: pcrejit discussion of the just-in-time optimization support
149: pcrelimits details of size and other limits
150: pcrematching discussion of the two matching algorithms
151: pcrepartial details of the partial matching facility
152: pcrepattern syntax and semantics of supported
153: regular expressions
154: pcreperform discussion of performance issues
155: pcreposix the POSIX-compatible C API for the 8-bit library
156: pcreprecompile details of saving and re-using precompiled patterns
157: pcresample discussion of the pcredemo program
158: pcrestack discussion of stack usage
159: pcresyntax quick syntax reference
160: pcretest description of the pcretest testing command
161: pcreunicode discussion of Unicode and UTF-8/16/32 support
162:
163: In addition, in the "man" and HTML formats, there is a short page for
164: each C library function, listing its arguments and results.
165:
166:
167: AUTHOR
168:
169: Philip Hazel
170: University Computing Service
171: Cambridge CB2 3QH, England.
172:
173: Putting an actual email address here seems to have been a spam magnet,
174: so I've taken it away. If you want to email me, use my two initials,
175: followed by the two digits 10, at the domain cam.ac.uk.
176:
177:
178: REVISION
179:
180: Last updated: 13 May 2013
181: Copyright (c) 1997-2013 University of Cambridge.
182: ------------------------------------------------------------------------------
183:
184:
185: PCRE(3) Library Functions Manual PCRE(3)
186:
187:
188:
189: NAME
190: PCRE - Perl-compatible regular expressions
191:
192: #include <pcre.h>
193:
194:
195: PCRE 16-BIT API BASIC FUNCTIONS
196:
197: pcre16 *pcre16_compile(PCRE_SPTR16 pattern, int options,
198: const char **errptr, int *erroffset,
199: const unsigned char *tableptr);
200:
201: pcre16 *pcre16_compile2(PCRE_SPTR16 pattern, int options,
202: int *errorcodeptr,
203: const char **errptr, int *erroffset,
204: const unsigned char *tableptr);
205:
206: pcre16_extra *pcre16_study(const pcre16 *code, int options,
207: const char **errptr);
208:
209: void pcre16_free_study(pcre16_extra *extra);
210:
211: int pcre16_exec(const pcre16 *code, const pcre16_extra *extra,
212: PCRE_SPTR16 subject, int length, int startoffset,
213: int options, int *ovector, int ovecsize);
214:
215: int pcre16_dfa_exec(const pcre16 *code, const pcre16_extra *extra,
216: PCRE_SPTR16 subject, int length, int startoffset,
217: int options, int *ovector, int ovecsize,
218: int *workspace, int wscount);
219:
220:
221: PCRE 16-BIT API STRING EXTRACTION FUNCTIONS
222:
223: int pcre16_copy_named_substring(const pcre16 *code,
224: PCRE_SPTR16 subject, int *ovector,
225: int stringcount, PCRE_SPTR16 stringname,
226: PCRE_UCHAR16 *buffer, int buffersize);
227:
228: int pcre16_copy_substring(PCRE_SPTR16 subject, int *ovector,
229: int stringcount, int stringnumber, PCRE_UCHAR16 *buffer,
230: int buffersize);
231:
232: int pcre16_get_named_substring(const pcre16 *code,
233: PCRE_SPTR16 subject, int *ovector,
234: int stringcount, PCRE_SPTR16 stringname,
235: PCRE_SPTR16 *stringptr);
236:
237: int pcre16_get_stringnumber(const pcre16 *code,
238: PCRE_SPTR16 name);
239:
240: int pcre16_get_stringtable_entries(const pcre16 *code,
241: PCRE_SPTR16 name, PCRE_UCHAR16 **first, PCRE_UCHAR16 **last);
242:
243: int pcre16_get_substring(PCRE_SPTR16 subject, int *ovector,
244: int stringcount, int stringnumber,
245: PCRE_SPTR16 *stringptr);
246:
247: int pcre16_get_substring_list(PCRE_SPTR16 subject,
248: int *ovector, int stringcount, PCRE_SPTR16 **listptr);
249:
250: void pcre16_free_substring(PCRE_SPTR16 stringptr);
251:
252: void pcre16_free_substring_list(PCRE_SPTR16 *stringptr);
253:
254:
255: PCRE 16-BIT API AUXILIARY FUNCTIONS
256:
257: pcre16_jit_stack *pcre16_jit_stack_alloc(int startsize, int maxsize);
258:
259: void pcre16_jit_stack_free(pcre16_jit_stack *stack);
260:
261: void pcre16_assign_jit_stack(pcre16_extra *extra,
262: pcre16_jit_callback callback, void *data);
263:
264: const unsigned char *pcre16_maketables(void);
265:
266: int pcre16_fullinfo(const pcre16 *code, const pcre16_extra *extra,
267: int what, void *where);
268:
269: int pcre16_refcount(pcre16 *code, int adjust);
270:
271: int pcre16_config(int what, void *where);
272:
273: const char *pcre16_version(void);
274:
275: int pcre16_pattern_to_host_byte_order(pcre16 *code,
276: pcre16_extra *extra, const unsigned char *tables);
277:
278:
279: PCRE 16-BIT API INDIRECTED FUNCTIONS
280:
281: void *(*pcre16_malloc)(size_t);
282:
283: void (*pcre16_free)(void *);
284:
285: void *(*pcre16_stack_malloc)(size_t);
286:
287: void (*pcre16_stack_free)(void *);
288:
289: int (*pcre16_callout)(pcre16_callout_block *);
290:
291:
292: PCRE 16-BIT API 16-BIT-ONLY FUNCTION
293:
294: int pcre16_utf16_to_host_byte_order(PCRE_UCHAR16 *output,
295: PCRE_SPTR16 input, int length, int *byte_order,
296: int keep_boms);
297:
298:
299: THE PCRE 16-BIT LIBRARY
300:
301: Starting with release 8.30, it is possible to compile a PCRE library
302: that supports 16-bit character strings, including UTF-16 strings, as
303: well as or instead of the original 8-bit library. The majority of the
304: work to make this possible was done by Zoltan Herczeg. The two
305: libraries contain identical sets of functions, used in exactly the same
306: way. Only the names of the functions and the data types of their argu-
307: ments and results are different. To avoid over-complication and reduce
308: the documentation maintenance load, most of the PCRE documentation
309: describes the 8-bit library, with only occasional references to the
310: 16-bit library. This page describes what is different when you use the
311: 16-bit library.
312:
313: WARNING: A single application can be linked with both libraries, but
314: you must take care when processing any particular pattern to use func-
315: tions from just one library. For example, if you want to study a pat-
316: tern that was compiled with pcre16_compile(), you must do so with
317: pcre16_study(), not pcre_study(), and you must free the study data with
318: pcre16_free_study().
319:
320:
321: THE HEADER FILE
322:
323: There is only one header file, pcre.h. It contains prototypes for all
324: the functions in all libraries, as well as definitions of flags, struc-
325: tures, error codes, etc.
326:
327:
328: THE LIBRARY NAME
329:
330: In Unix-like systems, the 16-bit library is called libpcre16, and can
331: normally be accesss by adding -lpcre16 to the command for linking an
332: application that uses PCRE.
333:
334:
335: STRING TYPES
336:
337: In the 8-bit library, strings are passed to PCRE library functions as
338: vectors of bytes with the C type "char *". In the 16-bit library,
339: strings are passed as vectors of unsigned 16-bit quantities. The macro
340: PCRE_UCHAR16 specifies an appropriate data type, and PCRE_SPTR16 is
341: defined as "const PCRE_UCHAR16 *". In very many environments, "short
342: int" is a 16-bit data type. When PCRE is built, it defines PCRE_UCHAR16
343: as "unsigned short int", but checks that it really is a 16-bit data
344: type. If it is not, the build fails with an error message telling the
345: maintainer to modify the definition appropriately.
346:
347:
348: STRUCTURE TYPES
349:
350: The types of the opaque structures that are used for compiled 16-bit
351: patterns and JIT stacks are pcre16 and pcre16_jit_stack respectively.
352: The type of the user-accessible structure that is returned by
353: pcre16_study() is pcre16_extra, and the type of the structure that is
354: used for passing data to a callout function is pcre16_callout_block.
355: These structures contain the same fields, with the same names, as their
356: 8-bit counterparts. The only difference is that pointers to character
357: strings are 16-bit instead of 8-bit types.
358:
359:
360: 16-BIT FUNCTIONS
361:
362: For every function in the 8-bit library there is a corresponding func-
363: tion in the 16-bit library with a name that starts with pcre16_ instead
364: of pcre_. The prototypes are listed above. In addition, there is one
365: extra function, pcre16_utf16_to_host_byte_order(). This is a utility
366: function that converts a UTF-16 character string to host byte order if
367: necessary. The other 16-bit functions expect the strings they are
368: passed to be in host byte order.
369:
370: The input and output arguments of pcre16_utf16_to_host_byte_order() may
371: point to the same address, that is, conversion in place is supported.
372: The output buffer must be at least as long as the input.
373:
374: The length argument specifies the number of 16-bit data units in the
375: input string; a negative value specifies a zero-terminated string.
376:
377: If byte_order is NULL, it is assumed that the string starts off in host
378: byte order. This may be changed by byte-order marks (BOMs) anywhere in
379: the string (commonly as the first character).
380:
381: If byte_order is not NULL, a non-zero value of the integer to which it
382: points means that the input starts off in host byte order, otherwise
383: the opposite order is assumed. Again, BOMs in the string can change
384: this. The final byte order is passed back at the end of processing.
385:
386: If keep_boms is not zero, byte-order mark characters (0xfeff) are
387: copied into the output string. Otherwise they are discarded.
388:
389: The result of the function is the number of 16-bit units placed into
390: the output buffer, including the zero terminator if the string was
391: zero-terminated.
392:
393:
394: SUBJECT STRING OFFSETS
395:
396: The lengths and starting offsets of subject strings must be specified
397: in 16-bit data units, and the offsets within subject strings that are
398: returned by the matching functions are in also 16-bit units rather than
399: bytes.
400:
401:
402: NAMED SUBPATTERNS
403:
404: The name-to-number translation table that is maintained for named sub-
405: patterns uses 16-bit characters. The pcre16_get_stringtable_entries()
406: function returns the length of each entry in the table as the number of
407: 16-bit data units.
408:
409:
410: OPTION NAMES
411:
412: There are two new general option names, PCRE_UTF16 and
413: PCRE_NO_UTF16_CHECK, which correspond to PCRE_UTF8 and
414: PCRE_NO_UTF8_CHECK in the 8-bit library. In fact, these new options
415: define the same bits in the options word. There is a discussion about
416: the validity of UTF-16 strings in the pcreunicode page.
417:
418: For the pcre16_config() function there is an option PCRE_CONFIG_UTF16
419: that returns 1 if UTF-16 support is configured, otherwise 0. If this
420: option is given to pcre_config() or pcre32_config(), or if the
421: PCRE_CONFIG_UTF8 or PCRE_CONFIG_UTF32 option is given to pcre16_con-
422: fig(), the result is the PCRE_ERROR_BADOPTION error.
423:
424:
425: CHARACTER CODES
426:
427: In 16-bit mode, when PCRE_UTF16 is not set, character values are
428: treated in the same way as in 8-bit, non UTF-8 mode, except, of course,
429: that they can range from 0 to 0xffff instead of 0 to 0xff. Character
430: types for characters less than 0xff can therefore be influenced by the
431: locale in the same way as before. Characters greater than 0xff have
432: only one case, and no "type" (such as letter or digit).
433:
434: In UTF-16 mode, the character code is Unicode, in the range 0 to
435: 0x10ffff, with the exception of values in the range 0xd800 to 0xdfff
436: because those are "surrogate" values that are used in pairs to encode
437: values greater than 0xffff.
438:
439: A UTF-16 string can indicate its endianness by special code knows as a
440: byte-order mark (BOM). The PCRE functions do not handle this, expecting
441: strings to be in host byte order. A utility function called
442: pcre16_utf16_to_host_byte_order() is provided to help with this (see
443: above).
444:
445:
446: ERROR NAMES
447:
448: The errors PCRE_ERROR_BADUTF16_OFFSET and PCRE_ERROR_SHORTUTF16 corre-
449: spond to their 8-bit counterparts. The error PCRE_ERROR_BADMODE is
450: given when a compiled pattern is passed to a function that processes
451: patterns in the other mode, for example, if a pattern compiled with
452: pcre_compile() is passed to pcre16_exec().
453:
454: There are new error codes whose names begin with PCRE_UTF16_ERR for
455: invalid UTF-16 strings, corresponding to the PCRE_UTF8_ERR codes for
456: UTF-8 strings that are described in the section entitled "Reason codes
457: for invalid UTF-8 strings" in the main pcreapi page. The UTF-16 errors
458: are:
459:
460: PCRE_UTF16_ERR1 Missing low surrogate at end of string
461: PCRE_UTF16_ERR2 Invalid low surrogate follows high surrogate
462: PCRE_UTF16_ERR3 Isolated low surrogate
463: PCRE_UTF16_ERR4 Non-character
464:
465:
466: ERROR TEXTS
467:
468: If there is an error while compiling a pattern, the error text that is
469: passed back by pcre16_compile() or pcre16_compile2() is still an 8-bit
470: character string, zero-terminated.
471:
472:
473: CALLOUTS
474:
475: The subject and mark fields in the callout block that is passed to a
476: callout function point to 16-bit vectors.
477:
478:
479: TESTING
480:
481: The pcretest program continues to operate with 8-bit input and output
482: files, but it can be used for testing the 16-bit library. If it is run
483: with the command line option -16, patterns and subject strings are con-
484: verted from 8-bit to 16-bit before being passed to PCRE, and the 16-bit
485: library functions are used instead of the 8-bit ones. Returned 16-bit
486: strings are converted to 8-bit for output. If both the 8-bit and the
487: 32-bit libraries were not compiled, pcretest defaults to 16-bit and the
488: -16 option is ignored.
489:
490: When PCRE is being built, the RunTest script that is called by "make
491: check" uses the pcretest -C option to discover which of the 8-bit,
492: 16-bit and 32-bit libraries has been built, and runs the tests appro-
493: priately.
494:
495:
496: NOT SUPPORTED IN 16-BIT MODE
497:
498: Not all the features of the 8-bit library are available with the 16-bit
499: library. The C++ and POSIX wrapper functions support only the 8-bit
500: library, and the pcregrep program is at present 8-bit only.
501:
502:
503: AUTHOR
504:
505: Philip Hazel
506: University Computing Service
507: Cambridge CB2 3QH, England.
508:
509:
510: REVISION
511:
512: Last updated: 12 May 2013
513: Copyright (c) 1997-2013 University of Cambridge.
514: ------------------------------------------------------------------------------
515:
516:
517: PCRE(3) Library Functions Manual PCRE(3)
518:
519:
520:
521: NAME
522: PCRE - Perl-compatible regular expressions
523:
524: #include <pcre.h>
525:
526:
527: PCRE 32-BIT API BASIC FUNCTIONS
528:
529: pcre32 *pcre32_compile(PCRE_SPTR32 pattern, int options,
530: const char **errptr, int *erroffset,
531: const unsigned char *tableptr);
532:
533: pcre32 *pcre32_compile2(PCRE_SPTR32 pattern, int options,
534: int *errorcodeptr,
535: const unsigned char *tableptr);
536:
537: pcre32_extra *pcre32_study(const pcre32 *code, int options,
538: const char **errptr);
539:
540: void pcre32_free_study(pcre32_extra *extra);
541:
542: int pcre32_exec(const pcre32 *code, const pcre32_extra *extra,
543: PCRE_SPTR32 subject, int length, int startoffset,
544: int options, int *ovector, int ovecsize);
545:
546: int pcre32_dfa_exec(const pcre32 *code, const pcre32_extra *extra,
547: PCRE_SPTR32 subject, int length, int startoffset,
548: int options, int *ovector, int ovecsize,
549: int *workspace, int wscount);
550:
551:
552: PCRE 32-BIT API STRING EXTRACTION FUNCTIONS
553:
554: int pcre32_copy_named_substring(const pcre32 *code,
555: PCRE_SPTR32 subject, int *ovector,
556: int stringcount, PCRE_SPTR32 stringname,
557: PCRE_UCHAR32 *buffer, int buffersize);
558:
559: int pcre32_copy_substring(PCRE_SPTR32 subject, int *ovector,
560: int stringcount, int stringnumber, PCRE_UCHAR32 *buffer,
561: int buffersize);
562:
563: int pcre32_get_named_substring(const pcre32 *code,
564: PCRE_SPTR32 subject, int *ovector,
565: int stringcount, PCRE_SPTR32 stringname,
566: PCRE_SPTR32 *stringptr);
567:
568: int pcre32_get_stringnumber(const pcre32 *code,
569: PCRE_SPTR32 name);
570:
571: int pcre32_get_stringtable_entries(const pcre32 *code,
572: PCRE_SPTR32 name, PCRE_UCHAR32 **first, PCRE_UCHAR32 **last);
573:
574: int pcre32_get_substring(PCRE_SPTR32 subject, int *ovector,
575: int stringcount, int stringnumber,
576: PCRE_SPTR32 *stringptr);
577:
578: int pcre32_get_substring_list(PCRE_SPTR32 subject,
579: int *ovector, int stringcount, PCRE_SPTR32 **listptr);
580:
581: void pcre32_free_substring(PCRE_SPTR32 stringptr);
582:
583: void pcre32_free_substring_list(PCRE_SPTR32 *stringptr);
584:
585:
586: PCRE 32-BIT API AUXILIARY FUNCTIONS
587:
588: pcre32_jit_stack *pcre32_jit_stack_alloc(int startsize, int maxsize);
589:
590: void pcre32_jit_stack_free(pcre32_jit_stack *stack);
591:
592: void pcre32_assign_jit_stack(pcre32_extra *extra,
593: pcre32_jit_callback callback, void *data);
594:
595: const unsigned char *pcre32_maketables(void);
596:
597: int pcre32_fullinfo(const pcre32 *code, const pcre32_extra *extra,
598: int what, void *where);
599:
600: int pcre32_refcount(pcre32 *code, int adjust);
601:
602: int pcre32_config(int what, void *where);
603:
604: const char *pcre32_version(void);
605:
606: int pcre32_pattern_to_host_byte_order(pcre32 *code,
607: pcre32_extra *extra, const unsigned char *tables);
608:
609:
610: PCRE 32-BIT API INDIRECTED FUNCTIONS
611:
612: void *(*pcre32_malloc)(size_t);
613:
614: void (*pcre32_free)(void *);
615:
616: void *(*pcre32_stack_malloc)(size_t);
617:
618: void (*pcre32_stack_free)(void *);
619:
620: int (*pcre32_callout)(pcre32_callout_block *);
621:
622:
623: PCRE 32-BIT API 32-BIT-ONLY FUNCTION
624:
625: int pcre32_utf32_to_host_byte_order(PCRE_UCHAR32 *output,
626: PCRE_SPTR32 input, int length, int *byte_order,
627: int keep_boms);
628:
629:
630: THE PCRE 32-BIT LIBRARY
631:
632: Starting with release 8.32, it is possible to compile a PCRE library
633: that supports 32-bit character strings, including UTF-32 strings, as
634: well as or instead of the original 8-bit library. This work was done by
635: Christian Persch, based on the work done by Zoltan Herczeg for the
636: 16-bit library. All three libraries contain identical sets of func-
637: tions, used in exactly the same way. Only the names of the functions
638: and the data types of their arguments and results are different. To
639: avoid over-complication and reduce the documentation maintenance load,
640: most of the PCRE documentation describes the 8-bit library, with only
641: occasional references to the 16-bit and 32-bit libraries. This page
642: describes what is different when you use the 32-bit library.
643:
644: WARNING: A single application can be linked with all or any of the
645: three libraries, but you must take care when processing any particular
646: pattern to use functions from just one library. For example, if you
647: want to study a pattern that was compiled with pcre32_compile(), you
648: must do so with pcre32_study(), not pcre_study(), and you must free the
649: study data with pcre32_free_study().
650:
651:
652: THE HEADER FILE
653:
654: There is only one header file, pcre.h. It contains prototypes for all
655: the functions in all libraries, as well as definitions of flags, struc-
656: tures, error codes, etc.
657:
658:
659: THE LIBRARY NAME
660:
661: In Unix-like systems, the 32-bit library is called libpcre32, and can
662: normally be accesss by adding -lpcre32 to the command for linking an
663: application that uses PCRE.
664:
665:
666: STRING TYPES
667:
668: In the 8-bit library, strings are passed to PCRE library functions as
669: vectors of bytes with the C type "char *". In the 32-bit library,
670: strings are passed as vectors of unsigned 32-bit quantities. The macro
671: PCRE_UCHAR32 specifies an appropriate data type, and PCRE_SPTR32 is
672: defined as "const PCRE_UCHAR32 *". In very many environments, "unsigned
673: int" is a 32-bit data type. When PCRE is built, it defines PCRE_UCHAR32
674: as "unsigned int", but checks that it really is a 32-bit data type. If
675: it is not, the build fails with an error message telling the maintainer
676: to modify the definition appropriately.
677:
678:
679: STRUCTURE TYPES
680:
681: The types of the opaque structures that are used for compiled 32-bit
682: patterns and JIT stacks are pcre32 and pcre32_jit_stack respectively.
683: The type of the user-accessible structure that is returned by
684: pcre32_study() is pcre32_extra, and the type of the structure that is
685: used for passing data to a callout function is pcre32_callout_block.
686: These structures contain the same fields, with the same names, as their
687: 8-bit counterparts. The only difference is that pointers to character
688: strings are 32-bit instead of 8-bit types.
689:
690:
691: 32-BIT FUNCTIONS
692:
693: For every function in the 8-bit library there is a corresponding func-
694: tion in the 32-bit library with a name that starts with pcre32_ instead
695: of pcre_. The prototypes are listed above. In addition, there is one
696: extra function, pcre32_utf32_to_host_byte_order(). This is a utility
697: function that converts a UTF-32 character string to host byte order if
698: necessary. The other 32-bit functions expect the strings they are
699: passed to be in host byte order.
700:
701: The input and output arguments of pcre32_utf32_to_host_byte_order() may
702: point to the same address, that is, conversion in place is supported.
703: The output buffer must be at least as long as the input.
704:
705: The length argument specifies the number of 32-bit data units in the
706: input string; a negative value specifies a zero-terminated string.
707:
708: If byte_order is NULL, it is assumed that the string starts off in host
709: byte order. This may be changed by byte-order marks (BOMs) anywhere in
710: the string (commonly as the first character).
711:
712: If byte_order is not NULL, a non-zero value of the integer to which it
713: points means that the input starts off in host byte order, otherwise
714: the opposite order is assumed. Again, BOMs in the string can change
715: this. The final byte order is passed back at the end of processing.
716:
717: If keep_boms is not zero, byte-order mark characters (0xfeff) are
718: copied into the output string. Otherwise they are discarded.
719:
720: The result of the function is the number of 32-bit units placed into
721: the output buffer, including the zero terminator if the string was
722: zero-terminated.
723:
724:
725: SUBJECT STRING OFFSETS
726:
727: The lengths and starting offsets of subject strings must be specified
728: in 32-bit data units, and the offsets within subject strings that are
729: returned by the matching functions are in also 32-bit units rather than
730: bytes.
731:
732:
733: NAMED SUBPATTERNS
734:
735: The name-to-number translation table that is maintained for named sub-
736: patterns uses 32-bit characters. The pcre32_get_stringtable_entries()
737: function returns the length of each entry in the table as the number of
738: 32-bit data units.
739:
740:
741: OPTION NAMES
742:
743: There are two new general option names, PCRE_UTF32 and
744: PCRE_NO_UTF32_CHECK, which correspond to PCRE_UTF8 and
745: PCRE_NO_UTF8_CHECK in the 8-bit library. In fact, these new options
746: define the same bits in the options word. There is a discussion about
747: the validity of UTF-32 strings in the pcreunicode page.
748:
749: For the pcre32_config() function there is an option PCRE_CONFIG_UTF32
750: that returns 1 if UTF-32 support is configured, otherwise 0. If this
751: option is given to pcre_config() or pcre16_config(), or if the
752: PCRE_CONFIG_UTF8 or PCRE_CONFIG_UTF16 option is given to pcre32_con-
753: fig(), the result is the PCRE_ERROR_BADOPTION error.
754:
755:
756: CHARACTER CODES
757:
758: In 32-bit mode, when PCRE_UTF32 is not set, character values are
759: treated in the same way as in 8-bit, non UTF-8 mode, except, of course,
760: that they can range from 0 to 0x7fffffff instead of 0 to 0xff. Charac-
761: ter types for characters less than 0xff can therefore be influenced by
762: the locale in the same way as before. Characters greater than 0xff
763: have only one case, and no "type" (such as letter or digit).
764:
765: In UTF-32 mode, the character code is Unicode, in the range 0 to
766: 0x10ffff, with the exception of values in the range 0xd800 to 0xdfff
767: because those are "surrogate" values that are ill-formed in UTF-32.
768:
769: A UTF-32 string can indicate its endianness by special code knows as a
770: byte-order mark (BOM). The PCRE functions do not handle this, expecting
771: strings to be in host byte order. A utility function called
772: pcre32_utf32_to_host_byte_order() is provided to help with this (see
773: above).
774:
775:
776: ERROR NAMES
777:
778: The error PCRE_ERROR_BADUTF32 corresponds to its 8-bit counterpart.
779: The error PCRE_ERROR_BADMODE is given when a compiled pattern is passed
780: to a function that processes patterns in the other mode, for example,
781: if a pattern compiled with pcre_compile() is passed to pcre32_exec().
782:
783: There are new error codes whose names begin with PCRE_UTF32_ERR for
784: invalid UTF-32 strings, corresponding to the PCRE_UTF8_ERR codes for
785: UTF-8 strings that are described in the section entitled "Reason codes
786: for invalid UTF-8 strings" in the main pcreapi page. The UTF-32 errors
787: are:
788:
789: PCRE_UTF32_ERR1 Surrogate character (range from 0xd800 to 0xdfff)
790: PCRE_UTF32_ERR2 Non-character
791: PCRE_UTF32_ERR3 Character > 0x10ffff
792:
793:
794: ERROR TEXTS
795:
796: If there is an error while compiling a pattern, the error text that is
797: passed back by pcre32_compile() or pcre32_compile2() is still an 8-bit
798: character string, zero-terminated.
799:
800:
801: CALLOUTS
802:
803: The subject and mark fields in the callout block that is passed to a
804: callout function point to 32-bit vectors.
805:
806:
807: TESTING
808:
809: The pcretest program continues to operate with 8-bit input and output
810: files, but it can be used for testing the 32-bit library. If it is run
811: with the command line option -32, patterns and subject strings are con-
812: verted from 8-bit to 32-bit before being passed to PCRE, and the 32-bit
813: library functions are used instead of the 8-bit ones. Returned 32-bit
814: strings are converted to 8-bit for output. If both the 8-bit and the
815: 16-bit libraries were not compiled, pcretest defaults to 32-bit and the
816: -32 option is ignored.
817:
818: When PCRE is being built, the RunTest script that is called by "make
819: check" uses the pcretest -C option to discover which of the 8-bit,
820: 16-bit and 32-bit libraries has been built, and runs the tests appro-
821: priately.
822:
823:
824: NOT SUPPORTED IN 32-BIT MODE
825:
826: Not all the features of the 8-bit library are available with the 32-bit
827: library. The C++ and POSIX wrapper functions support only the 8-bit
828: library, and the pcregrep program is at present 8-bit only.
829:
830:
831: AUTHOR
832:
833: Philip Hazel
834: University Computing Service
835: Cambridge CB2 3QH, England.
836:
837:
838: REVISION
839:
840: Last updated: 12 May 2013
841: Copyright (c) 1997-2013 University of Cambridge.
842: ------------------------------------------------------------------------------
843:
844:
845: PCREBUILD(3) Library Functions Manual PCREBUILD(3)
846:
847:
848:
849: NAME
850: PCRE - Perl-compatible regular expressions
851:
852: BUILDING PCRE
853:
854: PCRE is distributed with a configure script that can be used to build
855: the library in Unix-like environments using the applications known as
856: Autotools. Also in the distribution are files to support building
857: using CMake instead of configure. The text file README contains general
858: information about building with Autotools (some of which is repeated
859: below), and also has some comments about building on various operating
860: systems. There is a lot more information about building PCRE without
861: using Autotools (including information about using CMake and building
862: "by hand") in the text file called NON-AUTOTOOLS-BUILD. You should
863: consult this file as well as the README file if you are building in a
864: non-Unix-like environment.
865:
866:
867: PCRE BUILD-TIME OPTIONS
868:
869: The rest of this document describes the optional features of PCRE that
870: can be selected when the library is compiled. It assumes use of the
871: configure script, where the optional features are selected or dese-
872: lected by providing options to configure before running the make com-
873: mand. However, the same options can be selected in both Unix-like and
874: non-Unix-like environments using the GUI facility of cmake-gui if you
875: are using CMake instead of configure to build PCRE.
876:
877: If you are not using Autotools or CMake, option selection can be done
878: by editing the config.h file, or by passing parameter settings to the
879: compiler, as described in NON-AUTOTOOLS-BUILD.
880:
881: The complete list of options for configure (which includes the standard
882: ones such as the selection of the installation directory) can be
883: obtained by running
884:
885: ./configure --help
886:
887: The following sections include descriptions of options whose names
888: begin with --enable or --disable. These settings specify changes to the
889: defaults for the configure command. Because of the way that configure
890: works, --enable and --disable always come in pairs, so the complemen-
891: tary option always exists as well, but as it specifies the default, it
892: is not described.
893:
894:
895: BUILDING 8-BIT, 16-BIT AND 32-BIT LIBRARIES
896:
897: By default, a library called libpcre is built, containing functions
898: that take string arguments contained in vectors of bytes, either as
899: single-byte characters, or interpreted as UTF-8 strings. You can also
900: build a separate library, called libpcre16, in which strings are con-
901: tained in vectors of 16-bit data units and interpreted either as sin-
902: gle-unit characters or UTF-16 strings, by adding
903:
904: --enable-pcre16
905:
906: to the configure command. You can also build yet another separate
907: library, called libpcre32, in which strings are contained in vectors of
908: 32-bit data units and interpreted either as single-unit characters or
909: UTF-32 strings, by adding
910:
911: --enable-pcre32
912:
913: to the configure command. If you do not want the 8-bit library, add
914:
915: --disable-pcre8
916:
917: as well. At least one of the three libraries must be built. Note that
918: the C++ and POSIX wrappers are for the 8-bit library only, and that
919: pcregrep is an 8-bit program. None of these are built if you select
920: only the 16-bit or 32-bit libraries.
921:
922:
923: BUILDING SHARED AND STATIC LIBRARIES
924:
925: The Autotools PCRE building process uses libtool to build both shared
926: and static libraries by default. You can suppress one of these by
927: adding one of
928:
929: --disable-shared
930: --disable-static
931:
932: to the configure command, as required.
933:
934:
935: C++ SUPPORT
936:
937: By default, if the 8-bit library is being built, the configure script
938: will search for a C++ compiler and C++ header files. If it finds them,
939: it automatically builds the C++ wrapper library (which supports only
940: 8-bit strings). You can disable this by adding
941:
942: --disable-cpp
943:
944: to the configure command.
945:
946:
947: UTF-8, UTF-16 AND UTF-32 SUPPORT
948:
949: To build PCRE with support for UTF Unicode character strings, add
950:
951: --enable-utf
952:
953: to the configure command. This setting applies to all three libraries,
954: adding support for UTF-8 to the 8-bit library, support for UTF-16 to
955: the 16-bit library, and support for UTF-32 to the to the 32-bit
956: library. There are no separate options for enabling UTF-8, UTF-16 and
957: UTF-32 independently because that would allow ridiculous settings such
958: as requesting UTF-16 support while building only the 8-bit library. It
959: is not possible to build one library with UTF support and another with-
960: out in the same configuration. (For backwards compatibility, --enable-
961: utf8 is a synonym of --enable-utf.)
962:
963: Of itself, this setting does not make PCRE treat strings as UTF-8,
964: UTF-16 or UTF-32. As well as compiling PCRE with this option, you also
965: have have to set the PCRE_UTF8, PCRE_UTF16 or PCRE_UTF32 option (as
966: appropriate) when you call one of the pattern compiling functions.
967:
968: If you set --enable-utf when compiling in an EBCDIC environment, PCRE
969: expects its input to be either ASCII or UTF-8 (depending on the run-
970: time option). It is not possible to support both EBCDIC and UTF-8 codes
971: in the same version of the library. Consequently, --enable-utf and
972: --enable-ebcdic are mutually exclusive.
973:
974:
975: UNICODE CHARACTER PROPERTY SUPPORT
976:
977: UTF support allows the libraries to process character codepoints up to
978: 0x10ffff in the strings that they handle. On its own, however, it does
979: not provide any facilities for accessing the properties of such charac-
980: ters. If you want to be able to use the pattern escapes \P, \p, and \X,
981: which refer to Unicode character properties, you must add
982:
983: --enable-unicode-properties
984:
985: to the configure command. This implies UTF support, even if you have
986: not explicitly requested it.
987:
988: Including Unicode property support adds around 30K of tables to the
989: PCRE library. Only the general category properties such as Lu and Nd
990: are supported. Details are given in the pcrepattern documentation.
991:
992:
993: JUST-IN-TIME COMPILER SUPPORT
994:
995: Just-in-time compiler support is included in the build by specifying
996:
997: --enable-jit
998:
999: This support is available only for certain hardware architectures. If
1000: this option is set for an unsupported architecture, a compile time
1001: error occurs. See the pcrejit documentation for a discussion of JIT
1002: usage. When JIT support is enabled, pcregrep automatically makes use of
1003: it, unless you add
1004:
1005: --disable-pcregrep-jit
1006:
1007: to the "configure" command.
1008:
1009:
1010: CODE VALUE OF NEWLINE
1011:
1012: By default, PCRE interprets the linefeed (LF) character as indicating
1013: the end of a line. This is the normal newline character on Unix-like
1014: systems. You can compile PCRE to use carriage return (CR) instead, by
1015: adding
1016:
1017: --enable-newline-is-cr
1018:
1019: to the configure command. There is also a --enable-newline-is-lf
1020: option, which explicitly specifies linefeed as the newline character.
1021:
1022: Alternatively, you can specify that line endings are to be indicated by
1023: the two character sequence CRLF. If you want this, add
1024:
1025: --enable-newline-is-crlf
1026:
1027: to the configure command. There is a fourth option, specified by
1028:
1029: --enable-newline-is-anycrlf
1030:
1031: which causes PCRE to recognize any of the three sequences CR, LF, or
1032: CRLF as indicating a line ending. Finally, a fifth option, specified by
1033:
1034: --enable-newline-is-any
1035:
1036: causes PCRE to recognize any Unicode newline sequence.
1037:
1038: Whatever line ending convention is selected when PCRE is built can be
1039: overridden when the library functions are called. At build time it is
1040: conventional to use the standard for your operating system.
1041:
1042:
1043: WHAT \R MATCHES
1044:
1045: By default, the sequence \R in a pattern matches any Unicode newline
1046: sequence, whatever has been selected as the line ending sequence. If
1047: you specify
1048:
1049: --enable-bsr-anycrlf
1050:
1051: the default is changed so that \R matches only CR, LF, or CRLF. What-
1052: ever is selected when PCRE is built can be overridden when the library
1053: functions are called.
1054:
1055:
1056: POSIX MALLOC USAGE
1057:
1058: When the 8-bit library is called through the POSIX interface (see the
1059: pcreposix documentation), additional working storage is required for
1060: holding the pointers to capturing substrings, because PCRE requires
1061: three integers per substring, whereas the POSIX interface provides only
1062: two. If the number of expected substrings is small, the wrapper func-
1063: tion uses space on the stack, because this is faster than using mal-
1064: loc() for each call. The default threshold above which the stack is no
1065: longer used is 10; it can be changed by adding a setting such as
1066:
1067: --with-posix-malloc-threshold=20
1068:
1069: to the configure command.
1070:
1071:
1072: HANDLING VERY LARGE PATTERNS
1073:
1074: Within a compiled pattern, offset values are used to point from one
1075: part to another (for example, from an opening parenthesis to an alter-
1076: nation metacharacter). By default, in the 8-bit and 16-bit libraries,
1077: two-byte values are used for these offsets, leading to a maximum size
1078: for a compiled pattern of around 64K. This is sufficient to handle all
1079: but the most gigantic patterns. Nevertheless, some people do want to
1080: process truly enormous patterns, so it is possible to compile PCRE to
1081: use three-byte or four-byte offsets by adding a setting such as
1082:
1083: --with-link-size=3
1084:
1085: to the configure command. The value given must be 2, 3, or 4. For the
1086: 16-bit library, a value of 3 is rounded up to 4. In these libraries,
1087: using longer offsets slows down the operation of PCRE because it has to
1088: load additional data when handling them. For the 32-bit library the
1089: value is always 4 and cannot be overridden; the value of --with-link-
1090: size is ignored.
1091:
1092:
1093: AVOIDING EXCESSIVE STACK USAGE
1094:
1095: When matching with the pcre_exec() function, PCRE implements backtrack-
1096: ing by making recursive calls to an internal function called match().
1097: In environments where the size of the stack is limited, this can se-
1098: verely limit PCRE's operation. (The Unix environment does not usually
1099: suffer from this problem, but it may sometimes be necessary to increase
1100: the maximum stack size. There is a discussion in the pcrestack docu-
1101: mentation.) An alternative approach to recursion that uses memory from
1102: the heap to remember data, instead of using recursive function calls,
1103: has been implemented to work round the problem of limited stack size.
1104: If you want to build a version of PCRE that works this way, add
1105:
1106: --disable-stack-for-recursion
1107:
1108: to the configure command. With this configuration, PCRE will use the
1109: pcre_stack_malloc and pcre_stack_free variables to call memory manage-
1110: ment functions. By default these point to malloc() and free(), but you
1111: can replace the pointers so that your own functions are used instead.
1112:
1113: Separate functions are provided rather than using pcre_malloc and
1114: pcre_free because the usage is very predictable: the block sizes
1115: requested are always the same, and the blocks are always freed in
1116: reverse order. A calling program might be able to implement optimized
1117: functions that perform better than malloc() and free(). PCRE runs
1118: noticeably more slowly when built in this way. This option affects only
1119: the pcre_exec() function; it is not relevant for pcre_dfa_exec().
1120:
1121:
1122: LIMITING PCRE RESOURCE USAGE
1123:
1124: Internally, PCRE has a function called match(), which it calls repeat-
1125: edly (sometimes recursively) when matching a pattern with the
1126: pcre_exec() function. By controlling the maximum number of times this
1127: function may be called during a single matching operation, a limit can
1128: be placed on the resources used by a single call to pcre_exec(). The
1129: limit can be changed at run time, as described in the pcreapi documen-
1130: tation. The default is 10 million, but this can be changed by adding a
1131: setting such as
1132:
1133: --with-match-limit=500000
1134:
1135: to the configure command. This setting has no effect on the
1136: pcre_dfa_exec() matching function.
1137:
1138: In some environments it is desirable to limit the depth of recursive
1139: calls of match() more strictly than the total number of calls, in order
1140: to restrict the maximum amount of stack (or heap, if --disable-stack-
1141: for-recursion is specified) that is used. A second limit controls this;
1142: it defaults to the value that is set for --with-match-limit, which
1143: imposes no additional constraints. However, you can set a lower limit
1144: by adding, for example,
1145:
1146: --with-match-limit-recursion=10000
1147:
1148: to the configure command. This value can also be overridden at run
1149: time.
1150:
1151:
1152: CREATING CHARACTER TABLES AT BUILD TIME
1153:
1154: PCRE uses fixed tables for processing characters whose code values are
1155: less than 256. By default, PCRE is built with a set of tables that are
1156: distributed in the file pcre_chartables.c.dist. These tables are for
1157: ASCII codes only. If you add
1158:
1159: --enable-rebuild-chartables
1160:
1161: to the configure command, the distributed tables are no longer used.
1162: Instead, a program called dftables is compiled and run. This outputs
1163: the source for new set of tables, created in the default locale of your
1164: C run-time system. (This method of replacing the tables does not work
1165: if you are cross compiling, because dftables is run on the local host.
1166: If you need to create alternative tables when cross compiling, you will
1167: have to do so "by hand".)
1168:
1169:
1170: USING EBCDIC CODE
1171:
1172: PCRE assumes by default that it will run in an environment where the
1173: character code is ASCII (or Unicode, which is a superset of ASCII).
1174: This is the case for most computer operating systems. PCRE can, how-
1175: ever, be compiled to run in an EBCDIC environment by adding
1176:
1177: --enable-ebcdic
1178:
1179: to the configure command. This setting implies --enable-rebuild-charta-
1180: bles. You should only use it if you know that you are in an EBCDIC
1181: environment (for example, an IBM mainframe operating system). The
1182: --enable-ebcdic option is incompatible with --enable-utf.
1183:
1184: The EBCDIC character that corresponds to an ASCII LF is assumed to have
1185: the value 0x15 by default. However, in some EBCDIC environments, 0x25
1186: is used. In such an environment you should use
1187:
1188: --enable-ebcdic-nl25
1189:
1190: as well as, or instead of, --enable-ebcdic. The EBCDIC character for CR
1191: has the same value as in ASCII, namely, 0x0d. Whichever of 0x15 and
1192: 0x25 is not chosen as LF is made to correspond to the Unicode NEL char-
1193: acter (which, in Unicode, is 0x85).
1194:
1195: The options that select newline behaviour, such as --enable-newline-is-
1196: cr, and equivalent run-time options, refer to these character values in
1197: an EBCDIC environment.
1198:
1199:
1200: PCREGREP OPTIONS FOR COMPRESSED FILE SUPPORT
1201:
1202: By default, pcregrep reads all files as plain text. You can build it so
1203: that it recognizes files whose names end in .gz or .bz2, and reads them
1204: with libz or libbz2, respectively, by adding one or both of
1205:
1206: --enable-pcregrep-libz
1207: --enable-pcregrep-libbz2
1208:
1209: to the configure command. These options naturally require that the rel-
1210: evant libraries are installed on your system. Configuration will fail
1211: if they are not.
1212:
1213:
1214: PCREGREP BUFFER SIZE
1215:
1216: pcregrep uses an internal buffer to hold a "window" on the file it is
1217: scanning, in order to be able to output "before" and "after" lines when
1218: it finds a match. The size of the buffer is controlled by a parameter
1219: whose default value is 20K. The buffer itself is three times this size,
1220: but because of the way it is used for holding "before" lines, the long-
1221: est line that is guaranteed to be processable is the parameter size.
1222: You can change the default parameter value by adding, for example,
1223:
1224: --with-pcregrep-bufsize=50K
1225:
1226: to the configure command. The caller of pcregrep can, however, override
1227: this value by specifying a run-time option.
1228:
1229:
1230: PCRETEST OPTION FOR LIBREADLINE SUPPORT
1231:
1232: If you add
1233:
1234: --enable-pcretest-libreadline
1235:
1236: to the configure command, pcretest is linked with the libreadline
1237: library, and when its input is from a terminal, it reads it using the
1238: readline() function. This provides line-editing and history facilities.
1239: Note that libreadline is GPL-licensed, so if you distribute a binary of
1240: pcretest linked in this way, there may be licensing issues.
1241:
1242: Setting this option causes the -lreadline option to be added to the
1243: pcretest build. In many operating environments with a sytem-installed
1244: libreadline this is sufficient. However, in some environments (e.g. if
1245: an unmodified distribution version of readline is in use), some extra
1246: configuration may be necessary. The INSTALL file for libreadline says
1247: this:
1248:
1249: "Readline uses the termcap functions, but does not link with the
1250: termcap or curses library itself, allowing applications which link
1251: with readline the to choose an appropriate library."
1252:
1253: If your environment has not been set up so that an appropriate library
1254: is automatically included, you may need to add something like
1255:
1256: LIBS="-ncurses"
1257:
1258: immediately before the configure command.
1259:
1260:
1261: DEBUGGING WITH VALGRIND SUPPORT
1262:
1263: By adding the
1264:
1265: --enable-valgrind
1266:
1267: option to to the configure command, PCRE will use valgrind annotations
1268: to mark certain memory regions as unaddressable. This allows it to
1269: detect invalid memory accesses, and is mostly useful for debugging PCRE
1270: itself.
1271:
1272:
1273: CODE COVERAGE REPORTING
1274:
1275: If your C compiler is gcc, you can build a version of PCRE that can
1276: generate a code coverage report for its test suite. To enable this, you
1277: must install lcov version 1.6 or above. Then specify
1278:
1279: --enable-coverage
1280:
1281: to the configure command and build PCRE in the usual way.
1282:
1283: Note that using ccache (a caching C compiler) is incompatible with code
1284: coverage reporting. If you have configured ccache to run automatically
1285: on your system, you must set the environment variable
1286:
1287: CCACHE_DISABLE=1
1288:
1289: before running make to build PCRE, so that ccache is not used.
1290:
1291: When --enable-coverage is used, the following addition targets are
1292: added to the Makefile:
1293:
1294: make coverage
1295:
1296: This creates a fresh coverage report for the PCRE test suite. It is
1297: equivalent to running "make coverage-reset", "make coverage-baseline",
1298: "make check", and then "make coverage-report".
1299:
1300: make coverage-reset
1301:
1302: This zeroes the coverage counters, but does nothing else.
1303:
1304: make coverage-baseline
1305:
1306: This captures baseline coverage information.
1307:
1308: make coverage-report
1309:
1310: This creates the coverage report.
1311:
1312: make coverage-clean-report
1313:
1314: This removes the generated coverage report without cleaning the cover-
1315: age data itself.
1316:
1317: make coverage-clean-data
1318:
1319: This removes the captured coverage data without removing the coverage
1320: files created at compile time (*.gcno).
1321:
1322: make coverage-clean
1323:
1324: This cleans all coverage data including the generated coverage report.
1325: For more information about code coverage, see the gcov and lcov docu-
1326: mentation.
1327:
1328:
1329: SEE ALSO
1330:
1331: pcreapi(3), pcre16, pcre32, pcre_config(3).
1332:
1333:
1334: AUTHOR
1335:
1336: Philip Hazel
1337: University Computing Service
1338: Cambridge CB2 3QH, England.
1339:
1340:
1341: REVISION
1342:
1343: Last updated: 12 May 2013
1344: Copyright (c) 1997-2013 University of Cambridge.
1345: ------------------------------------------------------------------------------
1346:
1347:
1348: PCREMATCHING(3) Library Functions Manual PCREMATCHING(3)
1349:
1350:
1351:
1352: NAME
1353: PCRE - Perl-compatible regular expressions
1354:
1355: PCRE MATCHING ALGORITHMS
1356:
1357: This document describes the two different algorithms that are available
1358: in PCRE for matching a compiled regular expression against a given sub-
1359: ject string. The "standard" algorithm is the one provided by the
1360: pcre_exec(), pcre16_exec() and pcre32_exec() functions. These work in
1361: the same as as Perl's matching function, and provide a Perl-compatible
1362: matching operation. The just-in-time (JIT) optimization that is
1363: described in the pcrejit documentation is compatible with these func-
1364: tions.
1365:
1366: An alternative algorithm is provided by the pcre_dfa_exec(),
1367: pcre16_dfa_exec() and pcre32_dfa_exec() functions; they operate in a
1368: different way, and are not Perl-compatible. This alternative has advan-
1369: tages and disadvantages compared with the standard algorithm, and these
1370: are described below.
1371:
1372: When there is only one possible way in which a given subject string can
1373: match a pattern, the two algorithms give the same answer. A difference
1374: arises, however, when there are multiple possibilities. For example, if
1375: the pattern
1376:
1377: ^<.*>
1378:
1379: is matched against the string
1380:
1381: <something> <something else> <something further>
1382:
1383: there are three possible answers. The standard algorithm finds only one
1384: of them, whereas the alternative algorithm finds all three.
1385:
1386:
1387: REGULAR EXPRESSIONS AS TREES
1388:
1389: The set of strings that are matched by a regular expression can be rep-
1390: resented as a tree structure. An unlimited repetition in the pattern
1391: makes the tree of infinite size, but it is still a tree. Matching the
1392: pattern to a given subject string (from a given starting point) can be
1393: thought of as a search of the tree. There are two ways to search a
1394: tree: depth-first and breadth-first, and these correspond to the two
1395: matching algorithms provided by PCRE.
1396:
1397:
1398: THE STANDARD MATCHING ALGORITHM
1399:
1400: In the terminology of Jeffrey Friedl's book "Mastering Regular Expres-
1401: sions", the standard algorithm is an "NFA algorithm". It conducts a
1402: depth-first search of the pattern tree. That is, it proceeds along a
1403: single path through the tree, checking that the subject matches what is
1404: required. When there is a mismatch, the algorithm tries any alterna-
1405: tives at the current point, and if they all fail, it backs up to the
1406: previous branch point in the tree, and tries the next alternative
1407: branch at that level. This often involves backing up (moving to the
1408: left) in the subject string as well. The order in which repetition
1409: branches are tried is controlled by the greedy or ungreedy nature of
1410: the quantifier.
1411:
1412: If a leaf node is reached, a matching string has been found, and at
1413: that point the algorithm stops. Thus, if there is more than one possi-
1414: ble match, this algorithm returns the first one that it finds. Whether
1415: this is the shortest, the longest, or some intermediate length depends
1416: on the way the greedy and ungreedy repetition quantifiers are specified
1417: in the pattern.
1418:
1419: Because it ends up with a single path through the tree, it is rela-
1420: tively straightforward for this algorithm to keep track of the sub-
1421: strings that are matched by portions of the pattern in parentheses.
1422: This provides support for capturing parentheses and back references.
1423:
1424:
1425: THE ALTERNATIVE MATCHING ALGORITHM
1426:
1427: This algorithm conducts a breadth-first search of the tree. Starting
1428: from the first matching point in the subject, it scans the subject
1429: string from left to right, once, character by character, and as it does
1430: this, it remembers all the paths through the tree that represent valid
1431: matches. In Friedl's terminology, this is a kind of "DFA algorithm",
1432: though it is not implemented as a traditional finite state machine (it
1433: keeps multiple states active simultaneously).
1434:
1435: Although the general principle of this matching algorithm is that it
1436: scans the subject string only once, without backtracking, there is one
1437: exception: when a lookaround assertion is encountered, the characters
1438: following or preceding the current point have to be independently
1439: inspected.
1440:
1441: The scan continues until either the end of the subject is reached, or
1442: there are no more unterminated paths. At this point, terminated paths
1443: represent the different matching possibilities (if there are none, the
1444: match has failed). Thus, if there is more than one possible match,
1445: this algorithm finds all of them, and in particular, it finds the long-
1446: est. The matches are returned in decreasing order of length. There is
1447: an option to stop the algorithm after the first match (which is neces-
1448: sarily the shortest) is found.
1449:
1450: Note that all the matches that are found start at the same point in the
1451: subject. If the pattern
1452:
1453: cat(er(pillar)?)?
1454:
1455: is matched against the string "the caterpillar catchment", the result
1456: will be the three strings "caterpillar", "cater", and "cat" that start
1457: at the fifth character of the subject. The algorithm does not automati-
1458: cally move on to find matches that start at later positions.
1459:
1460: PCRE's "auto-possessification" optimization usually applies to charac-
1461: ter repeats at the end of a pattern (as well as internally). For exam-
1462: ple, the pattern "a\d+" is compiled as if it were "a\d++" because there
1463: is no point even considering the possibility of backtracking into the
1464: repeated digits. For DFA matching, this means that only one possible
1465: match is found. If you really do want multiple matches in such cases,
1466: either use an ungreedy repeat ("a\d+?") or set the PCRE_NO_AUTO_POSSESS
1467: option when compiling.
1468:
1469: There are a number of features of PCRE regular expressions that are not
1470: supported by the alternative matching algorithm. They are as follows:
1471:
1472: 1. Because the algorithm finds all possible matches, the greedy or
1473: ungreedy nature of repetition quantifiers is not relevant. Greedy and
1474: ungreedy quantifiers are treated in exactly the same way. However, pos-
1475: sessive quantifiers can make a difference when what follows could also
1476: match what is quantified, for example in a pattern like this:
1477:
1478: ^a++\w!
1479:
1480: This pattern matches "aaab!" but not "aaa!", which would be matched by
1481: a non-possessive quantifier. Similarly, if an atomic group is present,
1482: it is matched as if it were a standalone pattern at the current point,
1483: and the longest match is then "locked in" for the rest of the overall
1484: pattern.
1485:
1486: 2. When dealing with multiple paths through the tree simultaneously, it
1487: is not straightforward to keep track of captured substrings for the
1488: different matching possibilities, and PCRE's implementation of this
1489: algorithm does not attempt to do this. This means that no captured sub-
1490: strings are available.
1491:
1492: 3. Because no substrings are captured, back references within the pat-
1493: tern are not supported, and cause errors if encountered.
1494:
1495: 4. For the same reason, conditional expressions that use a backrefer-
1496: ence as the condition or test for a specific group recursion are not
1497: supported.
1498:
1499: 5. Because many paths through the tree may be active, the \K escape
1500: sequence, which resets the start of the match when encountered (but may
1501: be on some paths and not on others), is not supported. It causes an
1502: error if encountered.
1503:
1504: 6. Callouts are supported, but the value of the capture_top field is
1505: always 1, and the value of the capture_last field is always -1.
1506:
1507: 7. The \C escape sequence, which (in the standard algorithm) always
1508: matches a single data unit, even in UTF-8, UTF-16 or UTF-32 modes, is
1509: not supported in these modes, because the alternative algorithm moves
1510: through the subject string one character (not data unit) at a time, for
1511: all active paths through the tree.
1512:
1513: 8. Except for (*FAIL), the backtracking control verbs such as (*PRUNE)
1514: are not supported. (*FAIL) is supported, and behaves like a failing
1515: negative assertion.
1516:
1517:
1518: ADVANTAGES OF THE ALTERNATIVE ALGORITHM
1519:
1520: Using the alternative matching algorithm provides the following advan-
1521: tages:
1522:
1523: 1. All possible matches (at a single point in the subject) are automat-
1524: ically found, and in particular, the longest match is found. To find
1525: more than one match using the standard algorithm, you have to do kludgy
1526: things with callouts.
1527:
1528: 2. Because the alternative algorithm scans the subject string just
1529: once, and never needs to backtrack (except for lookbehinds), it is pos-
1530: sible to pass very long subject strings to the matching function in
1531: several pieces, checking for partial matching each time. Although it is
1532: possible to do multi-segment matching using the standard algorithm by
1533: retaining partially matched substrings, it is more complicated. The
1534: pcrepartial documentation gives details of partial matching and dis-
1535: cusses multi-segment matching.
1536:
1537:
1538: DISADVANTAGES OF THE ALTERNATIVE ALGORITHM
1539:
1540: The alternative algorithm suffers from a number of disadvantages:
1541:
1542: 1. It is substantially slower than the standard algorithm. This is
1543: partly because it has to search for all possible matches, but is also
1544: because it is less susceptible to optimization.
1545:
1546: 2. Capturing parentheses and back references are not supported.
1547:
1548: 3. Although atomic groups are supported, their use does not provide the
1549: performance advantage that it does for the standard algorithm.
1550:
1551:
1552: AUTHOR
1553:
1554: Philip Hazel
1555: University Computing Service
1556: Cambridge CB2 3QH, England.
1557:
1558:
1559: REVISION
1560:
1561: Last updated: 12 November 2013
1562: Copyright (c) 1997-2012 University of Cambridge.
1563: ------------------------------------------------------------------------------
1564:
1565:
1566: PCREAPI(3) Library Functions Manual PCREAPI(3)
1567:
1568:
1569:
1570: NAME
1571: PCRE - Perl-compatible regular expressions
1572:
1573: #include <pcre.h>
1574:
1575:
1576: PCRE NATIVE API BASIC FUNCTIONS
1577:
1578: pcre *pcre_compile(const char *pattern, int options,
1579: const char **errptr, int *erroffset,
1580: const unsigned char *tableptr);
1581:
1582: pcre *pcre_compile2(const char *pattern, int options,
1583: int *errorcodeptr,
1584: const char **errptr, int *erroffset,
1585: const unsigned char *tableptr);
1586:
1587: pcre_extra *pcre_study(const pcre *code, int options,
1588: const char **errptr);
1589:
1590: void pcre_free_study(pcre_extra *extra);
1591:
1592: int pcre_exec(const pcre *code, const pcre_extra *extra,
1593: const char *subject, int length, int startoffset,
1594: int options, int *ovector, int ovecsize);
1595:
1596: int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
1597: const char *subject, int length, int startoffset,
1598: int options, int *ovector, int ovecsize,
1599: int *workspace, int wscount);
1600:
1601:
1602: PCRE NATIVE API STRING EXTRACTION FUNCTIONS
1603:
1604: int pcre_copy_named_substring(const pcre *code,
1605: const char *subject, int *ovector,
1606: int stringcount, const char *stringname,
1607: char *buffer, int buffersize);
1608:
1609: int pcre_copy_substring(const char *subject, int *ovector,
1610: int stringcount, int stringnumber, char *buffer,
1611: int buffersize);
1612:
1613: int pcre_get_named_substring(const pcre *code,
1614: const char *subject, int *ovector,
1615: int stringcount, const char *stringname,
1616: const char **stringptr);
1617:
1618: int pcre_get_stringnumber(const pcre *code,
1619: const char *name);
1620:
1621: int pcre_get_stringtable_entries(const pcre *code,
1622: const char *name, char **first, char **last);
1623:
1624: int pcre_get_substring(const char *subject, int *ovector,
1625: int stringcount, int stringnumber,
1626: const char **stringptr);
1627:
1628: int pcre_get_substring_list(const char *subject,
1629: int *ovector, int stringcount, const char ***listptr);
1630:
1631: void pcre_free_substring(const char *stringptr);
1632:
1633: void pcre_free_substring_list(const char **stringptr);
1634:
1635:
1636: PCRE NATIVE API AUXILIARY FUNCTIONS
1637:
1638: int pcre_jit_exec(const pcre *code, const pcre_extra *extra,
1639: const char *subject, int length, int startoffset,
1640: int options, int *ovector, int ovecsize,
1641: pcre_jit_stack *jstack);
1642:
1643: pcre_jit_stack *pcre_jit_stack_alloc(int startsize, int maxsize);
1644:
1645: void pcre_jit_stack_free(pcre_jit_stack *stack);
1646:
1647: void pcre_assign_jit_stack(pcre_extra *extra,
1648: pcre_jit_callback callback, void *data);
1649:
1650: const unsigned char *pcre_maketables(void);
1651:
1652: int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
1653: int what, void *where);
1654:
1655: int pcre_refcount(pcre *code, int adjust);
1656:
1657: int pcre_config(int what, void *where);
1658:
1659: const char *pcre_version(void);
1660:
1661: int pcre_pattern_to_host_byte_order(pcre *code,
1662: pcre_extra *extra, const unsigned char *tables);
1663:
1664:
1665: PCRE NATIVE API INDIRECTED FUNCTIONS
1666:
1667: void *(*pcre_malloc)(size_t);
1668:
1669: void (*pcre_free)(void *);
1670:
1671: void *(*pcre_stack_malloc)(size_t);
1672:
1673: void (*pcre_stack_free)(void *);
1674:
1675: int (*pcre_callout)(pcre_callout_block *);
1676:
1677:
1678: PCRE 8-BIT, 16-BIT, AND 32-BIT LIBRARIES
1679:
1680: As well as support for 8-bit character strings, PCRE also supports
1681: 16-bit strings (from release 8.30) and 32-bit strings (from release
1682: 8.32), by means of two additional libraries. They can be built as well
1683: as, or instead of, the 8-bit library. To avoid too much complication,
1684: this document describes the 8-bit versions of the functions, with only
1685: occasional references to the 16-bit and 32-bit libraries.
1686:
1687: The 16-bit and 32-bit functions operate in the same way as their 8-bit
1688: counterparts; they just use different data types for their arguments
1689: and results, and their names start with pcre16_ or pcre32_ instead of
1690: pcre_. For every option that has UTF8 in its name (for example,
1691: PCRE_UTF8), there are corresponding 16-bit and 32-bit names with UTF8
1692: replaced by UTF16 or UTF32, respectively. This facility is in fact just
1693: cosmetic; the 16-bit and 32-bit option names define the same bit val-
1694: ues.
1695:
1696: References to bytes and UTF-8 in this document should be read as refer-
1697: ences to 16-bit data units and UTF-16 when using the 16-bit library, or
1698: 32-bit data units and UTF-32 when using the 32-bit library, unless
1699: specified otherwise. More details of the specific differences for the
1700: 16-bit and 32-bit libraries are given in the pcre16 and pcre32 pages.
1701:
1702:
1703: PCRE API OVERVIEW
1704:
1705: PCRE has its own native API, which is described in this document. There
1706: are also some wrapper functions (for the 8-bit library only) that cor-
1707: respond to the POSIX regular expression API, but they do not give
1708: access to all the functionality. They are described in the pcreposix
1709: documentation. Both of these APIs define a set of C function calls. A
1710: C++ wrapper (again for the 8-bit library only) is also distributed with
1711: PCRE. It is documented in the pcrecpp page.
1712:
1713: The native API C function prototypes are defined in the header file
1714: pcre.h, and on Unix-like systems the (8-bit) library itself is called
1715: libpcre. It can normally be accessed by adding -lpcre to the command
1716: for linking an application that uses PCRE. The header file defines the
1717: macros PCRE_MAJOR and PCRE_MINOR to contain the major and minor release
1718: numbers for the library. Applications can use these to include support
1719: for different releases of PCRE.
1720:
1721: In a Windows environment, if you want to statically link an application
1722: program against a non-dll pcre.a file, you must define PCRE_STATIC
1723: before including pcre.h or pcrecpp.h, because otherwise the pcre_mal-
1724: loc() and pcre_free() exported functions will be declared
1725: __declspec(dllimport), with unwanted results.
1726:
1727: The functions pcre_compile(), pcre_compile2(), pcre_study(), and
1728: pcre_exec() are used for compiling and matching regular expressions in
1729: a Perl-compatible manner. A sample program that demonstrates the sim-
1730: plest way of using them is provided in the file called pcredemo.c in
1731: the PCRE source distribution. A listing of this program is given in the
1732: pcredemo documentation, and the pcresample documentation describes how
1733: to compile and run it.
1734:
1735: Just-in-time compiler support is an optional feature of PCRE that can
1736: be built in appropriate hardware environments. It greatly speeds up the
1737: matching performance of many patterns. Simple programs can easily
1738: request that it be used if available, by setting an option that is
1739: ignored when it is not relevant. More complicated programs might need
1740: to make use of the functions pcre_jit_stack_alloc(),
1741: pcre_jit_stack_free(), and pcre_assign_jit_stack() in order to control
1742: the JIT code's memory usage.
1743:
1744: From release 8.32 there is also a direct interface for JIT execution,
1745: which gives improved performance. The JIT-specific functions are dis-
1746: cussed in the pcrejit documentation.
1747:
1748: A second matching function, pcre_dfa_exec(), which is not Perl-compati-
1749: ble, is also provided. This uses a different algorithm for the match-
1750: ing. The alternative algorithm finds all possible matches (at a given
1751: point in the subject), and scans the subject just once (unless there
1752: are lookbehind assertions). However, this algorithm does not return
1753: captured substrings. A description of the two matching algorithms and
1754: their advantages and disadvantages is given in the pcrematching docu-
1755: mentation.
1756:
1757: In addition to the main compiling and matching functions, there are
1758: convenience functions for extracting captured substrings from a subject
1759: string that is matched by pcre_exec(). They are:
1760:
1761: pcre_copy_substring()
1762: pcre_copy_named_substring()
1763: pcre_get_substring()
1764: pcre_get_named_substring()
1765: pcre_get_substring_list()
1766: pcre_get_stringnumber()
1767: pcre_get_stringtable_entries()
1768:
1769: pcre_free_substring() and pcre_free_substring_list() are also provided,
1770: to free the memory used for extracted strings.
1771:
1772: The function pcre_maketables() is used to build a set of character
1773: tables in the current locale for passing to pcre_compile(),
1774: pcre_exec(), or pcre_dfa_exec(). This is an optional facility that is
1775: provided for specialist use. Most commonly, no special tables are
1776: passed, in which case internal tables that are generated when PCRE is
1777: built are used.
1778:
1779: The function pcre_fullinfo() is used to find out information about a
1780: compiled pattern. The function pcre_version() returns a pointer to a
1781: string containing the version of PCRE and its date of release.
1782:
1783: The function pcre_refcount() maintains a reference count in a data
1784: block containing a compiled pattern. This is provided for the benefit
1785: of object-oriented applications.
1786:
1787: The global variables pcre_malloc and pcre_free initially contain the
1788: entry points of the standard malloc() and free() functions, respec-
1789: tively. PCRE calls the memory management functions via these variables,
1790: so a calling program can replace them if it wishes to intercept the
1791: calls. This should be done before calling any PCRE functions.
1792:
1793: The global variables pcre_stack_malloc and pcre_stack_free are also
1794: indirections to memory management functions. These special functions
1795: are used only when PCRE is compiled to use the heap for remembering
1796: data, instead of recursive function calls, when running the pcre_exec()
1797: function. See the pcrebuild documentation for details of how to do
1798: this. It is a non-standard way of building PCRE, for use in environ-
1799: ments that have limited stacks. Because of the greater use of memory
1800: management, it runs more slowly. Separate functions are provided so
1801: that special-purpose external code can be used for this case. When
1802: used, these functions are always called in a stack-like manner (last
1803: obtained, first freed), and always for memory blocks of the same size.
1804: There is a discussion about PCRE's stack usage in the pcrestack docu-
1805: mentation.
1806:
1807: The global variable pcre_callout initially contains NULL. It can be set
1808: by the caller to a "callout" function, which PCRE will then call at
1809: specified points during a matching operation. Details are given in the
1810: pcrecallout documentation.
1811:
1812:
1813: NEWLINES
1814:
1815: PCRE supports five different conventions for indicating line breaks in
1816: strings: a single CR (carriage return) character, a single LF (line-
1817: feed) character, the two-character sequence CRLF, any of the three pre-
1818: ceding, or any Unicode newline sequence. The Unicode newline sequences
1819: are the three just mentioned, plus the single characters VT (vertical
1820: tab, U+000B), FF (form feed, U+000C), NEL (next line, U+0085), LS (line
1821: separator, U+2028), and PS (paragraph separator, U+2029).
1822:
1823: Each of the first three conventions is used by at least one operating
1824: system as its standard newline sequence. When PCRE is built, a default
1825: can be specified. The default default is LF, which is the Unix stan-
1826: dard. When PCRE is run, the default can be overridden, either when a
1827: pattern is compiled, or when it is matched.
1828:
1829: At compile time, the newline convention can be specified by the options
1830: argument of pcre_compile(), or it can be specified by special text at
1831: the start of the pattern itself; this overrides any other settings. See
1832: the pcrepattern page for details of the special character sequences.
1833:
1834: In the PCRE documentation the word "newline" is used to mean "the char-
1835: acter or pair of characters that indicate a line break". The choice of
1836: newline convention affects the handling of the dot, circumflex, and
1837: dollar metacharacters, the handling of #-comments in /x mode, and, when
1838: CRLF is a recognized line ending sequence, the match position advance-
1839: ment for a non-anchored pattern. There is more detail about this in the
1840: section on pcre_exec() options below.
1841:
1842: The choice of newline convention does not affect the interpretation of
1843: the \n or \r escape sequences, nor does it affect what \R matches,
1844: which is controlled in a similar way, but by separate options.
1845:
1846:
1847: MULTITHREADING
1848:
1849: The PCRE functions can be used in multi-threading applications, with
1850: the proviso that the memory management functions pointed to by
1851: pcre_malloc, pcre_free, pcre_stack_malloc, and pcre_stack_free, and the
1852: callout function pointed to by pcre_callout, are shared by all threads.
1853:
1854: The compiled form of a regular expression is not altered during match-
1855: ing, so the same compiled pattern can safely be used by several threads
1856: at once.
1857:
1858: If the just-in-time optimization feature is being used, it needs sepa-
1859: rate memory stack areas for each thread. See the pcrejit documentation
1860: for more details.
1861:
1862:
1863: SAVING PRECOMPILED PATTERNS FOR LATER USE
1864:
1865: The compiled form of a regular expression can be saved and re-used at a
1866: later time, possibly by a different program, and even on a host other
1867: than the one on which it was compiled. Details are given in the
1868: pcreprecompile documentation, which includes a description of the
1869: pcre_pattern_to_host_byte_order() function. However, compiling a regu-
1870: lar expression with one version of PCRE for use with a different ver-
1871: sion is not guaranteed to work and may cause crashes.
1872:
1873:
1874: CHECKING BUILD-TIME OPTIONS
1875:
1876: int pcre_config(int what, void *where);
1877:
1878: The function pcre_config() makes it possible for a PCRE client to dis-
1879: cover which optional features have been compiled into the PCRE library.
1880: The pcrebuild documentation has more details about these optional fea-
1881: tures.
1882:
1883: The first argument for pcre_config() is an integer, specifying which
1884: information is required; the second argument is a pointer to a variable
1885: into which the information is placed. The returned value is zero on
1886: success, or the negative error code PCRE_ERROR_BADOPTION if the value
1887: in the first argument is not recognized. The following information is
1888: available:
1889:
1890: PCRE_CONFIG_UTF8
1891:
1892: The output is an integer that is set to one if UTF-8 support is avail-
1893: able; otherwise it is set to zero. This value should normally be given
1894: to the 8-bit version of this function, pcre_config(). If it is given to
1895: the 16-bit or 32-bit version of this function, the result is
1896: PCRE_ERROR_BADOPTION.
1897:
1898: PCRE_CONFIG_UTF16
1899:
1900: The output is an integer that is set to one if UTF-16 support is avail-
1901: able; otherwise it is set to zero. This value should normally be given
1902: to the 16-bit version of this function, pcre16_config(). If it is given
1903: to the 8-bit or 32-bit version of this function, the result is
1904: PCRE_ERROR_BADOPTION.
1905:
1906: PCRE_CONFIG_UTF32
1907:
1908: The output is an integer that is set to one if UTF-32 support is avail-
1909: able; otherwise it is set to zero. This value should normally be given
1910: to the 32-bit version of this function, pcre32_config(). If it is given
1911: to the 8-bit or 16-bit version of this function, the result is
1912: PCRE_ERROR_BADOPTION.
1913:
1914: PCRE_CONFIG_UNICODE_PROPERTIES
1915:
1916: The output is an integer that is set to one if support for Unicode
1917: character properties is available; otherwise it is set to zero.
1918:
1919: PCRE_CONFIG_JIT
1920:
1921: The output is an integer that is set to one if support for just-in-time
1922: compiling is available; otherwise it is set to zero.
1923:
1924: PCRE_CONFIG_JITTARGET
1925:
1926: The output is a pointer to a zero-terminated "const char *" string. If
1927: JIT support is available, the string contains the name of the architec-
1928: ture for which the JIT compiler is configured, for example "x86 32bit
1929: (little endian + unaligned)". If JIT support is not available, the
1930: result is NULL.
1931:
1932: PCRE_CONFIG_NEWLINE
1933:
1934: The output is an integer whose value specifies the default character
1935: sequence that is recognized as meaning "newline". The values that are
1936: supported in ASCII/Unicode environments are: 10 for LF, 13 for CR, 3338
1937: for CRLF, -2 for ANYCRLF, and -1 for ANY. In EBCDIC environments, CR,
1938: ANYCRLF, and ANY yield the same values. However, the value for LF is
1939: normally 21, though some EBCDIC environments use 37. The corresponding
1940: values for CRLF are 3349 and 3365. The default should normally corre-
1941: spond to the standard sequence for your operating system.
1942:
1943: PCRE_CONFIG_BSR
1944:
1945: The output is an integer whose value indicates what character sequences
1946: the \R escape sequence matches by default. A value of 0 means that \R
1947: matches any Unicode line ending sequence; a value of 1 means that \R
1948: matches only CR, LF, or CRLF. The default can be overridden when a pat-
1949: tern is compiled or matched.
1950:
1951: PCRE_CONFIG_LINK_SIZE
1952:
1953: The output is an integer that contains the number of bytes used for
1954: internal linkage in compiled regular expressions. For the 8-bit
1955: library, the value can be 2, 3, or 4. For the 16-bit library, the value
1956: is either 2 or 4 and is still a number of bytes. For the 32-bit
1957: library, the value is either 2 or 4 and is still a number of bytes. The
1958: default value of 2 is sufficient for all but the most massive patterns,
1959: since it allows the compiled pattern to be up to 64K in size. Larger
1960: values allow larger regular expressions to be compiled, at the expense
1961: of slower matching.
1962:
1963: PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
1964:
1965: The output is an integer that contains the threshold above which the
1966: POSIX interface uses malloc() for output vectors. Further details are
1967: given in the pcreposix documentation.
1968:
1969: PCRE_CONFIG_PARENS_LIMIT
1970:
1971: The output is a long integer that gives the maximum depth of nesting of
1972: parentheses (of any kind) in a pattern. This limit is imposed to cap
1973: the amount of system stack used when a pattern is compiled. It is spec-
1974: ified when PCRE is built; the default is 250.
1975:
1976: PCRE_CONFIG_MATCH_LIMIT
1977:
1978: The output is a long integer that gives the default limit for the num-
1979: ber of internal matching function calls in a pcre_exec() execution.
1980: Further details are given with pcre_exec() below.
1981:
1982: PCRE_CONFIG_MATCH_LIMIT_RECURSION
1983:
1984: The output is a long integer that gives the default limit for the depth
1985: of recursion when calling the internal matching function in a
1986: pcre_exec() execution. Further details are given with pcre_exec()
1987: below.
1988:
1989: PCRE_CONFIG_STACKRECURSE
1990:
1991: The output is an integer that is set to one if internal recursion when
1992: running pcre_exec() is implemented by recursive function calls that use
1993: the stack to remember their state. This is the usual way that PCRE is
1994: compiled. The output is zero if PCRE was compiled to use blocks of data
1995: on the heap instead of recursive function calls. In this case,
1996: pcre_stack_malloc and pcre_stack_free are called to manage memory
1997: blocks on the heap, thus avoiding the use of the stack.
1998:
1999:
2000: COMPILING A PATTERN
2001:
2002: pcre *pcre_compile(const char *pattern, int options,
2003: const char **errptr, int *erroffset,
2004: const unsigned char *tableptr);
2005:
2006: pcre *pcre_compile2(const char *pattern, int options,
2007: int *errorcodeptr,
2008: const char **errptr, int *erroffset,
2009: const unsigned char *tableptr);
2010:
2011: Either of the functions pcre_compile() or pcre_compile2() can be called
2012: to compile a pattern into an internal form. The only difference between
2013: the two interfaces is that pcre_compile2() has an additional argument,
2014: errorcodeptr, via which a numerical error code can be returned. To
2015: avoid too much repetition, we refer just to pcre_compile() below, but
2016: the information applies equally to pcre_compile2().
2017:
2018: The pattern is a C string terminated by a binary zero, and is passed in
2019: the pattern argument. A pointer to a single block of memory that is
2020: obtained via pcre_malloc is returned. This contains the compiled code
2021: and related data. The pcre type is defined for the returned block; this
2022: is a typedef for a structure whose contents are not externally defined.
2023: It is up to the caller to free the memory (via pcre_free) when it is no
2024: longer required.
2025:
2026: Although the compiled code of a PCRE regex is relocatable, that is, it
2027: does not depend on memory location, the complete pcre data block is not
2028: fully relocatable, because it may contain a copy of the tableptr argu-
2029: ment, which is an address (see below).
2030:
2031: The options argument contains various bit settings that affect the com-
2032: pilation. It should be zero if no options are required. The available
2033: options are described below. Some of them (in particular, those that
2034: are compatible with Perl, but some others as well) can also be set and
2035: unset from within the pattern (see the detailed description in the
2036: pcrepattern documentation). For those options that can be different in
2037: different parts of the pattern, the contents of the options argument
2038: specifies their settings at the start of compilation and execution. The
2039: PCRE_ANCHORED, PCRE_BSR_xxx, PCRE_NEWLINE_xxx, PCRE_NO_UTF8_CHECK, and
2040: PCRE_NO_START_OPTIMIZE options can be set at the time of matching as
2041: well as at compile time.
2042:
2043: If errptr is NULL, pcre_compile() returns NULL immediately. Otherwise,
2044: if compilation of a pattern fails, pcre_compile() returns NULL, and
2045: sets the variable pointed to by errptr to point to a textual error mes-
2046: sage. This is a static string that is part of the library. You must not
2047: try to free it. Normally, the offset from the start of the pattern to
2048: the data unit that was being processed when the error was discovered is
2049: placed in the variable pointed to by erroffset, which must not be NULL
2050: (if it is, an immediate error is given). However, for an invalid UTF-8
2051: or UTF-16 string, the offset is that of the first data unit of the
2052: failing character.
2053:
2054: Some errors are not detected until the whole pattern has been scanned;
2055: in these cases, the offset passed back is the length of the pattern.
2056: Note that the offset is in data units, not characters, even in a UTF
2057: mode. It may sometimes point into the middle of a UTF-8 or UTF-16 char-
2058: acter.
2059:
2060: If pcre_compile2() is used instead of pcre_compile(), and the error-
2061: codeptr argument is not NULL, a non-zero error code number is returned
2062: via this argument in the event of an error. This is in addition to the
2063: textual error message. Error codes and messages are listed below.
2064:
2065: If the final argument, tableptr, is NULL, PCRE uses a default set of
2066: character tables that are built when PCRE is compiled, using the
2067: default C locale. Otherwise, tableptr must be an address that is the
2068: result of a call to pcre_maketables(). This value is stored with the
2069: compiled pattern, and used again by pcre_exec() and pcre_dfa_exec()
2070: when the pattern is matched. For more discussion, see the section on
2071: locale support below.
2072:
2073: This code fragment shows a typical straightforward call to pcre_com-
2074: pile():
2075:
2076: pcre *re;
2077: const char *error;
2078: int erroffset;
2079: re = pcre_compile(
2080: "^A.*Z", /* the pattern */
2081: 0, /* default options */
2082: &error, /* for error message */
2083: &erroffset, /* for error offset */
2084: NULL); /* use default character tables */
2085:
2086: The following names for option bits are defined in the pcre.h header
2087: file:
2088:
2089: PCRE_ANCHORED
2090:
2091: If this bit is set, the pattern is forced to be "anchored", that is, it
2092: is constrained to match only at the first matching point in the string
2093: that is being searched (the "subject string"). This effect can also be
2094: achieved by appropriate constructs in the pattern itself, which is the
2095: only way to do it in Perl.
2096:
2097: PCRE_AUTO_CALLOUT
2098:
2099: If this bit is set, pcre_compile() automatically inserts callout items,
2100: all with number 255, before each pattern item. For discussion of the
2101: callout facility, see the pcrecallout documentation.
2102:
2103: PCRE_BSR_ANYCRLF
2104: PCRE_BSR_UNICODE
2105:
2106: These options (which are mutually exclusive) control what the \R escape
2107: sequence matches. The choice is either to match only CR, LF, or CRLF,
2108: or to match any Unicode newline sequence. The default is specified when
2109: PCRE is built. It can be overridden from within the pattern, or by set-
2110: ting an option when a compiled pattern is matched.
2111:
2112: PCRE_CASELESS
2113:
2114: If this bit is set, letters in the pattern match both upper and lower
2115: case letters. It is equivalent to Perl's /i option, and it can be
2116: changed within a pattern by a (?i) option setting. In UTF-8 mode, PCRE
2117: always understands the concept of case for characters whose values are
2118: less than 128, so caseless matching is always possible. For characters
2119: with higher values, the concept of case is supported if PCRE is com-
2120: piled with Unicode property support, but not otherwise. If you want to
2121: use caseless matching for characters 128 and above, you must ensure
2122: that PCRE is compiled with Unicode property support as well as with
2123: UTF-8 support.
2124:
2125: PCRE_DOLLAR_ENDONLY
2126:
2127: If this bit is set, a dollar metacharacter in the pattern matches only
2128: at the end of the subject string. Without this option, a dollar also
2129: matches immediately before a newline at the end of the string (but not
2130: before any other newlines). The PCRE_DOLLAR_ENDONLY option is ignored
2131: if PCRE_MULTILINE is set. There is no equivalent to this option in
2132: Perl, and no way to set it within a pattern.
2133:
2134: PCRE_DOTALL
2135:
2136: If this bit is set, a dot metacharacter in the pattern matches a char-
2137: acter of any value, including one that indicates a newline. However, it
2138: only ever matches one character, even if newlines are coded as CRLF.
2139: Without this option, a dot does not match when the current position is
2140: at a newline. This option is equivalent to Perl's /s option, and it can
2141: be changed within a pattern by a (?s) option setting. A negative class
2142: such as [^a] always matches newline characters, independent of the set-
2143: ting of this option.
2144:
2145: PCRE_DUPNAMES
2146:
2147: If this bit is set, names used to identify capturing subpatterns need
2148: not be unique. This can be helpful for certain types of pattern when it
2149: is known that only one instance of the named subpattern can ever be
2150: matched. There are more details of named subpatterns below; see also
2151: the pcrepattern documentation.
2152:
2153: PCRE_EXTENDED
2154:
2155: If this bit is set, most white space characters in the pattern are
2156: totally ignored except when escaped or inside a character class. How-
2157: ever, white space is not allowed within sequences such as (?> that
2158: introduce various parenthesized subpatterns, nor within a numerical
2159: quantifier such as {1,3}. However, ignorable white space is permitted
2160: between an item and a following quantifier and between a quantifier and
2161: a following + that indicates possessiveness.
2162:
2163: White space did not used to include the VT character (code 11), because
2164: Perl did not treat this character as white space. However, Perl changed
2165: at release 5.18, so PCRE followed at release 8.34, and VT is now
2166: treated as white space.
2167:
2168: PCRE_EXTENDED also causes characters between an unescaped # outside a
2169: character class and the next newline, inclusive, to be ignored.
2170: PCRE_EXTENDED is equivalent to Perl's /x option, and it can be changed
2171: within a pattern by a (?x) option setting.
2172:
2173: Which characters are interpreted as newlines is controlled by the
2174: options passed to pcre_compile() or by a special sequence at the start
2175: of the pattern, as described in the section entitled "Newline conven-
2176: tions" in the pcrepattern documentation. Note that the end of this type
2177: of comment is a literal newline sequence in the pattern; escape
2178: sequences that happen to represent a newline do not count.
2179:
2180: This option makes it possible to include comments inside complicated
2181: patterns. Note, however, that this applies only to data characters.
2182: White space characters may never appear within special character
2183: sequences in a pattern, for example within the sequence (?( that intro-
2184: duces a conditional subpattern.
2185:
2186: PCRE_EXTRA
2187:
2188: This option was invented in order to turn on additional functionality
2189: of PCRE that is incompatible with Perl, but it is currently of very
2190: little use. When set, any backslash in a pattern that is followed by a
2191: letter that has no special meaning causes an error, thus reserving
2192: these combinations for future expansion. By default, as in Perl, a
2193: backslash followed by a letter with no special meaning is treated as a
2194: literal. (Perl can, however, be persuaded to give an error for this, by
2195: running it with the -w option.) There are at present no other features
2196: controlled by this option. It can also be set by a (?X) option setting
2197: within a pattern.
2198:
2199: PCRE_FIRSTLINE
2200:
2201: If this option is set, an unanchored pattern is required to match
2202: before or at the first newline in the subject string, though the
2203: matched text may continue over the newline.
2204:
2205: PCRE_JAVASCRIPT_COMPAT
2206:
2207: If this option is set, PCRE's behaviour is changed in some ways so that
2208: it is compatible with JavaScript rather than Perl. The changes are as
2209: follows:
2210:
2211: (1) A lone closing square bracket in a pattern causes a compile-time
2212: error, because this is illegal in JavaScript (by default it is treated
2213: as a data character). Thus, the pattern AB]CD becomes illegal when this
2214: option is set.
2215:
2216: (2) At run time, a back reference to an unset subpattern group matches
2217: an empty string (by default this causes the current matching alterna-
2218: tive to fail). A pattern such as (\1)(a) succeeds when this option is
2219: set (assuming it can find an "a" in the subject), whereas it fails by
2220: default, for Perl compatibility.
2221:
2222: (3) \U matches an upper case "U" character; by default \U causes a com-
2223: pile time error (Perl uses \U to upper case subsequent characters).
2224:
2225: (4) \u matches a lower case "u" character unless it is followed by four
2226: hexadecimal digits, in which case the hexadecimal number defines the
2227: code point to match. By default, \u causes a compile time error (Perl
2228: uses it to upper case the following character).
2229:
2230: (5) \x matches a lower case "x" character unless it is followed by two
2231: hexadecimal digits, in which case the hexadecimal number defines the
2232: code point to match. By default, as in Perl, a hexadecimal number is
2233: always expected after \x, but it may have zero, one, or two digits (so,
2234: for example, \xz matches a binary zero character followed by z).
2235:
2236: PCRE_MULTILINE
2237:
2238: By default, for the purposes of matching "start of line" and "end of
2239: line", PCRE treats the subject string as consisting of a single line of
2240: characters, even if it actually contains newlines. The "start of line"
2241: metacharacter (^) matches only at the start of the string, and the "end
2242: of line" metacharacter ($) matches only at the end of the string, or
2243: before a terminating newline (except when PCRE_DOLLAR_ENDONLY is set).
2244: Note, however, that unless PCRE_DOTALL is set, the "any character"
2245: metacharacter (.) does not match at a newline. This behaviour (for ^,
2246: $, and dot) is the same as Perl.
2247:
2248: When PCRE_MULTILINE it is set, the "start of line" and "end of line"
2249: constructs match immediately following or immediately before internal
2250: newlines in the subject string, respectively, as well as at the very
2251: start and end. This is equivalent to Perl's /m option, and it can be
2252: changed within a pattern by a (?m) option setting. If there are no new-
2253: lines in a subject string, or no occurrences of ^ or $ in a pattern,
2254: setting PCRE_MULTILINE has no effect.
2255:
2256: PCRE_NEVER_UTF
2257:
2258: This option locks out interpretation of the pattern as UTF-8 (or UTF-16
2259: or UTF-32 in the 16-bit and 32-bit libraries). In particular, it pre-
2260: vents the creator of the pattern from switching to UTF interpretation
2261: by starting the pattern with (*UTF). This may be useful in applications
2262: that process patterns from external sources. The combination of
2263: PCRE_UTF8 and PCRE_NEVER_UTF also causes an error.
2264:
2265: PCRE_NEWLINE_CR
2266: PCRE_NEWLINE_LF
2267: PCRE_NEWLINE_CRLF
2268: PCRE_NEWLINE_ANYCRLF
2269: PCRE_NEWLINE_ANY
2270:
2271: These options override the default newline definition that was chosen
2272: when PCRE was built. Setting the first or the second specifies that a
2273: newline is indicated by a single character (CR or LF, respectively).
2274: Setting PCRE_NEWLINE_CRLF specifies that a newline is indicated by the
2275: two-character CRLF sequence. Setting PCRE_NEWLINE_ANYCRLF specifies
2276: that any of the three preceding sequences should be recognized. Setting
2277: PCRE_NEWLINE_ANY specifies that any Unicode newline sequence should be
2278: recognized.
2279:
2280: In an ASCII/Unicode environment, the Unicode newline sequences are the
2281: three just mentioned, plus the single characters VT (vertical tab,
2282: U+000B), FF (form feed, U+000C), NEL (next line, U+0085), LS (line sep-
2283: arator, U+2028), and PS (paragraph separator, U+2029). For the 8-bit
2284: library, the last two are recognized only in UTF-8 mode.
2285:
2286: When PCRE is compiled to run in an EBCDIC (mainframe) environment, the
2287: code for CR is 0x0d, the same as ASCII. However, the character code for
2288: LF is normally 0x15, though in some EBCDIC environments 0x25 is used.
2289: Whichever of these is not LF is made to correspond to Unicode's NEL
2290: character. EBCDIC codes are all less than 256. For more details, see
2291: the pcrebuild documentation.
2292:
2293: The newline setting in the options word uses three bits that are
2294: treated as a number, giving eight possibilities. Currently only six are
2295: used (default plus the five values above). This means that if you set
2296: more than one newline option, the combination may or may not be sensi-
2297: ble. For example, PCRE_NEWLINE_CR with PCRE_NEWLINE_LF is equivalent to
2298: PCRE_NEWLINE_CRLF, but other combinations may yield unused numbers and
2299: cause an error.
2300:
2301: The only time that a line break in a pattern is specially recognized
2302: when compiling is when PCRE_EXTENDED is set. CR and LF are white space
2303: characters, and so are ignored in this mode. Also, an unescaped # out-
2304: side a character class indicates a comment that lasts until after the
2305: next line break sequence. In other circumstances, line break sequences
2306: in patterns are treated as literal data.
2307:
2308: The newline option that is set at compile time becomes the default that
2309: is used for pcre_exec() and pcre_dfa_exec(), but it can be overridden.
2310:
2311: PCRE_NO_AUTO_CAPTURE
2312:
2313: If this option is set, it disables the use of numbered capturing paren-
2314: theses in the pattern. Any opening parenthesis that is not followed by
2315: ? behaves as if it were followed by ?: but named parentheses can still
2316: be used for capturing (and they acquire numbers in the usual way).
2317: There is no equivalent of this option in Perl.
2318:
2319: PCRE_NO_AUTO_POSSESS
2320:
2321: If this option is set, it disables "auto-possessification". This is an
2322: optimization that, for example, turns a+b into a++b in order to avoid
2323: backtracks into a+ that can never be successful. However, if callouts
2324: are in use, auto-possessification means that some of them are never
2325: taken. You can set this option if you want the matching functions to do
2326: a full unoptimized search and run all the callouts, but it is mainly
2327: provided for testing purposes.
2328:
2329: PCRE_NO_START_OPTIMIZE
2330:
2331: This is an option that acts at matching time; that is, it is really an
2332: option for pcre_exec() or pcre_dfa_exec(). If it is set at compile
2333: time, it is remembered with the compiled pattern and assumed at match-
2334: ing time. This is necessary if you want to use JIT execution, because
2335: the JIT compiler needs to know whether or not this option is set. For
2336: details see the discussion of PCRE_NO_START_OPTIMIZE below.
2337:
2338: PCRE_UCP
2339:
2340: This option changes the way PCRE processes \B, \b, \D, \d, \S, \s, \W,
2341: \w, and some of the POSIX character classes. By default, only ASCII
2342: characters are recognized, but if PCRE_UCP is set, Unicode properties
2343: are used instead to classify characters. More details are given in the
2344: section on generic character types in the pcrepattern page. If you set
2345: PCRE_UCP, matching one of the items it affects takes much longer. The
2346: option is available only if PCRE has been compiled with Unicode prop-
2347: erty support.
2348:
2349: PCRE_UNGREEDY
2350:
2351: This option inverts the "greediness" of the quantifiers so that they
2352: are not greedy by default, but become greedy if followed by "?". It is
2353: not compatible with Perl. It can also be set by a (?U) option setting
2354: within the pattern.
2355:
2356: PCRE_UTF8
2357:
2358: This option causes PCRE to regard both the pattern and the subject as
2359: strings of UTF-8 characters instead of single-byte strings. However, it
2360: is available only when PCRE is built to include UTF support. If not,
2361: the use of this option provokes an error. Details of how this option
2362: changes the behaviour of PCRE are given in the pcreunicode page.
2363:
2364: PCRE_NO_UTF8_CHECK
2365:
2366: When PCRE_UTF8 is set, the validity of the pattern as a UTF-8 string is
2367: automatically checked. There is a discussion about the validity of
2368: UTF-8 strings in the pcreunicode page. If an invalid UTF-8 sequence is
2369: found, pcre_compile() returns an error. If you already know that your
2370: pattern is valid, and you want to skip this check for performance rea-
2371: sons, you can set the PCRE_NO_UTF8_CHECK option. When it is set, the
2372: effect of passing an invalid UTF-8 string as a pattern is undefined. It
2373: may cause your program to crash or loop. Note that this option can also
2374: be passed to pcre_exec() and pcre_dfa_exec(), to suppress the validity
2375: checking of subject strings only. If the same string is being matched
2376: many times, the option can be safely set for the second and subsequent
2377: matchings to improve performance.
2378:
2379:
2380: COMPILATION ERROR CODES
2381:
2382: The following table lists the error codes than may be returned by
2383: pcre_compile2(), along with the error messages that may be returned by
2384: both compiling functions. Note that error messages are always 8-bit
2385: ASCII strings, even in 16-bit or 32-bit mode. As PCRE has developed,
2386: some error codes have fallen out of use. To avoid confusion, they have
2387: not been re-used.
2388:
2389: 0 no error
2390: 1 \ at end of pattern
2391: 2 \c at end of pattern
2392: 3 unrecognized character follows \
2393: 4 numbers out of order in {} quantifier
2394: 5 number too big in {} quantifier
2395: 6 missing terminating ] for character class
2396: 7 invalid escape sequence in character class
2397: 8 range out of order in character class
2398: 9 nothing to repeat
2399: 10 [this code is not in use]
2400: 11 internal error: unexpected repeat
2401: 12 unrecognized character after (? or (?-
2402: 13 POSIX named classes are supported only within a class
2403: 14 missing )
2404: 15 reference to non-existent subpattern
2405: 16 erroffset passed as NULL
2406: 17 unknown option bit(s) set
2407: 18 missing ) after comment
2408: 19 [this code is not in use]
2409: 20 regular expression is too large
2410: 21 failed to get memory
2411: 22 unmatched parentheses
2412: 23 internal error: code overflow
2413: 24 unrecognized character after (?<
2414: 25 lookbehind assertion is not fixed length
2415: 26 malformed number or name after (?(
2416: 27 conditional group contains more than two branches
2417: 28 assertion expected after (?(
2418: 29 (?R or (?[+-]digits must be followed by )
2419: 30 unknown POSIX class name
2420: 31 POSIX collating elements are not supported
2421: 32 this version of PCRE is compiled without UTF support
2422: 33 [this code is not in use]
2423: 34 character value in \x{} or \o{} is too large
2424: 35 invalid condition (?(0)
2425: 36 \C not allowed in lookbehind assertion
2426: 37 PCRE does not support \L, \l, \N{name}, \U, or \u
2427: 38 number after (?C is > 255
2428: 39 closing ) for (?C expected
2429: 40 recursive call could loop indefinitely
2430: 41 unrecognized character after (?P
2431: 42 syntax error in subpattern name (missing terminator)
2432: 43 two named subpatterns have the same name
2433: 44 invalid UTF-8 string (specifically UTF-8)
2434: 45 support for \P, \p, and \X has not been compiled
2435: 46 malformed \P or \p sequence
2436: 47 unknown property name after \P or \p
2437: 48 subpattern name is too long (maximum 32 characters)
2438: 49 too many named subpatterns (maximum 10000)
2439: 50 [this code is not in use]
2440: 51 octal value is greater than \377 in 8-bit non-UTF-8 mode
2441: 52 internal error: overran compiling workspace
2442: 53 internal error: previously-checked referenced subpattern
2443: not found
2444: 54 DEFINE group contains more than one branch
2445: 55 repeating a DEFINE group is not allowed
2446: 56 inconsistent NEWLINE options
2447: 57 \g is not followed by a braced, angle-bracketed, or quoted
2448: name/number or by a plain number
2449: 58 a numbered reference must not be zero
2450: 59 an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)
2451: 60 (*VERB) not recognized or malformed
2452: 61 number is too big
2453: 62 subpattern name expected
2454: 63 digit expected after (?+
2455: 64 ] is an invalid data character in JavaScript compatibility mode
2456: 65 different names for subpatterns of the same number are
2457: not allowed
2458: 66 (*MARK) must have an argument
2459: 67 this version of PCRE is not compiled with Unicode property
2460: support
2461: 68 \c must be followed by an ASCII character
2462: 69 \k is not followed by a braced, angle-bracketed, or quoted name
2463: 70 internal error: unknown opcode in find_fixedlength()
2464: 71 \N is not supported in a class
2465: 72 too many forward references
2466: 73 disallowed Unicode code point (>= 0xd800 && <= 0xdfff)
2467: 74 invalid UTF-16 string (specifically UTF-16)
2468: 75 name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)
2469: 76 character value in \u.... sequence is too large
2470: 77 invalid UTF-32 string (specifically UTF-32)
2471: 78 setting UTF is disabled by the application
2472: 79 non-hex character in \x{} (closing brace missing?)
2473: 80 non-octal character in \o{} (closing brace missing?)
2474: 81 missing opening brace after \o
2475: 82 parentheses are too deeply nested
2476: 83 invalid range in character class
2477:
2478: The numbers 32 and 10000 in errors 48 and 49 are defaults; different
2479: values may be used if the limits were changed when PCRE was built.
2480:
2481:
2482: STUDYING A PATTERN
2483:
2484: pcre_extra *pcre_study(const pcre *code, int options,
2485: const char **errptr);
2486:
2487: If a compiled pattern is going to be used several times, it is worth
2488: spending more time analyzing it in order to speed up the time taken for
2489: matching. The function pcre_study() takes a pointer to a compiled pat-
2490: tern as its first argument. If studying the pattern produces additional
2491: information that will help speed up matching, pcre_study() returns a
2492: pointer to a pcre_extra block, in which the study_data field points to
2493: the results of the study.
2494:
2495: The returned value from pcre_study() can be passed directly to
2496: pcre_exec() or pcre_dfa_exec(). However, a pcre_extra block also con-
2497: tains other fields that can be set by the caller before the block is
2498: passed; these are described below in the section on matching a pattern.
2499:
2500: If studying the pattern does not produce any useful information,
2501: pcre_study() returns NULL by default. In that circumstance, if the
2502: calling program wants to pass any of the other fields to pcre_exec() or
2503: pcre_dfa_exec(), it must set up its own pcre_extra block. However, if
2504: pcre_study() is called with the PCRE_STUDY_EXTRA_NEEDED option, it
2505: returns a pcre_extra block even if studying did not find any additional
2506: information. It may still return NULL, however, if an error occurs in
2507: pcre_study().
2508:
2509: The second argument of pcre_study() contains option bits. There are
2510: three further options in addition to PCRE_STUDY_EXTRA_NEEDED:
2511:
2512: PCRE_STUDY_JIT_COMPILE
2513: PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE
2514: PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE
2515:
2516: If any of these are set, and the just-in-time compiler is available,
2517: the pattern is further compiled into machine code that executes much
2518: faster than the pcre_exec() interpretive matching function. If the
2519: just-in-time compiler is not available, these options are ignored. All
2520: undefined bits in the options argument must be zero.
2521:
2522: JIT compilation is a heavyweight optimization. It can take some time
2523: for patterns to be analyzed, and for one-off matches and simple pat-
2524: terns the benefit of faster execution might be offset by a much slower
2525: study time. Not all patterns can be optimized by the JIT compiler. For
2526: those that cannot be handled, matching automatically falls back to the
2527: pcre_exec() interpreter. For more details, see the pcrejit documenta-
2528: tion.
2529:
2530: The third argument for pcre_study() is a pointer for an error message.
2531: If studying succeeds (even if no data is returned), the variable it
2532: points to is set to NULL. Otherwise it is set to point to a textual
2533: error message. This is a static string that is part of the library. You
2534: must not try to free it. You should test the error pointer for NULL
2535: after calling pcre_study(), to be sure that it has run successfully.
2536:
2537: When you are finished with a pattern, you can free the memory used for
2538: the study data by calling pcre_free_study(). This function was added to
2539: the API for release 8.20. For earlier versions, the memory could be
2540: freed with pcre_free(), just like the pattern itself. This will still
2541: work in cases where JIT optimization is not used, but it is advisable
2542: to change to the new function when convenient.
2543:
2544: This is a typical way in which pcre_study() is used (except that in a
2545: real application there should be tests for errors):
2546:
2547: int rc;
2548: pcre *re;
2549: pcre_extra *sd;
2550: re = pcre_compile("pattern", 0, &error, &erroroffset, NULL);
2551: sd = pcre_study(
2552: re, /* result of pcre_compile() */
2553: 0, /* no options */
2554: &error); /* set to NULL or points to a message */
2555: rc = pcre_exec( /* see below for details of pcre_exec() options */
2556: re, sd, "subject", 7, 0, 0, ovector, 30);
2557: ...
2558: pcre_free_study(sd);
2559: pcre_free(re);
2560:
2561: Studying a pattern does two things: first, a lower bound for the length
2562: of subject string that is needed to match the pattern is computed. This
2563: does not mean that there are any strings of that length that match, but
2564: it does guarantee that no shorter strings match. The value is used to
2565: avoid wasting time by trying to match strings that are shorter than the
2566: lower bound. You can find out the value in a calling program via the
2567: pcre_fullinfo() function.
2568:
2569: Studying a pattern is also useful for non-anchored patterns that do not
2570: have a single fixed starting character. A bitmap of possible starting
2571: bytes is created. This speeds up finding a position in the subject at
2572: which to start matching. (In 16-bit mode, the bitmap is used for 16-bit
2573: values less than 256. In 32-bit mode, the bitmap is used for 32-bit
2574: values less than 256.)
2575:
2576: These two optimizations apply to both pcre_exec() and pcre_dfa_exec(),
2577: and the information is also used by the JIT compiler. The optimiza-
2578: tions can be disabled by setting the PCRE_NO_START_OPTIMIZE option.
2579: You might want to do this if your pattern contains callouts or (*MARK)
2580: and you want to make use of these facilities in cases where matching
2581: fails.
2582:
2583: PCRE_NO_START_OPTIMIZE can be specified at either compile time or exe-
2584: cution time. However, if PCRE_NO_START_OPTIMIZE is passed to
2585: pcre_exec(), (that is, after any JIT compilation has happened) JIT exe-
2586: cution is disabled. For JIT execution to work with PCRE_NO_START_OPTI-
2587: MIZE, the option must be set at compile time.
2588:
2589: There is a longer discussion of PCRE_NO_START_OPTIMIZE below.
2590:
2591:
2592: LOCALE SUPPORT
2593:
2594: PCRE handles caseless matching, and determines whether characters are
2595: letters, digits, or whatever, by reference to a set of tables, indexed
2596: by character code point. When running in UTF-8 mode, or in the 16- or
2597: 32-bit libraries, this applies only to characters with code points less
2598: than 256. By default, higher-valued code points never match escapes
2599: such as \w or \d. However, if PCRE is built with Unicode property sup-
2600: port, all characters can be tested with \p and \P, or, alternatively,
2601: the PCRE_UCP option can be set when a pattern is compiled; this causes
2602: \w and friends to use Unicode property support instead of the built-in
2603: tables.
2604:
2605: The use of locales with Unicode is discouraged. If you are handling
2606: characters with code points greater than 128, you should either use
2607: Unicode support, or use locales, but not try to mix the two.
2608:
2609: PCRE contains an internal set of tables that are used when the final
2610: argument of pcre_compile() is NULL. These are sufficient for many
2611: applications. Normally, the internal tables recognize only ASCII char-
2612: acters. However, when PCRE is built, it is possible to cause the inter-
2613: nal tables to be rebuilt in the default "C" locale of the local system,
2614: which may cause them to be different.
2615:
2616: The internal tables can always be overridden by tables supplied by the
2617: application that calls PCRE. These may be created in a different locale
2618: from the default. As more and more applications change to using Uni-
2619: code, the need for this locale support is expected to die away.
2620:
2621: External tables are built by calling the pcre_maketables() function,
2622: which has no arguments, in the relevant locale. The result can then be
2623: passed to pcre_compile() as often as necessary. For example, to build
2624: and use tables that are appropriate for the French locale (where
2625: accented characters with values greater than 128 are treated as let-
2626: ters), the following code could be used:
2627:
2628: setlocale(LC_CTYPE, "fr_FR");
2629: tables = pcre_maketables();
2630: re = pcre_compile(..., tables);
2631:
2632: The locale name "fr_FR" is used on Linux and other Unix-like systems;
2633: if you are using Windows, the name for the French locale is "french".
2634:
2635: When pcre_maketables() runs, the tables are built in memory that is
2636: obtained via pcre_malloc. It is the caller's responsibility to ensure
2637: that the memory containing the tables remains available for as long as
2638: it is needed.
2639:
2640: The pointer that is passed to pcre_compile() is saved with the compiled
2641: pattern, and the same tables are used via this pointer by pcre_study()
2642: and also by pcre_exec() and pcre_dfa_exec(). Thus, for any single pat-
2643: tern, compilation, studying and matching all happen in the same locale,
2644: but different patterns can be processed in different locales.
2645:
2646: It is possible to pass a table pointer or NULL (indicating the use of
2647: the internal tables) to pcre_exec() or pcre_dfa_exec() (see the discus-
2648: sion below in the section on matching a pattern). This facility is pro-
2649: vided for use with pre-compiled patterns that have been saved and
2650: reloaded. Character tables are not saved with patterns, so if a non-
2651: standard table was used at compile time, it must be provided again when
2652: the reloaded pattern is matched. Attempting to use this facility to
2653: match a pattern in a different locale from the one in which it was com-
2654: piled is likely to lead to anomalous (usually incorrect) results.
2655:
2656:
2657: INFORMATION ABOUT A PATTERN
2658:
2659: int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
2660: int what, void *where);
2661:
2662: The pcre_fullinfo() function returns information about a compiled pat-
2663: tern. It replaces the pcre_info() function, which was removed from the
2664: library at version 8.30, after more than 10 years of obsolescence.
2665:
2666: The first argument for pcre_fullinfo() is a pointer to the compiled
2667: pattern. The second argument is the result of pcre_study(), or NULL if
2668: the pattern was not studied. The third argument specifies which piece
2669: of information is required, and the fourth argument is a pointer to a
2670: variable to receive the data. The yield of the function is zero for
2671: success, or one of the following negative numbers:
2672:
2673: PCRE_ERROR_NULL the argument code was NULL
2674: the argument where was NULL
2675: PCRE_ERROR_BADMAGIC the "magic number" was not found
2676: PCRE_ERROR_BADENDIANNESS the pattern was compiled with different
2677: endianness
2678: PCRE_ERROR_BADOPTION the value of what was invalid
2679: PCRE_ERROR_UNSET the requested field is not set
2680:
2681: The "magic number" is placed at the start of each compiled pattern as
2682: an simple check against passing an arbitrary memory pointer. The endi-
2683: anness error can occur if a compiled pattern is saved and reloaded on a
2684: different host. Here is a typical call of pcre_fullinfo(), to obtain
2685: the length of the compiled pattern:
2686:
2687: int rc;
2688: size_t length;
2689: rc = pcre_fullinfo(
2690: re, /* result of pcre_compile() */
2691: sd, /* result of pcre_study(), or NULL */
2692: PCRE_INFO_SIZE, /* what is required */
2693: &length); /* where to put the data */
2694:
2695: The possible values for the third argument are defined in pcre.h, and
2696: are as follows:
2697:
2698: PCRE_INFO_BACKREFMAX
2699:
2700: Return the number of the highest back reference in the pattern. The
2701: fourth argument should point to an int variable. Zero is returned if
2702: there are no back references.
2703:
2704: PCRE_INFO_CAPTURECOUNT
2705:
2706: Return the number of capturing subpatterns in the pattern. The fourth
2707: argument should point to an int variable.
2708:
2709: PCRE_INFO_DEFAULT_TABLES
2710:
2711: Return a pointer to the internal default character tables within PCRE.
2712: The fourth argument should point to an unsigned char * variable. This
2713: information call is provided for internal use by the pcre_study() func-
2714: tion. External callers can cause PCRE to use its internal tables by
2715: passing a NULL table pointer.
2716:
2717: PCRE_INFO_FIRSTBYTE
2718:
2719: Return information about the first data unit of any matched string, for
2720: a non-anchored pattern. (The name of this option refers to the 8-bit
2721: library, where data units are bytes.) The fourth argument should point
2722: to an int variable.
2723:
2724: If there is a fixed first value, for example, the letter "c" from a
2725: pattern such as (cat|cow|coyote), its value is returned. In the 8-bit
2726: library, the value is always less than 256. In the 16-bit library the
2727: value can be up to 0xffff. In the 32-bit library the value can be up to
2728: 0x10ffff.
2729:
2730: If there is no fixed first value, and if either
2731:
2732: (a) the pattern was compiled with the PCRE_MULTILINE option, and every
2733: branch starts with "^", or
2734:
2735: (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
2736: set (if it were set, the pattern would be anchored),
2737:
2738: -1 is returned, indicating that the pattern matches only at the start
2739: of a subject string or after any newline within the string. Otherwise
2740: -2 is returned. For anchored patterns, -2 is returned.
2741:
2742: Since for the 32-bit library using the non-UTF-32 mode, this function
2743: is unable to return the full 32-bit range of the character, this value
2744: is deprecated; instead the PCRE_INFO_FIRSTCHARACTERFLAGS and
2745: PCRE_INFO_FIRSTCHARACTER values should be used.
2746:
2747: PCRE_INFO_FIRSTTABLE
2748:
2749: If the pattern was studied, and this resulted in the construction of a
2750: 256-bit table indicating a fixed set of values for the first data unit
2751: in any matching string, a pointer to the table is returned. Otherwise
2752: NULL is returned. The fourth argument should point to an unsigned char
2753: * variable.
2754:
2755: PCRE_INFO_HASCRORLF
2756:
2757: Return 1 if the pattern contains any explicit matches for CR or LF
2758: characters, otherwise 0. The fourth argument should point to an int
2759: variable. An explicit match is either a literal CR or LF character, or
2760: \r or \n.
2761:
2762: PCRE_INFO_JCHANGED
2763:
2764: Return 1 if the (?J) or (?-J) option setting is used in the pattern,
2765: otherwise 0. The fourth argument should point to an int variable. (?J)
2766: and (?-J) set and unset the local PCRE_DUPNAMES option, respectively.
2767:
2768: PCRE_INFO_JIT
2769:
2770: Return 1 if the pattern was studied with one of the JIT options, and
2771: just-in-time compiling was successful. The fourth argument should point
2772: to an int variable. A return value of 0 means that JIT support is not
2773: available in this version of PCRE, or that the pattern was not studied
2774: with a JIT option, or that the JIT compiler could not handle this par-
2775: ticular pattern. See the pcrejit documentation for details of what can
2776: and cannot be handled.
2777:
2778: PCRE_INFO_JITSIZE
2779:
2780: If the pattern was successfully studied with a JIT option, return the
2781: size of the JIT compiled code, otherwise return zero. The fourth argu-
2782: ment should point to a size_t variable.
2783:
2784: PCRE_INFO_LASTLITERAL
2785:
2786: Return the value of the rightmost literal data unit that must exist in
2787: any matched string, other than at its start, if such a value has been
2788: recorded. The fourth argument should point to an int variable. If there
2789: is no such value, -1 is returned. For anchored patterns, a last literal
2790: value is recorded only if it follows something of variable length. For
2791: example, for the pattern /^a\d+z\d+/ the returned value is "z", but for
2792: /^a\dz\d/ the returned value is -1.
2793:
2794: Since for the 32-bit library using the non-UTF-32 mode, this function
2795: is unable to return the full 32-bit range of characters, this value is
2796: deprecated; instead the PCRE_INFO_REQUIREDCHARFLAGS and
2797: PCRE_INFO_REQUIREDCHAR values should be used.
2798:
2799: PCRE_INFO_MATCH_EMPTY
2800:
2801: Return 1 if the pattern can match an empty string, otherwise 0. The
2802: fourth argument should point to an int variable.
2803:
2804: PCRE_INFO_MATCHLIMIT
2805:
2806: If the pattern set a match limit by including an item of the form
2807: (*LIMIT_MATCH=nnnn) at the start, the value is returned. The fourth
2808: argument should point to an unsigned 32-bit integer. If no such value
2809: has been set, the call to pcre_fullinfo() returns the error
2810: PCRE_ERROR_UNSET.
2811:
2812: PCRE_INFO_MAXLOOKBEHIND
2813:
2814: Return the number of characters (NB not data units) in the longest
2815: lookbehind assertion in the pattern. This information is useful when
2816: doing multi-segment matching using the partial matching facilities.
2817: Note that the simple assertions \b and \B require a one-character look-
2818: behind. \A also registers a one-character lookbehind, though it does
2819: not actually inspect the previous character. This is to ensure that at
2820: least one character from the old segment is retained when a new segment
2821: is processed. Otherwise, if there are no lookbehinds in the pattern, \A
2822: might match incorrectly at the start of a new segment.
2823:
2824: PCRE_INFO_MINLENGTH
2825:
2826: If the pattern was studied and a minimum length for matching subject
2827: strings was computed, its value is returned. Otherwise the returned
2828: value is -1. The value is a number of characters, which in UTF mode may
2829: be different from the number of data units. The fourth argument should
2830: point to an int variable. A non-negative value is a lower bound to the
2831: length of any matching string. There may not be any strings of that
2832: length that do actually match, but every string that does match is at
2833: least that long.
2834:
2835: PCRE_INFO_NAMECOUNT
2836: PCRE_INFO_NAMEENTRYSIZE
2837: PCRE_INFO_NAMETABLE
2838:
2839: PCRE supports the use of named as well as numbered capturing parenthe-
2840: ses. The names are just an additional way of identifying the parenthe-
2841: ses, which still acquire numbers. Several convenience functions such as
2842: pcre_get_named_substring() are provided for extracting captured sub-
2843: strings by name. It is also possible to extract the data directly, by
2844: first converting the name to a number in order to access the correct
2845: pointers in the output vector (described with pcre_exec() below). To do
2846: the conversion, you need to use the name-to-number map, which is
2847: described by these three values.
2848:
2849: The map consists of a number of fixed-size entries. PCRE_INFO_NAMECOUNT
2850: gives the number of entries, and PCRE_INFO_NAMEENTRYSIZE gives the size
2851: of each entry; both of these return an int value. The entry size
2852: depends on the length of the longest name. PCRE_INFO_NAMETABLE returns
2853: a pointer to the first entry of the table. This is a pointer to char in
2854: the 8-bit library, where the first two bytes of each entry are the num-
2855: ber of the capturing parenthesis, most significant byte first. In the
2856: 16-bit library, the pointer points to 16-bit data units, the first of
2857: which contains the parenthesis number. In the 32-bit library, the
2858: pointer points to 32-bit data units, the first of which contains the
2859: parenthesis number. The rest of the entry is the corresponding name,
2860: zero terminated.
2861:
2862: The names are in alphabetical order. If (?| is used to create multiple
2863: groups with the same number, as described in the section on duplicate
2864: subpattern numbers in the pcrepattern page, the groups may be given the
2865: same name, but there is only one entry in the table. Different names
2866: for groups of the same number are not permitted. Duplicate names for
2867: subpatterns with different numbers are permitted, but only if PCRE_DUP-
2868: NAMES is set. They appear in the table in the order in which they were
2869: found in the pattern. In the absence of (?| this is the order of
2870: increasing number; when (?| is used this is not necessarily the case
2871: because later subpatterns may have lower numbers.
2872:
2873: As a simple example of the name/number table, consider the following
2874: pattern after compilation by the 8-bit library (assume PCRE_EXTENDED is
2875: set, so white space - including newlines - is ignored):
2876:
2877: (?<date> (?<year>(\d\d)?\d\d) -
2878: (?<month>\d\d) - (?<day>\d\d) )
2879:
2880: There are four named subpatterns, so the table has four entries, and
2881: each entry in the table is eight bytes long. The table is as follows,
2882: with non-printing bytes shows in hexadecimal, and undefined bytes shown
2883: as ??:
2884:
2885: 00 01 d a t e 00 ??
2886: 00 05 d a y 00 ?? ??
2887: 00 04 m o n t h 00
2888: 00 02 y e a r 00 ??
2889:
2890: When writing code to extract data from named subpatterns using the
2891: name-to-number map, remember that the length of the entries is likely
2892: to be different for each compiled pattern.
2893:
2894: PCRE_INFO_OKPARTIAL
2895:
2896: Return 1 if the pattern can be used for partial matching with
2897: pcre_exec(), otherwise 0. The fourth argument should point to an int
2898: variable. From release 8.00, this always returns 1, because the
2899: restrictions that previously applied to partial matching have been
2900: lifted. The pcrepartial documentation gives details of partial match-
2901: ing.
2902:
2903: PCRE_INFO_OPTIONS
2904:
2905: Return a copy of the options with which the pattern was compiled. The
2906: fourth argument should point to an unsigned long int variable. These
2907: option bits are those specified in the call to pcre_compile(), modified
2908: by any top-level option settings at the start of the pattern itself. In
2909: other words, they are the options that will be in force when matching
2910: starts. For example, if the pattern /(?im)abc(?-i)d/ is compiled with
2911: the PCRE_EXTENDED option, the result is PCRE_CASELESS, PCRE_MULTILINE,
2912: and PCRE_EXTENDED.
2913:
2914: A pattern is automatically anchored by PCRE if all of its top-level
2915: alternatives begin with one of the following:
2916:
2917: ^ unless PCRE_MULTILINE is set
2918: \A always
2919: \G always
2920: .* if PCRE_DOTALL is set and there are no back
2921: references to the subpattern in which .* appears
2922:
2923: For such patterns, the PCRE_ANCHORED bit is set in the options returned
2924: by pcre_fullinfo().
2925:
2926: PCRE_INFO_RECURSIONLIMIT
2927:
2928: If the pattern set a recursion limit by including an item of the form
2929: (*LIMIT_RECURSION=nnnn) at the start, the value is returned. The fourth
2930: argument should point to an unsigned 32-bit integer. If no such value
2931: has been set, the call to pcre_fullinfo() returns the error
2932: PCRE_ERROR_UNSET.
2933:
2934: PCRE_INFO_SIZE
2935:
2936: Return the size of the compiled pattern in bytes (for all three
2937: libraries). The fourth argument should point to a size_t variable. This
2938: value does not include the size of the pcre structure that is returned
2939: by pcre_compile(). The value that is passed as the argument to
2940: pcre_malloc() when pcre_compile() is getting memory in which to place
2941: the compiled data is the value returned by this option plus the size of
2942: the pcre structure. Studying a compiled pattern, with or without JIT,
2943: does not alter the value returned by this option.
2944:
2945: PCRE_INFO_STUDYSIZE
2946:
2947: Return the size in bytes (for all three libraries) of the data block
2948: pointed to by the study_data field in a pcre_extra block. If pcre_extra
2949: is NULL, or there is no study data, zero is returned. The fourth argu-
2950: ment should point to a size_t variable. The study_data field is set by
2951: pcre_study() to record information that will speed up matching (see the
2952: section entitled "Studying a pattern" above). The format of the
2953: study_data block is private, but its length is made available via this
2954: option so that it can be saved and restored (see the pcreprecompile
2955: documentation for details).
2956:
2957: PCRE_INFO_FIRSTCHARACTERFLAGS
2958:
2959: Return information about the first data unit of any matched string, for
2960: a non-anchored pattern. The fourth argument should point to an int
2961: variable.
2962:
2963: If there is a fixed first value, for example, the letter "c" from a
2964: pattern such as (cat|cow|coyote), 1 is returned, and the character
2965: value can be retrieved using PCRE_INFO_FIRSTCHARACTER.
2966:
2967: If there is no fixed first value, and if either
2968:
2969: (a) the pattern was compiled with the PCRE_MULTILINE option, and every
2970: branch starts with "^", or
2971:
2972: (b) every branch of the pattern starts with ".*" and PCRE_DOTALL is not
2973: set (if it were set, the pattern would be anchored),
2974:
2975: 2 is returned, indicating that the pattern matches only at the start of
2976: a subject string or after any newline within the string. Otherwise 0 is
2977: returned. For anchored patterns, 0 is returned.
2978:
2979: PCRE_INFO_FIRSTCHARACTER
2980:
2981: Return the fixed first character value in the situation where
2982: PCRE_INFO_FIRSTCHARACTERFLAGS returns 1; otherwise return 0. The fourth
2983: argument should point to an uint_t variable.
2984:
2985: In the 8-bit library, the value is always less than 256. In the 16-bit
2986: library the value can be up to 0xffff. In the 32-bit library in UTF-32
2987: mode the value can be up to 0x10ffff, and up to 0xffffffff when not
2988: using UTF-32 mode.
2989:
2990: PCRE_INFO_REQUIREDCHARFLAGS
2991:
2992: Returns 1 if there is a rightmost literal data unit that must exist in
2993: any matched string, other than at its start. The fourth argument should
2994: point to an int variable. If there is no such value, 0 is returned. If
2995: returning 1, the character value itself can be retrieved using
2996: PCRE_INFO_REQUIREDCHAR.
2997:
2998: For anchored patterns, a last literal value is recorded only if it fol-
2999: lows something of variable length. For example, for the pattern
3000: /^a\d+z\d+/ the returned value 1 (with "z" returned from
3001: PCRE_INFO_REQUIREDCHAR), but for /^a\dz\d/ the returned value is 0.
3002:
3003: PCRE_INFO_REQUIREDCHAR
3004:
3005: Return the value of the rightmost literal data unit that must exist in
3006: any matched string, other than at its start, if such a value has been
3007: recorded. The fourth argument should point to an uint32_t variable. If
3008: there is no such value, 0 is returned.
3009:
3010:
3011: REFERENCE COUNTS
3012:
3013: int pcre_refcount(pcre *code, int adjust);
3014:
3015: The pcre_refcount() function is used to maintain a reference count in
3016: the data block that contains a compiled pattern. It is provided for the
3017: benefit of applications that operate in an object-oriented manner,
3018: where different parts of the application may be using the same compiled
3019: pattern, but you want to free the block when they are all done.
3020:
3021: When a pattern is compiled, the reference count field is initialized to
3022: zero. It is changed only by calling this function, whose action is to
3023: add the adjust value (which may be positive or negative) to it. The
3024: yield of the function is the new value. However, the value of the count
3025: is constrained to lie between 0 and 65535, inclusive. If the new value
3026: is outside these limits, it is forced to the appropriate limit value.
3027:
3028: Except when it is zero, the reference count is not correctly preserved
3029: if a pattern is compiled on one host and then transferred to a host
3030: whose byte-order is different. (This seems a highly unlikely scenario.)
3031:
3032:
3033: MATCHING A PATTERN: THE TRADITIONAL FUNCTION
3034:
3035: int pcre_exec(const pcre *code, const pcre_extra *extra,
3036: const char *subject, int length, int startoffset,
3037: int options, int *ovector, int ovecsize);
3038:
3039: The function pcre_exec() is called to match a subject string against a
3040: compiled pattern, which is passed in the code argument. If the pattern
3041: was studied, the result of the study should be passed in the extra
3042: argument. You can call pcre_exec() with the same code and extra argu-
3043: ments as many times as you like, in order to match different subject
3044: strings with the same pattern.
3045:
3046: This function is the main matching facility of the library, and it
3047: operates in a Perl-like manner. For specialist use there is also an
3048: alternative matching function, which is described below in the section
3049: about the pcre_dfa_exec() function.
3050:
3051: In most applications, the pattern will have been compiled (and option-
3052: ally studied) in the same process that calls pcre_exec(). However, it
3053: is possible to save compiled patterns and study data, and then use them
3054: later in different processes, possibly even on different hosts. For a
3055: discussion about this, see the pcreprecompile documentation.
3056:
3057: Here is an example of a simple call to pcre_exec():
3058:
3059: int rc;
3060: int ovector[30];
3061: rc = pcre_exec(
3062: re, /* result of pcre_compile() */
3063: NULL, /* we didn't study the pattern */
3064: "some string", /* the subject string */
3065: 11, /* the length of the subject string */
3066: 0, /* start at offset 0 in the subject */
3067: 0, /* default options */
3068: ovector, /* vector of integers for substring information */
3069: 30); /* number of elements (NOT size in bytes) */
3070:
3071: Extra data for pcre_exec()
3072:
3073: If the extra argument is not NULL, it must point to a pcre_extra data
3074: block. The pcre_study() function returns such a block (when it doesn't
3075: return NULL), but you can also create one for yourself, and pass addi-
3076: tional information in it. The pcre_extra block contains the following
3077: fields (not necessarily in this order):
3078:
3079: unsigned long int flags;
3080: void *study_data;
3081: void *executable_jit;
3082: unsigned long int match_limit;
3083: unsigned long int match_limit_recursion;
3084: void *callout_data;
3085: const unsigned char *tables;
3086: unsigned char **mark;
3087:
3088: In the 16-bit version of this structure, the mark field has type
3089: "PCRE_UCHAR16 **".
3090:
3091: In the 32-bit version of this structure, the mark field has type
3092: "PCRE_UCHAR32 **".
3093:
3094: The flags field is used to specify which of the other fields are set.
3095: The flag bits are:
3096:
3097: PCRE_EXTRA_CALLOUT_DATA
3098: PCRE_EXTRA_EXECUTABLE_JIT
3099: PCRE_EXTRA_MARK
3100: PCRE_EXTRA_MATCH_LIMIT
3101: PCRE_EXTRA_MATCH_LIMIT_RECURSION
3102: PCRE_EXTRA_STUDY_DATA
3103: PCRE_EXTRA_TABLES
3104:
3105: Other flag bits should be set to zero. The study_data field and some-
3106: times the executable_jit field are set in the pcre_extra block that is
3107: returned by pcre_study(), together with the appropriate flag bits. You
3108: should not set these yourself, but you may add to the block by setting
3109: other fields and their corresponding flag bits.
3110:
3111: The match_limit field provides a means of preventing PCRE from using up
3112: a vast amount of resources when running patterns that are not going to
3113: match, but which have a very large number of possibilities in their
3114: search trees. The classic example is a pattern that uses nested unlim-
3115: ited repeats.
3116:
3117: Internally, pcre_exec() uses a function called match(), which it calls
3118: repeatedly (sometimes recursively). The limit set by match_limit is
3119: imposed on the number of times this function is called during a match,
3120: which has the effect of limiting the amount of backtracking that can
3121: take place. For patterns that are not anchored, the count restarts from
3122: zero for each position in the subject string.
3123:
3124: When pcre_exec() is called with a pattern that was successfully studied
3125: with a JIT option, the way that the matching is executed is entirely
3126: different. However, there is still the possibility of runaway matching
3127: that goes on for a very long time, and so the match_limit value is also
3128: used in this case (but in a different way) to limit how long the match-
3129: ing can continue.
3130:
3131: The default value for the limit can be set when PCRE is built; the
3132: default default is 10 million, which handles all but the most extreme
3133: cases. You can override the default by suppling pcre_exec() with a
3134: pcre_extra block in which match_limit is set, and
3135: PCRE_EXTRA_MATCH_LIMIT is set in the flags field. If the limit is
3136: exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
3137:
3138: A value for the match limit may also be supplied by an item at the
3139: start of a pattern of the form
3140:
3141: (*LIMIT_MATCH=d)
3142:
3143: where d is a decimal number. However, such a setting is ignored unless
3144: d is less than the limit set by the caller of pcre_exec() or, if no
3145: such limit is set, less than the default.
3146:
3147: The match_limit_recursion field is similar to match_limit, but instead
3148: of limiting the total number of times that match() is called, it limits
3149: the depth of recursion. The recursion depth is a smaller number than
3150: the total number of calls, because not all calls to match() are recur-
3151: sive. This limit is of use only if it is set smaller than match_limit.
3152:
3153: Limiting the recursion depth limits the amount of machine stack that
3154: can be used, or, when PCRE has been compiled to use memory on the heap
3155: instead of the stack, the amount of heap memory that can be used. This
3156: limit is not relevant, and is ignored, when matching is done using JIT
3157: compiled code.
3158:
3159: The default value for match_limit_recursion can be set when PCRE is
3160: built; the default default is the same value as the default for
3161: match_limit. You can override the default by suppling pcre_exec() with
3162: a pcre_extra block in which match_limit_recursion is set, and
3163: PCRE_EXTRA_MATCH_LIMIT_RECURSION is set in the flags field. If the
3164: limit is exceeded, pcre_exec() returns PCRE_ERROR_RECURSIONLIMIT.
3165:
3166: A value for the recursion limit may also be supplied by an item at the
3167: start of a pattern of the form
3168:
3169: (*LIMIT_RECURSION=d)
3170:
3171: where d is a decimal number. However, such a setting is ignored unless
3172: d is less than the limit set by the caller of pcre_exec() or, if no
3173: such limit is set, less than the default.
3174:
3175: The callout_data field is used in conjunction with the "callout" fea-
3176: ture, and is described in the pcrecallout documentation.
3177:
3178: The tables field is provided for use with patterns that have been pre-
3179: compiled using custom character tables, saved to disc or elsewhere, and
3180: then reloaded, because the tables that were used to compile a pattern
3181: are not saved with it. See the pcreprecompile documentation for a dis-
3182: cussion of saving compiled patterns for later use. If NULL is passed
3183: using this mechanism, it forces PCRE's internal tables to be used.
3184:
3185: Warning: The tables that pcre_exec() uses must be the same as those
3186: that were used when the pattern was compiled. If this is not the case,
3187: the behaviour of pcre_exec() is undefined. Therefore, when a pattern is
3188: compiled and matched in the same process, this field should never be
3189: set. In this (the most common) case, the correct table pointer is auto-
3190: matically passed with the compiled pattern from pcre_compile() to
3191: pcre_exec().
3192:
3193: If PCRE_EXTRA_MARK is set in the flags field, the mark field must be
3194: set to point to a suitable variable. If the pattern contains any back-
3195: tracking control verbs such as (*MARK:NAME), and the execution ends up
3196: with a name to pass back, a pointer to the name string (zero termi-
3197: nated) is placed in the variable pointed to by the mark field. The
3198: names are within the compiled pattern; if you wish to retain such a
3199: name you must copy it before freeing the memory of a compiled pattern.
3200: If there is no name to pass back, the variable pointed to by the mark
3201: field is set to NULL. For details of the backtracking control verbs,
3202: see the section entitled "Backtracking control" in the pcrepattern doc-
3203: umentation.
3204:
3205: Option bits for pcre_exec()
3206:
3207: The unused bits of the options argument for pcre_exec() must be zero.
3208: The only bits that may be set are PCRE_ANCHORED, PCRE_NEWLINE_xxx,
3209: PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NOTEMPTY_ATSTART,
3210: PCRE_NO_START_OPTIMIZE, PCRE_NO_UTF8_CHECK, PCRE_PARTIAL_HARD, and
3211: PCRE_PARTIAL_SOFT.
3212:
3213: If the pattern was successfully studied with one of the just-in-time
3214: (JIT) compile options, the only supported options for JIT execution are
3215: PCRE_NO_UTF8_CHECK, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY,
3216: PCRE_NOTEMPTY_ATSTART, PCRE_PARTIAL_HARD, and PCRE_PARTIAL_SOFT. If an
3217: unsupported option is used, JIT execution is disabled and the normal
3218: interpretive code in pcre_exec() is run.
3219:
3220: PCRE_ANCHORED
3221:
3222: The PCRE_ANCHORED option limits pcre_exec() to matching at the first
3223: matching position. If a pattern was compiled with PCRE_ANCHORED, or
3224: turned out to be anchored by virtue of its contents, it cannot be made
3225: unachored at matching time.
3226:
3227: PCRE_BSR_ANYCRLF
3228: PCRE_BSR_UNICODE
3229:
3230: These options (which are mutually exclusive) control what the \R escape
3231: sequence matches. The choice is either to match only CR, LF, or CRLF,
3232: or to match any Unicode newline sequence. These options override the
3233: choice that was made or defaulted when the pattern was compiled.
3234:
3235: PCRE_NEWLINE_CR
3236: PCRE_NEWLINE_LF
3237: PCRE_NEWLINE_CRLF
3238: PCRE_NEWLINE_ANYCRLF
3239: PCRE_NEWLINE_ANY
3240:
3241: These options override the newline definition that was chosen or
3242: defaulted when the pattern was compiled. For details, see the descrip-
3243: tion of pcre_compile() above. During matching, the newline choice
3244: affects the behaviour of the dot, circumflex, and dollar metacharac-
3245: ters. It may also alter the way the match position is advanced after a
3246: match failure for an unanchored pattern.
3247:
3248: When PCRE_NEWLINE_CRLF, PCRE_NEWLINE_ANYCRLF, or PCRE_NEWLINE_ANY is
3249: set, and a match attempt for an unanchored pattern fails when the cur-
3250: rent position is at a CRLF sequence, and the pattern contains no
3251: explicit matches for CR or LF characters, the match position is
3252: advanced by two characters instead of one, in other words, to after the
3253: CRLF.
3254:
3255: The above rule is a compromise that makes the most common cases work as
3256: expected. For example, if the pattern is .+A (and the PCRE_DOTALL
3257: option is not set), it does not match the string "\r\nA" because, after
3258: failing at the start, it skips both the CR and the LF before retrying.
3259: However, the pattern [\r\n]A does match that string, because it con-
3260: tains an explicit CR or LF reference, and so advances only by one char-
3261: acter after the first failure.
3262:
3263: An explicit match for CR of LF is either a literal appearance of one of
3264: those characters, or one of the \r or \n escape sequences. Implicit
3265: matches such as [^X] do not count, nor does \s (which includes CR and
3266: LF in the characters that it matches).
3267:
3268: Notwithstanding the above, anomalous effects may still occur when CRLF
3269: is a valid newline sequence and explicit \r or \n escapes appear in the
3270: pattern.
3271:
3272: PCRE_NOTBOL
3273:
3274: This option specifies that first character of the subject string is not
3275: the beginning of a line, so the circumflex metacharacter should not
3276: match before it. Setting this without PCRE_MULTILINE (at compile time)
3277: causes circumflex never to match. This option affects only the behav-
3278: iour of the circumflex metacharacter. It does not affect \A.
3279:
3280: PCRE_NOTEOL
3281:
3282: This option specifies that the end of the subject string is not the end
3283: of a line, so the dollar metacharacter should not match it nor (except
3284: in multiline mode) a newline immediately before it. Setting this with-
3285: out PCRE_MULTILINE (at compile time) causes dollar never to match. This
3286: option affects only the behaviour of the dollar metacharacter. It does
3287: not affect \Z or \z.
3288:
3289: PCRE_NOTEMPTY
3290:
3291: An empty string is not considered to be a valid match if this option is
3292: set. If there are alternatives in the pattern, they are tried. If all
3293: the alternatives match the empty string, the entire match fails. For
3294: example, if the pattern
3295:
3296: a?b?
3297:
3298: is applied to a string not beginning with "a" or "b", it matches an
3299: empty string at the start of the subject. With PCRE_NOTEMPTY set, this
3300: match is not valid, so PCRE searches further into the string for occur-
3301: rences of "a" or "b".
3302:
3303: PCRE_NOTEMPTY_ATSTART
3304:
3305: This is like PCRE_NOTEMPTY, except that an empty string match that is
3306: not at the start of the subject is permitted. If the pattern is
3307: anchored, such a match can occur only if the pattern contains \K.
3308:
3309: Perl has no direct equivalent of PCRE_NOTEMPTY or
3310: PCRE_NOTEMPTY_ATSTART, but it does make a special case of a pattern
3311: match of the empty string within its split() function, and when using
3312: the /g modifier. It is possible to emulate Perl's behaviour after
3313: matching a null string by first trying the match again at the same off-
3314: set with PCRE_NOTEMPTY_ATSTART and PCRE_ANCHORED, and then if that
3315: fails, by advancing the starting offset (see below) and trying an ordi-
3316: nary match again. There is some code that demonstrates how to do this
3317: in the pcredemo sample program. In the most general case, you have to
3318: check to see if the newline convention recognizes CRLF as a newline,
3319: and if so, and the current character is CR followed by LF, advance the
3320: starting offset by two characters instead of one.
3321:
3322: PCRE_NO_START_OPTIMIZE
3323:
3324: There are a number of optimizations that pcre_exec() uses at the start
3325: of a match, in order to speed up the process. For example, if it is
3326: known that an unanchored match must start with a specific character, it
3327: searches the subject for that character, and fails immediately if it
3328: cannot find it, without actually running the main matching function.
3329: This means that a special item such as (*COMMIT) at the start of a pat-
3330: tern is not considered until after a suitable starting point for the
3331: match has been found. Also, when callouts or (*MARK) items are in use,
3332: these "start-up" optimizations can cause them to be skipped if the pat-
3333: tern is never actually used. The start-up optimizations are in effect a
3334: pre-scan of the subject that takes place before the pattern is run.
3335:
3336: The PCRE_NO_START_OPTIMIZE option disables the start-up optimizations,
3337: possibly causing performance to suffer, but ensuring that in cases
3338: where the result is "no match", the callouts do occur, and that items
3339: such as (*COMMIT) and (*MARK) are considered at every possible starting
3340: position in the subject string. If PCRE_NO_START_OPTIMIZE is set at
3341: compile time, it cannot be unset at matching time. The use of
3342: PCRE_NO_START_OPTIMIZE at matching time (that is, passing it to
3343: pcre_exec()) disables JIT execution; in this situation, matching is
3344: always done using interpretively.
3345:
3346: Setting PCRE_NO_START_OPTIMIZE can change the outcome of a matching
3347: operation. Consider the pattern
3348:
3349: (*COMMIT)ABC
3350:
3351: When this is compiled, PCRE records the fact that a match must start
3352: with the character "A". Suppose the subject string is "DEFABC". The
3353: start-up optimization scans along the subject, finds "A" and runs the
3354: first match attempt from there. The (*COMMIT) item means that the pat-
3355: tern must match the current starting position, which in this case, it
3356: does. However, if the same match is run with PCRE_NO_START_OPTIMIZE
3357: set, the initial scan along the subject string does not happen. The
3358: first match attempt is run starting from "D" and when this fails,
3359: (*COMMIT) prevents any further matches being tried, so the overall
3360: result is "no match". If the pattern is studied, more start-up opti-
3361: mizations may be used. For example, a minimum length for the subject
3362: may be recorded. Consider the pattern
3363:
3364: (*MARK:A)(X|Y)
3365:
3366: The minimum length for a match is one character. If the subject is
3367: "ABC", there will be attempts to match "ABC", "BC", "C", and then
3368: finally an empty string. If the pattern is studied, the final attempt
3369: does not take place, because PCRE knows that the subject is too short,
3370: and so the (*MARK) is never encountered. In this case, studying the
3371: pattern does not affect the overall match result, which is still "no
3372: match", but it does affect the auxiliary information that is returned.
3373:
3374: PCRE_NO_UTF8_CHECK
3375:
3376: When PCRE_UTF8 is set at compile time, the validity of the subject as a
3377: UTF-8 string is automatically checked when pcre_exec() is subsequently
3378: called. The entire string is checked before any other processing takes
3379: place. The value of startoffset is also checked to ensure that it
3380: points to the start of a UTF-8 character. There is a discussion about
3381: the validity of UTF-8 strings in the pcreunicode page. If an invalid
3382: sequence of bytes is found, pcre_exec() returns the error
3383: PCRE_ERROR_BADUTF8 or, if PCRE_PARTIAL_HARD is set and the problem is a
3384: truncated character at the end of the subject, PCRE_ERROR_SHORTUTF8. In
3385: both cases, information about the precise nature of the error may also
3386: be returned (see the descriptions of these errors in the section enti-
3387: tled Error return values from pcre_exec() below). If startoffset con-
3388: tains a value that does not point to the start of a UTF-8 character (or
3389: to the end of the subject), PCRE_ERROR_BADUTF8_OFFSET is returned.
3390:
3391: If you already know that your subject is valid, and you want to skip
3392: these checks for performance reasons, you can set the
3393: PCRE_NO_UTF8_CHECK option when calling pcre_exec(). You might want to
3394: do this for the second and subsequent calls to pcre_exec() if you are
3395: making repeated calls to find all the matches in a single subject
3396: string. However, you should be sure that the value of startoffset
3397: points to the start of a character (or the end of the subject). When
3398: PCRE_NO_UTF8_CHECK is set, the effect of passing an invalid string as a
3399: subject or an invalid value of startoffset is undefined. Your program
3400: may crash or loop.
3401:
3402: PCRE_PARTIAL_HARD
3403: PCRE_PARTIAL_SOFT
3404:
3405: These options turn on the partial matching feature. For backwards com-
3406: patibility, PCRE_PARTIAL is a synonym for PCRE_PARTIAL_SOFT. A partial
3407: match occurs if the end of the subject string is reached successfully,
3408: but there are not enough subject characters to complete the match. If
3409: this happens when PCRE_PARTIAL_SOFT (but not PCRE_PARTIAL_HARD) is set,
3410: matching continues by testing any remaining alternatives. Only if no
3411: complete match can be found is PCRE_ERROR_PARTIAL returned instead of
3412: PCRE_ERROR_NOMATCH. In other words, PCRE_PARTIAL_SOFT says that the
3413: caller is prepared to handle a partial match, but only if no complete
3414: match can be found.
3415:
3416: If PCRE_PARTIAL_HARD is set, it overrides PCRE_PARTIAL_SOFT. In this
3417: case, if a partial match is found, pcre_exec() immediately returns
3418: PCRE_ERROR_PARTIAL, without considering any other alternatives. In
3419: other words, when PCRE_PARTIAL_HARD is set, a partial match is consid-
3420: ered to be more important that an alternative complete match.
3421:
3422: In both cases, the portion of the string that was inspected when the
3423: partial match was found is set as the first matching string. There is a
3424: more detailed discussion of partial and multi-segment matching, with
3425: examples, in the pcrepartial documentation.
3426:
3427: The string to be matched by pcre_exec()
3428:
3429: The subject string is passed to pcre_exec() as a pointer in subject, a
3430: length in length, and a starting offset in startoffset. The units for
3431: length and startoffset are bytes for the 8-bit library, 16-bit data
3432: items for the 16-bit library, and 32-bit data items for the 32-bit
3433: library.
3434:
3435: If startoffset is negative or greater than the length of the subject,
3436: pcre_exec() returns PCRE_ERROR_BADOFFSET. When the starting offset is
3437: zero, the search for a match starts at the beginning of the subject,
3438: and this is by far the most common case. In UTF-8 or UTF-16 mode, the
3439: offset must point to the start of a character, or the end of the sub-
3440: ject (in UTF-32 mode, one data unit equals one character, so all off-
3441: sets are valid). Unlike the pattern string, the subject may contain
3442: binary zeroes.
3443:
3444: A non-zero starting offset is useful when searching for another match
3445: in the same subject by calling pcre_exec() again after a previous suc-
3446: cess. Setting startoffset differs from just passing over a shortened
3447: string and setting PCRE_NOTBOL in the case of a pattern that begins
3448: with any kind of lookbehind. For example, consider the pattern
3449:
3450: \Biss\B
3451:
3452: which finds occurrences of "iss" in the middle of words. (\B matches
3453: only if the current position in the subject is not a word boundary.)
3454: When applied to the string "Mississipi" the first call to pcre_exec()
3455: finds the first occurrence. If pcre_exec() is called again with just
3456: the remainder of the subject, namely "issipi", it does not match,
3457: because \B is always false at the start of the subject, which is deemed
3458: to be a word boundary. However, if pcre_exec() is passed the entire
3459: string again, but with startoffset set to 4, it finds the second occur-
3460: rence of "iss" because it is able to look behind the starting point to
3461: discover that it is preceded by a letter.
3462:
3463: Finding all the matches in a subject is tricky when the pattern can
3464: match an empty string. It is possible to emulate Perl's /g behaviour by
3465: first trying the match again at the same offset, with the
3466: PCRE_NOTEMPTY_ATSTART and PCRE_ANCHORED options, and then if that
3467: fails, advancing the starting offset and trying an ordinary match
3468: again. There is some code that demonstrates how to do this in the pcre-
3469: demo sample program. In the most general case, you have to check to see
3470: if the newline convention recognizes CRLF as a newline, and if so, and
3471: the current character is CR followed by LF, advance the starting offset
3472: by two characters instead of one.
3473:
3474: If a non-zero starting offset is passed when the pattern is anchored,
3475: one attempt to match at the given offset is made. This can only succeed
3476: if the pattern does not require the match to be at the start of the
3477: subject.
3478:
3479: How pcre_exec() returns captured substrings
3480:
3481: In general, a pattern matches a certain portion of the subject, and in
3482: addition, further substrings from the subject may be picked out by
3483: parts of the pattern. Following the usage in Jeffrey Friedl's book,
3484: this is called "capturing" in what follows, and the phrase "capturing
3485: subpattern" is used for a fragment of a pattern that picks out a sub-
3486: string. PCRE supports several other kinds of parenthesized subpattern
3487: that do not cause substrings to be captured.
3488:
3489: Captured substrings are returned to the caller via a vector of integers
3490: whose address is passed in ovector. The number of elements in the vec-
3491: tor is passed in ovecsize, which must be a non-negative number. Note:
3492: this argument is NOT the size of ovector in bytes.
3493:
3494: The first two-thirds of the vector is used to pass back captured sub-
3495: strings, each substring using a pair of integers. The remaining third
3496: of the vector is used as workspace by pcre_exec() while matching cap-
3497: turing subpatterns, and is not available for passing back information.
3498: The number passed in ovecsize should always be a multiple of three. If
3499: it is not, it is rounded down.
3500:
3501: When a match is successful, information about captured substrings is
3502: returned in pairs of integers, starting at the beginning of ovector,
3503: and continuing up to two-thirds of its length at the most. The first
3504: element of each pair is set to the offset of the first character in a
3505: substring, and the second is set to the offset of the first character
3506: after the end of a substring. These values are always data unit off-
3507: sets, even in UTF mode. They are byte offsets in the 8-bit library,
3508: 16-bit data item offsets in the 16-bit library, and 32-bit data item
3509: offsets in the 32-bit library. Note: they are not character counts.
3510:
3511: The first pair of integers, ovector[0] and ovector[1], identify the
3512: portion of the subject string matched by the entire pattern. The next
3513: pair is used for the first capturing subpattern, and so on. The value
3514: returned by pcre_exec() is one more than the highest numbered pair that
3515: has been set. For example, if two substrings have been captured, the
3516: returned value is 3. If there are no capturing subpatterns, the return
3517: value from a successful match is 1, indicating that just the first pair
3518: of offsets has been set.
3519:
3520: If a capturing subpattern is matched repeatedly, it is the last portion
3521: of the string that it matched that is returned.
3522:
3523: If the vector is too small to hold all the captured substring offsets,
3524: it is used as far as possible (up to two-thirds of its length), and the
3525: function returns a value of zero. If neither the actual string matched
3526: nor any captured substrings are of interest, pcre_exec() may be called
3527: with ovector passed as NULL and ovecsize as zero. However, if the pat-
3528: tern contains back references and the ovector is not big enough to
3529: remember the related substrings, PCRE has to get additional memory for
3530: use during matching. Thus it is usually advisable to supply an ovector
3531: of reasonable size.
3532:
3533: There are some cases where zero is returned (indicating vector over-
3534: flow) when in fact the vector is exactly the right size for the final
3535: match. For example, consider the pattern
3536:
3537: (a)(?:(b)c|bd)
3538:
3539: If a vector of 6 elements (allowing for only 1 captured substring) is
3540: given with subject string "abd", pcre_exec() will try to set the second
3541: captured string, thereby recording a vector overflow, before failing to
3542: match "c" and backing up to try the second alternative. The zero
3543: return, however, does correctly indicate that the maximum number of
3544: slots (namely 2) have been filled. In similar cases where there is tem-
3545: porary overflow, but the final number of used slots is actually less
3546: than the maximum, a non-zero value is returned.
3547:
3548: The pcre_fullinfo() function can be used to find out how many capturing
3549: subpatterns there are in a compiled pattern. The smallest size for
3550: ovector that will allow for n captured substrings, in addition to the
3551: offsets of the substring matched by the whole pattern, is (n+1)*3.
3552:
3553: It is possible for capturing subpattern number n+1 to match some part
3554: of the subject when subpattern n has not been used at all. For example,
3555: if the string "abc" is matched against the pattern (a|(z))(bc) the
3556: return from the function is 4, and subpatterns 1 and 3 are matched, but
3557: 2 is not. When this happens, both values in the offset pairs corre-
3558: sponding to unused subpatterns are set to -1.
3559:
3560: Offset values that correspond to unused subpatterns at the end of the
3561: expression are also set to -1. For example, if the string "abc" is
3562: matched against the pattern (abc)(x(yz)?)? subpatterns 2 and 3 are not
3563: matched. The return from the function is 2, because the highest used
3564: capturing subpattern number is 1, and the offsets for for the second
3565: and third capturing subpatterns (assuming the vector is large enough,
3566: of course) are set to -1.
3567:
3568: Note: Elements in the first two-thirds of ovector that do not corre-
3569: spond to capturing parentheses in the pattern are never changed. That
3570: is, if a pattern contains n capturing parentheses, no more than ovec-
3571: tor[0] to ovector[2n+1] are set by pcre_exec(). The other elements (in
3572: the first two-thirds) retain whatever values they previously had.
3573:
3574: Some convenience functions are provided for extracting the captured
3575: substrings as separate strings. These are described below.
3576:
3577: Error return values from pcre_exec()
3578:
3579: If pcre_exec() fails, it returns a negative number. The following are
3580: defined in the header file:
3581:
3582: PCRE_ERROR_NOMATCH (-1)
3583:
3584: The subject string did not match the pattern.
3585:
3586: PCRE_ERROR_NULL (-2)
3587:
3588: Either code or subject was passed as NULL, or ovector was NULL and
3589: ovecsize was not zero.
3590:
3591: PCRE_ERROR_BADOPTION (-3)
3592:
3593: An unrecognized bit was set in the options argument.
3594:
3595: PCRE_ERROR_BADMAGIC (-4)
3596:
3597: PCRE stores a 4-byte "magic number" at the start of the compiled code,
3598: to catch the case when it is passed a junk pointer and to detect when a
3599: pattern that was compiled in an environment of one endianness is run in
3600: an environment with the other endianness. This is the error that PCRE
3601: gives when the magic number is not present.
3602:
3603: PCRE_ERROR_UNKNOWN_OPCODE (-5)
3604:
3605: While running the pattern match, an unknown item was encountered in the
3606: compiled pattern. This error could be caused by a bug in PCRE or by
3607: overwriting of the compiled pattern.
3608:
3609: PCRE_ERROR_NOMEMORY (-6)
3610:
3611: If a pattern contains back references, but the ovector that is passed
3612: to pcre_exec() is not big enough to remember the referenced substrings,
3613: PCRE gets a block of memory at the start of matching to use for this
3614: purpose. If the call via pcre_malloc() fails, this error is given. The
3615: memory is automatically freed at the end of matching.
3616:
3617: This error is also given if pcre_stack_malloc() fails in pcre_exec().
3618: This can happen only when PCRE has been compiled with --disable-stack-
3619: for-recursion.
3620:
3621: PCRE_ERROR_NOSUBSTRING (-7)
3622:
3623: This error is used by the pcre_copy_substring(), pcre_get_substring(),
3624: and pcre_get_substring_list() functions (see below). It is never
3625: returned by pcre_exec().
3626:
3627: PCRE_ERROR_MATCHLIMIT (-8)
3628:
3629: The backtracking limit, as specified by the match_limit field in a
3630: pcre_extra structure (or defaulted) was reached. See the description
3631: above.
3632:
3633: PCRE_ERROR_CALLOUT (-9)
3634:
3635: This error is never generated by pcre_exec() itself. It is provided for
3636: use by callout functions that want to yield a distinctive error code.
3637: See the pcrecallout documentation for details.
3638:
3639: PCRE_ERROR_BADUTF8 (-10)
3640:
3641: A string that contains an invalid UTF-8 byte sequence was passed as a
3642: subject, and the PCRE_NO_UTF8_CHECK option was not set. If the size of
3643: the output vector (ovecsize) is at least 2, the byte offset to the
3644: start of the the invalid UTF-8 character is placed in the first ele-
3645: ment, and a reason code is placed in the second element. The reason
3646: codes are listed in the following section. For backward compatibility,
3647: if PCRE_PARTIAL_HARD is set and the problem is a truncated UTF-8 char-
3648: acter at the end of the subject (reason codes 1 to 5),
3649: PCRE_ERROR_SHORTUTF8 is returned instead of PCRE_ERROR_BADUTF8.
3650:
3651: PCRE_ERROR_BADUTF8_OFFSET (-11)
3652:
3653: The UTF-8 byte sequence that was passed as a subject was checked and
3654: found to be valid (the PCRE_NO_UTF8_CHECK option was not set), but the
3655: value of startoffset did not point to the beginning of a UTF-8 charac-
3656: ter or the end of the subject.
3657:
3658: PCRE_ERROR_PARTIAL (-12)
3659:
3660: The subject string did not match, but it did match partially. See the
3661: pcrepartial documentation for details of partial matching.
3662:
3663: PCRE_ERROR_BADPARTIAL (-13)
3664:
3665: This code is no longer in use. It was formerly returned when the
3666: PCRE_PARTIAL option was used with a compiled pattern containing items
3667: that were not supported for partial matching. From release 8.00
3668: onwards, there are no restrictions on partial matching.
3669:
3670: PCRE_ERROR_INTERNAL (-14)
3671:
3672: An unexpected internal error has occurred. This error could be caused
3673: by a bug in PCRE or by overwriting of the compiled pattern.
3674:
3675: PCRE_ERROR_BADCOUNT (-15)
3676:
3677: This error is given if the value of the ovecsize argument is negative.
3678:
3679: PCRE_ERROR_RECURSIONLIMIT (-21)
3680:
3681: The internal recursion limit, as specified by the match_limit_recursion
3682: field in a pcre_extra structure (or defaulted) was reached. See the
3683: description above.
3684:
3685: PCRE_ERROR_BADNEWLINE (-23)
3686:
3687: An invalid combination of PCRE_NEWLINE_xxx options was given.
3688:
3689: PCRE_ERROR_BADOFFSET (-24)
3690:
3691: The value of startoffset was negative or greater than the length of the
3692: subject, that is, the value in length.
3693:
3694: PCRE_ERROR_SHORTUTF8 (-25)
3695:
3696: This error is returned instead of PCRE_ERROR_BADUTF8 when the subject
3697: string ends with a truncated UTF-8 character and the PCRE_PARTIAL_HARD
3698: option is set. Information about the failure is returned as for
3699: PCRE_ERROR_BADUTF8. It is in fact sufficient to detect this case, but
3700: this special error code for PCRE_PARTIAL_HARD precedes the implementa-
3701: tion of returned information; it is retained for backwards compatibil-
3702: ity.
3703:
3704: PCRE_ERROR_RECURSELOOP (-26)
3705:
3706: This error is returned when pcre_exec() detects a recursion loop within
3707: the pattern. Specifically, it means that either the whole pattern or a
3708: subpattern has been called recursively for the second time at the same
3709: position in the subject string. Some simple patterns that might do this
3710: are detected and faulted at compile time, but more complicated cases,
3711: in particular mutual recursions between two different subpatterns, can-
3712: not be detected until run time.
3713:
3714: PCRE_ERROR_JIT_STACKLIMIT (-27)
3715:
3716: This error is returned when a pattern that was successfully studied
3717: using a JIT compile option is being matched, but the memory available
3718: for the just-in-time processing stack is not large enough. See the
3719: pcrejit documentation for more details.
3720:
3721: PCRE_ERROR_BADMODE (-28)
3722:
3723: This error is given if a pattern that was compiled by the 8-bit library
3724: is passed to a 16-bit or 32-bit library function, or vice versa.
3725:
3726: PCRE_ERROR_BADENDIANNESS (-29)
3727:
3728: This error is given if a pattern that was compiled and saved is
3729: reloaded on a host with different endianness. The utility function
3730: pcre_pattern_to_host_byte_order() can be used to convert such a pattern
3731: so that it runs on the new host.
3732:
3733: PCRE_ERROR_JIT_BADOPTION
3734:
3735: This error is returned when a pattern that was successfully studied
3736: using a JIT compile option is being matched, but the matching mode
3737: (partial or complete match) does not correspond to any JIT compilation
3738: mode. When the JIT fast path function is used, this error may be also
3739: given for invalid options. See the pcrejit documentation for more
3740: details.
3741:
3742: PCRE_ERROR_BADLENGTH (-32)
3743:
3744: This error is given if pcre_exec() is called with a negative value for
3745: the length argument.
3746:
3747: Error numbers -16 to -20, -22, and 30 are not used by pcre_exec().
3748:
3749: Reason codes for invalid UTF-8 strings
3750:
3751: This section applies only to the 8-bit library. The corresponding
3752: information for the 16-bit and 32-bit libraries is given in the pcre16
3753: and pcre32 pages.
3754:
3755: When pcre_exec() returns either PCRE_ERROR_BADUTF8 or PCRE_ERROR_SHORT-
3756: UTF8, and the size of the output vector (ovecsize) is at least 2, the
3757: offset of the start of the invalid UTF-8 character is placed in the
3758: first output vector element (ovector[0]) and a reason code is placed in
3759: the second element (ovector[1]). The reason codes are given names in
3760: the pcre.h header file:
3761:
3762: PCRE_UTF8_ERR1
3763: PCRE_UTF8_ERR2
3764: PCRE_UTF8_ERR3
3765: PCRE_UTF8_ERR4
3766: PCRE_UTF8_ERR5
3767:
3768: The string ends with a truncated UTF-8 character; the code specifies
3769: how many bytes are missing (1 to 5). Although RFC 3629 restricts UTF-8
3770: characters to be no longer than 4 bytes, the encoding scheme (origi-
3771: nally defined by RFC 2279) allows for up to 6 bytes, and this is
3772: checked first; hence the possibility of 4 or 5 missing bytes.
3773:
3774: PCRE_UTF8_ERR6
3775: PCRE_UTF8_ERR7
3776: PCRE_UTF8_ERR8
3777: PCRE_UTF8_ERR9
3778: PCRE_UTF8_ERR10
3779:
3780: The two most significant bits of the 2nd, 3rd, 4th, 5th, or 6th byte of
3781: the character do not have the binary value 0b10 (that is, either the
3782: most significant bit is 0, or the next bit is 1).
3783:
3784: PCRE_UTF8_ERR11
3785: PCRE_UTF8_ERR12
3786:
3787: A character that is valid by the RFC 2279 rules is either 5 or 6 bytes
3788: long; these code points are excluded by RFC 3629.
3789:
3790: PCRE_UTF8_ERR13
3791:
3792: A 4-byte character has a value greater than 0x10fff; these code points
3793: are excluded by RFC 3629.
3794:
3795: PCRE_UTF8_ERR14
3796:
3797: A 3-byte character has a value in the range 0xd800 to 0xdfff; this
3798: range of code points are reserved by RFC 3629 for use with UTF-16, and
3799: so are excluded from UTF-8.
3800:
3801: PCRE_UTF8_ERR15
3802: PCRE_UTF8_ERR16
3803: PCRE_UTF8_ERR17
3804: PCRE_UTF8_ERR18
3805: PCRE_UTF8_ERR19
3806:
3807: A 2-, 3-, 4-, 5-, or 6-byte character is "overlong", that is, it codes
3808: for a value that can be represented by fewer bytes, which is invalid.
3809: For example, the two bytes 0xc0, 0xae give the value 0x2e, whose cor-
3810: rect coding uses just one byte.
3811:
3812: PCRE_UTF8_ERR20
3813:
3814: The two most significant bits of the first byte of a character have the
3815: binary value 0b10 (that is, the most significant bit is 1 and the sec-
3816: ond is 0). Such a byte can only validly occur as the second or subse-
3817: quent byte of a multi-byte character.
3818:
3819: PCRE_UTF8_ERR21
3820:
3821: The first byte of a character has the value 0xfe or 0xff. These values
3822: can never occur in a valid UTF-8 string.
3823:
3824: PCRE_UTF8_ERR22
3825:
3826: This error code was formerly used when the presence of a so-called
3827: "non-character" caused an error. Unicode corrigendum #9 makes it clear
3828: that such characters should not cause a string to be rejected, and so
3829: this code is no longer in use and is never returned.
3830:
3831:
3832: EXTRACTING CAPTURED SUBSTRINGS BY NUMBER
3833:
3834: int pcre_copy_substring(const char *subject, int *ovector,
3835: int stringcount, int stringnumber, char *buffer,
3836: int buffersize);
3837:
3838: int pcre_get_substring(const char *subject, int *ovector,
3839: int stringcount, int stringnumber,
3840: const char **stringptr);
3841:
3842: int pcre_get_substring_list(const char *subject,
3843: int *ovector, int stringcount, const char ***listptr);
3844:
3845: Captured substrings can be accessed directly by using the offsets
3846: returned by pcre_exec() in ovector. For convenience, the functions
3847: pcre_copy_substring(), pcre_get_substring(), and pcre_get_sub-
3848: string_list() are provided for extracting captured substrings as new,
3849: separate, zero-terminated strings. These functions identify substrings
3850: by number. The next section describes functions for extracting named
3851: substrings.
3852:
3853: A substring that contains a binary zero is correctly extracted and has
3854: a further zero added on the end, but the result is not, of course, a C
3855: string. However, you can process such a string by referring to the
3856: length that is returned by pcre_copy_substring() and pcre_get_sub-
3857: string(). Unfortunately, the interface to pcre_get_substring_list() is
3858: not adequate for handling strings containing binary zeros, because the
3859: end of the final string is not independently indicated.
3860:
3861: The first three arguments are the same for all three of these func-
3862: tions: subject is the subject string that has just been successfully
3863: matched, ovector is a pointer to the vector of integer offsets that was
3864: passed to pcre_exec(), and stringcount is the number of substrings that
3865: were captured by the match, including the substring that matched the
3866: entire regular expression. This is the value returned by pcre_exec() if
3867: it is greater than zero. If pcre_exec() returned zero, indicating that
3868: it ran out of space in ovector, the value passed as stringcount should
3869: be the number of elements in the vector divided by three.
3870:
3871: The functions pcre_copy_substring() and pcre_get_substring() extract a
3872: single substring, whose number is given as stringnumber. A value of
3873: zero extracts the substring that matched the entire pattern, whereas
3874: higher values extract the captured substrings. For pcre_copy_sub-
3875: string(), the string is placed in buffer, whose length is given by
3876: buffersize, while for pcre_get_substring() a new block of memory is
3877: obtained via pcre_malloc, and its address is returned via stringptr.
3878: The yield of the function is the length of the string, not including
3879: the terminating zero, or one of these error codes:
3880:
3881: PCRE_ERROR_NOMEMORY (-6)
3882:
3883: The buffer was too small for pcre_copy_substring(), or the attempt to
3884: get memory failed for pcre_get_substring().
3885:
3886: PCRE_ERROR_NOSUBSTRING (-7)
3887:
3888: There is no substring whose number is stringnumber.
3889:
3890: The pcre_get_substring_list() function extracts all available sub-
3891: strings and builds a list of pointers to them. All this is done in a
3892: single block of memory that is obtained via pcre_malloc. The address of
3893: the memory block is returned via listptr, which is also the start of
3894: the list of string pointers. The end of the list is marked by a NULL
3895: pointer. The yield of the function is zero if all went well, or the
3896: error code
3897:
3898: PCRE_ERROR_NOMEMORY (-6)
3899:
3900: if the attempt to get the memory block failed.
3901:
3902: When any of these functions encounter a substring that is unset, which
3903: can happen when capturing subpattern number n+1 matches some part of
3904: the subject, but subpattern n has not been used at all, they return an
3905: empty string. This can be distinguished from a genuine zero-length sub-
3906: string by inspecting the appropriate offset in ovector, which is nega-
3907: tive for unset substrings.
3908:
3909: The two convenience functions pcre_free_substring() and pcre_free_sub-
3910: string_list() can be used to free the memory returned by a previous
3911: call of pcre_get_substring() or pcre_get_substring_list(), respec-
3912: tively. They do nothing more than call the function pointed to by
3913: pcre_free, which of course could be called directly from a C program.
3914: However, PCRE is used in some situations where it is linked via a spe-
3915: cial interface to another programming language that cannot use
3916: pcre_free directly; it is for these cases that the functions are pro-
3917: vided.
3918:
3919:
3920: EXTRACTING CAPTURED SUBSTRINGS BY NAME
3921:
3922: int pcre_get_stringnumber(const pcre *code,
3923: const char *name);
3924:
3925: int pcre_copy_named_substring(const pcre *code,
3926: const char *subject, int *ovector,
3927: int stringcount, const char *stringname,
3928: char *buffer, int buffersize);
3929:
3930: int pcre_get_named_substring(const pcre *code,
3931: const char *subject, int *ovector,
3932: int stringcount, const char *stringname,
3933: const char **stringptr);
3934:
3935: To extract a substring by name, you first have to find associated num-
3936: ber. For example, for this pattern
3937:
3938: (a+)b(?<xxx>\d+)...
3939:
3940: the number of the subpattern called "xxx" is 2. If the name is known to
3941: be unique (PCRE_DUPNAMES was not set), you can find the number from the
3942: name by calling pcre_get_stringnumber(). The first argument is the com-
3943: piled pattern, and the second is the name. The yield of the function is
3944: the subpattern number, or PCRE_ERROR_NOSUBSTRING (-7) if there is no
3945: subpattern of that name.
3946:
3947: Given the number, you can extract the substring directly, or use one of
3948: the functions described in the previous section. For convenience, there
3949: are also two functions that do the whole job.
3950:
3951: Most of the arguments of pcre_copy_named_substring() and
3952: pcre_get_named_substring() are the same as those for the similarly
3953: named functions that extract by number. As these are described in the
3954: previous section, they are not re-described here. There are just two
3955: differences:
3956:
3957: First, instead of a substring number, a substring name is given. Sec-
3958: ond, there is an extra argument, given at the start, which is a pointer
3959: to the compiled pattern. This is needed in order to gain access to the
3960: name-to-number translation table.
3961:
3962: These functions call pcre_get_stringnumber(), and if it succeeds, they
3963: then call pcre_copy_substring() or pcre_get_substring(), as appropri-
3964: ate. NOTE: If PCRE_DUPNAMES is set and there are duplicate names, the
3965: behaviour may not be what you want (see the next section).
3966:
3967: Warning: If the pattern uses the (?| feature to set up multiple subpat-
3968: terns with the same number, as described in the section on duplicate
3969: subpattern numbers in the pcrepattern page, you cannot use names to
3970: distinguish the different subpatterns, because names are not included
3971: in the compiled code. The matching process uses only numbers. For this
3972: reason, the use of different names for subpatterns of the same number
3973: causes an error at compile time.
3974:
3975:
3976: DUPLICATE SUBPATTERN NAMES
3977:
3978: int pcre_get_stringtable_entries(const pcre *code,
3979: const char *name, char **first, char **last);
3980:
3981: When a pattern is compiled with the PCRE_DUPNAMES option, names for
3982: subpatterns are not required to be unique. (Duplicate names are always
3983: allowed for subpatterns with the same number, created by using the (?|
3984: feature. Indeed, if such subpatterns are named, they are required to
3985: use the same names.)
3986:
3987: Normally, patterns with duplicate names are such that in any one match,
3988: only one of the named subpatterns participates. An example is shown in
3989: the pcrepattern documentation.
3990:
3991: When duplicates are present, pcre_copy_named_substring() and
3992: pcre_get_named_substring() return the first substring corresponding to
3993: the given name that is set. If none are set, PCRE_ERROR_NOSUBSTRING
3994: (-7) is returned; no data is returned. The pcre_get_stringnumber()
3995: function returns one of the numbers that are associated with the name,
3996: but it is not defined which it is.
3997:
3998: If you want to get full details of all captured substrings for a given
3999: name, you must use the pcre_get_stringtable_entries() function. The
4000: first argument is the compiled pattern, and the second is the name. The
4001: third and fourth are pointers to variables which are updated by the
4002: function. After it has run, they point to the first and last entries in
4003: the name-to-number table for the given name. The function itself
4004: returns the length of each entry, or PCRE_ERROR_NOSUBSTRING (-7) if
4005: there are none. The format of the table is described above in the sec-
4006: tion entitled Information about a pattern above. Given all the rele-
4007: vant entries for the name, you can extract each of their numbers, and
4008: hence the captured data, if any.
4009:
4010:
4011: FINDING ALL POSSIBLE MATCHES
4012:
4013: The traditional matching function uses a similar algorithm to Perl,
4014: which stops when it finds the first match, starting at a given point in
4015: the subject. If you want to find all possible matches, or the longest
4016: possible match, consider using the alternative matching function (see
4017: below) instead. If you cannot use the alternative function, but still
4018: need to find all possible matches, you can kludge it up by making use
4019: of the callout facility, which is described in the pcrecallout documen-
4020: tation.
4021:
4022: What you have to do is to insert a callout right at the end of the pat-
4023: tern. When your callout function is called, extract and save the cur-
4024: rent matched substring. Then return 1, which forces pcre_exec() to
4025: backtrack and try other alternatives. Ultimately, when it runs out of
4026: matches, pcre_exec() will yield PCRE_ERROR_NOMATCH.
4027:
4028:
4029: OBTAINING AN ESTIMATE OF STACK USAGE
4030:
4031: Matching certain patterns using pcre_exec() can use a lot of process
4032: stack, which in certain environments can be rather limited in size.
4033: Some users find it helpful to have an estimate of the amount of stack
4034: that is used by pcre_exec(), to help them set recursion limits, as
4035: described in the pcrestack documentation. The estimate that is output
4036: by pcretest when called with the -m and -C options is obtained by call-
4037: ing pcre_exec with the values NULL, NULL, NULL, -999, and -999 for its
4038: first five arguments.
4039:
4040: Normally, if its first argument is NULL, pcre_exec() immediately
4041: returns the negative error code PCRE_ERROR_NULL, but with this special
4042: combination of arguments, it returns instead a negative number whose
4043: absolute value is the approximate stack frame size in bytes. (A nega-
4044: tive number is used so that it is clear that no match has happened.)
4045: The value is approximate because in some cases, recursive calls to
4046: pcre_exec() occur when there are one or two additional variables on the
4047: stack.
4048:
4049: If PCRE has been compiled to use the heap instead of the stack for
4050: recursion, the value returned is the size of each block that is
4051: obtained from the heap.
4052:
4053:
4054: MATCHING A PATTERN: THE ALTERNATIVE FUNCTION
4055:
4056: int pcre_dfa_exec(const pcre *code, const pcre_extra *extra,
4057: const char *subject, int length, int startoffset,
4058: int options, int *ovector, int ovecsize,
4059: int *workspace, int wscount);
4060:
4061: The function pcre_dfa_exec() is called to match a subject string
4062: against a compiled pattern, using a matching algorithm that scans the
4063: subject string just once, and does not backtrack. This has different
4064: characteristics to the normal algorithm, and is not compatible with
4065: Perl. Some of the features of PCRE patterns are not supported. Never-
4066: theless, there are times when this kind of matching can be useful. For
4067: a discussion of the two matching algorithms, and a list of features
4068: that pcre_dfa_exec() does not support, see the pcrematching documenta-
4069: tion.
4070:
4071: The arguments for the pcre_dfa_exec() function are the same as for
4072: pcre_exec(), plus two extras. The ovector argument is used in a differ-
4073: ent way, and this is described below. The other common arguments are
4074: used in the same way as for pcre_exec(), so their description is not
4075: repeated here.
4076:
4077: The two additional arguments provide workspace for the function. The
4078: workspace vector should contain at least 20 elements. It is used for
4079: keeping track of multiple paths through the pattern tree. More
4080: workspace will be needed for patterns and subjects where there are a
4081: lot of potential matches.
4082:
4083: Here is an example of a simple call to pcre_dfa_exec():
4084:
4085: int rc;
4086: int ovector[10];
4087: int wspace[20];
4088: rc = pcre_dfa_exec(
4089: re, /* result of pcre_compile() */
4090: NULL, /* we didn't study the pattern */
4091: "some string", /* the subject string */
4092: 11, /* the length of the subject string */
4093: 0, /* start at offset 0 in the subject */
4094: 0, /* default options */
4095: ovector, /* vector of integers for substring information */
4096: 10, /* number of elements (NOT size in bytes) */
4097: wspace, /* working space vector */
4098: 20); /* number of elements (NOT size in bytes) */
4099:
4100: Option bits for pcre_dfa_exec()
4101:
4102: The unused bits of the options argument for pcre_dfa_exec() must be
4103: zero. The only bits that may be set are PCRE_ANCHORED, PCRE_NEW-
4104: LINE_xxx, PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY,
4105: PCRE_NOTEMPTY_ATSTART, PCRE_NO_UTF8_CHECK, PCRE_BSR_ANYCRLF,
4106: PCRE_BSR_UNICODE, PCRE_NO_START_OPTIMIZE, PCRE_PARTIAL_HARD, PCRE_PAR-
4107: TIAL_SOFT, PCRE_DFA_SHORTEST, and PCRE_DFA_RESTART. All but the last
4108: four of these are exactly the same as for pcre_exec(), so their
4109: description is not repeated here.
4110:
4111: PCRE_PARTIAL_HARD
4112: PCRE_PARTIAL_SOFT
4113:
4114: These have the same general effect as they do for pcre_exec(), but the
4115: details are slightly different. When PCRE_PARTIAL_HARD is set for
4116: pcre_dfa_exec(), it returns PCRE_ERROR_PARTIAL if the end of the sub-
4117: ject is reached and there is still at least one matching possibility
4118: that requires additional characters. This happens even if some complete
4119: matches have also been found. When PCRE_PARTIAL_SOFT is set, the return
4120: code PCRE_ERROR_NOMATCH is converted into PCRE_ERROR_PARTIAL if the end
4121: of the subject is reached, there have been no complete matches, but
4122: there is still at least one matching possibility. The portion of the
4123: string that was inspected when the longest partial match was found is
4124: set as the first matching string in both cases. There is a more
4125: detailed discussion of partial and multi-segment matching, with exam-
4126: ples, in the pcrepartial documentation.
4127:
4128: PCRE_DFA_SHORTEST
4129:
4130: Setting the PCRE_DFA_SHORTEST option causes the matching algorithm to
4131: stop as soon as it has found one match. Because of the way the alterna-
4132: tive algorithm works, this is necessarily the shortest possible match
4133: at the first possible matching point in the subject string.
4134:
4135: PCRE_DFA_RESTART
4136:
4137: When pcre_dfa_exec() returns a partial match, it is possible to call it
4138: again, with additional subject characters, and have it continue with
4139: the same match. The PCRE_DFA_RESTART option requests this action; when
4140: it is set, the workspace and wscount options must reference the same
4141: vector as before because data about the match so far is left in them
4142: after a partial match. There is more discussion of this facility in the
4143: pcrepartial documentation.
4144:
4145: Successful returns from pcre_dfa_exec()
4146:
4147: When pcre_dfa_exec() succeeds, it may have matched more than one sub-
4148: string in the subject. Note, however, that all the matches from one run
4149: of the function start at the same point in the subject. The shorter
4150: matches are all initial substrings of the longer matches. For example,
4151: if the pattern
4152:
4153: <.*>
4154:
4155: is matched against the string
4156:
4157: This is <something> <something else> <something further> no more
4158:
4159: the three matched strings are
4160:
4161: <something>
4162: <something> <something else>
4163: <something> <something else> <something further>
4164:
4165: On success, the yield of the function is a number greater than zero,
4166: which is the number of matched substrings. The substrings themselves
4167: are returned in ovector. Each string uses two elements; the first is
4168: the offset to the start, and the second is the offset to the end. In
4169: fact, all the strings have the same start offset. (Space could have
4170: been saved by giving this only once, but it was decided to retain some
4171: compatibility with the way pcre_exec() returns data, even though the
4172: meaning of the strings is different.)
4173:
4174: The strings are returned in reverse order of length; that is, the long-
4175: est matching string is given first. If there were too many matches to
4176: fit into ovector, the yield of the function is zero, and the vector is
4177: filled with the longest matches. Unlike pcre_exec(), pcre_dfa_exec()
4178: can use the entire ovector for returning matched strings.
4179:
4180: NOTE: PCRE's "auto-possessification" optimization usually applies to
4181: character repeats at the end of a pattern (as well as internally). For
4182: example, the pattern "a\d+" is compiled as if it were "a\d++" because
4183: there is no point even considering the possibility of backtracking into
4184: the repeated digits. For DFA matching, this means that only one possi-
4185: ble match is found. If you really do want multiple matches in such
4186: cases, either use an ungreedy repeat ("a\d+?") or set the
4187: PCRE_NO_AUTO_POSSESS option when compiling.
4188:
4189: Error returns from pcre_dfa_exec()
4190:
4191: The pcre_dfa_exec() function returns a negative number when it fails.
4192: Many of the errors are the same as for pcre_exec(), and these are
4193: described above. There are in addition the following errors that are
4194: specific to pcre_dfa_exec():
4195:
4196: PCRE_ERROR_DFA_UITEM (-16)
4197:
4198: This return is given if pcre_dfa_exec() encounters an item in the pat-
4199: tern that it does not support, for instance, the use of \C or a back
4200: reference.
4201:
4202: PCRE_ERROR_DFA_UCOND (-17)
4203:
4204: This return is given if pcre_dfa_exec() encounters a condition item
4205: that uses a back reference for the condition, or a test for recursion
4206: in a specific group. These are not supported.
4207:
4208: PCRE_ERROR_DFA_UMLIMIT (-18)
4209:
4210: This return is given if pcre_dfa_exec() is called with an extra block
4211: that contains a setting of the match_limit or match_limit_recursion
4212: fields. This is not supported (these fields are meaningless for DFA
4213: matching).
4214:
4215: PCRE_ERROR_DFA_WSSIZE (-19)
4216:
4217: This return is given if pcre_dfa_exec() runs out of space in the
4218: workspace vector.
4219:
4220: PCRE_ERROR_DFA_RECURSE (-20)
4221:
4222: When a recursive subpattern is processed, the matching function calls
4223: itself recursively, using private vectors for ovector and workspace.
4224: This error is given if the output vector is not large enough. This
4225: should be extremely rare, as a vector of size 1000 is used.
4226:
4227: PCRE_ERROR_DFA_BADRESTART (-30)
4228:
4229: When pcre_dfa_exec() is called with the PCRE_DFA_RESTART option, some
4230: plausibility checks are made on the contents of the workspace, which
4231: should contain data about the previous partial match. If any of these
4232: checks fail, this error is given.
4233:
4234:
4235: SEE ALSO
4236:
4237: pcre16(3), pcre32(3), pcrebuild(3), pcrecallout(3), pcrecpp(3)(3),
4238: pcrematching(3), pcrepartial(3), pcreposix(3), pcreprecompile(3), pcre-
4239: sample(3), pcrestack(3).
4240:
4241:
4242: AUTHOR
4243:
4244: Philip Hazel
4245: University Computing Service
4246: Cambridge CB2 3QH, England.
4247:
4248:
4249: REVISION
4250:
4251: Last updated: 12 November 2013
4252: Copyright (c) 1997-2013 University of Cambridge.
4253: ------------------------------------------------------------------------------
4254:
4255:
4256: PCRECALLOUT(3) Library Functions Manual PCRECALLOUT(3)
4257:
4258:
4259:
4260: NAME
4261: PCRE - Perl-compatible regular expressions
4262:
4263: SYNOPSIS
4264:
4265: #include <pcre.h>
4266:
4267: int (*pcre_callout)(pcre_callout_block *);
4268:
4269: int (*pcre16_callout)(pcre16_callout_block *);
4270:
4271: int (*pcre32_callout)(pcre32_callout_block *);
4272:
4273:
4274: DESCRIPTION
4275:
4276: PCRE provides a feature called "callout", which is a means of temporar-
4277: ily passing control to the caller of PCRE in the middle of pattern
4278: matching. The caller of PCRE provides an external function by putting
4279: its entry point in the global variable pcre_callout (pcre16_callout for
4280: the 16-bit library, pcre32_callout for the 32-bit library). By default,
4281: this variable contains NULL, which disables all calling out.
4282:
4283: Within a regular expression, (?C) indicates the points at which the
4284: external function is to be called. Different callout points can be
4285: identified by putting a number less than 256 after the letter C. The
4286: default value is zero. For example, this pattern has two callout
4287: points:
4288:
4289: (?C1)abc(?C2)def
4290:
4291: If the PCRE_AUTO_CALLOUT option bit is set when a pattern is compiled,
4292: PCRE automatically inserts callouts, all with number 255, before each
4293: item in the pattern. For example, if PCRE_AUTO_CALLOUT is used with the
4294: pattern
4295:
4296: A(\d{2}|--)
4297:
4298: it is processed as if it were
4299:
4300: (?C255)A(?C255)((?C255)\d{2}(?C255)|(?C255)-(?C255)-(?C255))(?C255)
4301:
4302: Notice that there is a callout before and after each parenthesis and
4303: alternation bar. If the pattern contains a conditional group whose con-
4304: dition is an assertion, an automatic callout is inserted immediately
4305: before the condition. Such a callout may also be inserted explicitly,
4306: for example:
4307:
4308: (?(?C9)(?=a)ab|de)
4309:
4310: This applies only to assertion conditions (because they are themselves
4311: independent groups).
4312:
4313: Automatic callouts can be used for tracking the progress of pattern
4314: matching. The pcretest program has a pattern qualifier (/C) that sets
4315: automatic callouts; when it is used, the output indicates how the pat-
4316: tern is being matched. This is useful information when you are trying
4317: to optimize the performance of a particular pattern.
4318:
4319:
4320: MISSING CALLOUTS
4321:
4322: You should be aware that, because of optimizations in the way PCRE com-
4323: piles and matches patterns, callouts sometimes do not happen exactly as
4324: you might expect.
4325:
4326: At compile time, PCRE "auto-possessifies" repeated items when it knows
4327: that what follows cannot be part of the repeat. For example, a+[bc] is
4328: compiled as if it were a++[bc]. The pcretest output when this pattern
4329: is anchored and then applied with automatic callouts to the string
4330: "aaaa" is:
4331:
4332: --->aaaa
4333: +0 ^ ^
4334: +1 ^ a+
4335: +3 ^ ^ [bc]
4336: No match
4337:
4338: This indicates that when matching [bc] fails, there is no backtracking
4339: into a+ and therefore the callouts that would be taken for the back-
4340: tracks do not occur. You can disable the auto-possessify feature by
4341: passing PCRE_NO_AUTO_POSSESS to pcre_compile(), or starting the pattern
4342: with (*NO_AUTO_POSSESS). If this is done in pcretest (using the /O
4343: qualifier), the output changes to this:
4344:
4345: --->aaaa
4346: +0 ^ ^
4347: +1 ^ a+
4348: +3 ^ ^ [bc]
4349: +3 ^ ^ [bc]
4350: +3 ^ ^ [bc]
4351: +3 ^^ [bc]
4352: No match
4353:
4354: This time, when matching [bc] fails, the matcher backtracks into a+ and
4355: tries again, repeatedly, until a+ itself fails.
4356:
4357: Other optimizations that provide fast "no match" results also affect
4358: callouts. For example, if the pattern is
4359:
4360: ab(?C4)cd
4361:
4362: PCRE knows that any matching string must contain the letter "d". If the
4363: subject string is "abyz", the lack of "d" means that matching doesn't
4364: ever start, and the callout is never reached. However, with "abyd",
4365: though the result is still no match, the callout is obeyed.
4366:
4367: If the pattern is studied, PCRE knows the minimum length of a matching
4368: string, and will immediately give a "no match" return without actually
4369: running a match if the subject is not long enough, or, for unanchored
4370: patterns, if it has been scanned far enough.
4371:
4372: You can disable these optimizations by passing the PCRE_NO_START_OPTI-
4373: MIZE option to the matching function, or by starting the pattern with
4374: (*NO_START_OPT). This slows down the matching process, but does ensure
4375: that callouts such as the example above are obeyed.
4376:
4377:
4378: THE CALLOUT INTERFACE
4379:
4380: During matching, when PCRE reaches a callout point, the external func-
4381: tion defined by pcre_callout or pcre[16|32]_callout is called (if it is
4382: set). This applies to both normal and DFA matching. The only argument
4383: to the callout function is a pointer to a pcre_callout or
4384: pcre[16|32]_callout block. These structures contains the following
4385: fields:
4386:
4387: int version;
4388: int callout_number;
4389: int *offset_vector;
4390: const char *subject; (8-bit version)
4391: PCRE_SPTR16 subject; (16-bit version)
4392: PCRE_SPTR32 subject; (32-bit version)
4393: int subject_length;
4394: int start_match;
4395: int current_position;
4396: int capture_top;
4397: int capture_last;
4398: void *callout_data;
4399: int pattern_position;
4400: int next_item_length;
4401: const unsigned char *mark; (8-bit version)
4402: const PCRE_UCHAR16 *mark; (16-bit version)
4403: const PCRE_UCHAR32 *mark; (32-bit version)
4404:
4405: The version field is an integer containing the version number of the
4406: block format. The initial version was 0; the current version is 2. The
4407: version number will change again in future if additional fields are
4408: added, but the intention is never to remove any of the existing fields.
4409:
4410: The callout_number field contains the number of the callout, as com-
4411: piled into the pattern (that is, the number after ?C for manual call-
4412: outs, and 255 for automatically generated callouts).
4413:
4414: The offset_vector field is a pointer to the vector of offsets that was
4415: passed by the caller to the matching function. When pcre_exec() or
4416: pcre[16|32]_exec() is used, the contents can be inspected, in order to
4417: extract substrings that have been matched so far, in the same way as
4418: for extracting substrings after a match has completed. For the DFA
4419: matching functions, this field is not useful.
4420:
4421: The subject and subject_length fields contain copies of the values that
4422: were passed to the matching function.
4423:
4424: The start_match field normally contains the offset within the subject
4425: at which the current match attempt started. However, if the escape
4426: sequence \K has been encountered, this value is changed to reflect the
4427: modified starting point. If the pattern is not anchored, the callout
4428: function may be called several times from the same point in the pattern
4429: for different starting points in the subject.
4430:
4431: The current_position field contains the offset within the subject of
4432: the current match pointer.
4433:
4434: When the pcre_exec() or pcre[16|32]_exec() is used, the capture_top
4435: field contains one more than the number of the highest numbered cap-
4436: tured substring so far. If no substrings have been captured, the value
4437: of capture_top is one. This is always the case when the DFA functions
4438: are used, because they do not support captured substrings.
4439:
4440: The capture_last field contains the number of the most recently cap-
4441: tured substring. However, when a recursion exits, the value reverts to
4442: what it was outside the recursion, as do the values of all captured
4443: substrings. If no substrings have been captured, the value of cap-
4444: ture_last is -1. This is always the case for the DFA matching func-
4445: tions.
4446:
4447: The callout_data field contains a value that is passed to a matching
4448: function specifically so that it can be passed back in callouts. It is
4449: passed in the callout_data field of a pcre_extra or pcre[16|32]_extra
4450: data structure. If no such data was passed, the value of callout_data
4451: in a callout block is NULL. There is a description of the pcre_extra
4452: structure in the pcreapi documentation.
4453:
4454: The pattern_position field is present from version 1 of the callout
4455: structure. It contains the offset to the next item to be matched in the
4456: pattern string.
4457:
4458: The next_item_length field is present from version 1 of the callout
4459: structure. It contains the length of the next item to be matched in the
4460: pattern string. When the callout immediately precedes an alternation
4461: bar, a closing parenthesis, or the end of the pattern, the length is
4462: zero. When the callout precedes an opening parenthesis, the length is
4463: that of the entire subpattern.
4464:
4465: The pattern_position and next_item_length fields are intended to help
4466: in distinguishing between different automatic callouts, which all have
4467: the same callout number. However, they are set for all callouts.
4468:
4469: The mark field is present from version 2 of the callout structure. In
4470: callouts from pcre_exec() or pcre[16|32]_exec() it contains a pointer
4471: to the zero-terminated name of the most recently passed (*MARK),
4472: (*PRUNE), or (*THEN) item in the match, or NULL if no such items have
4473: been passed. Instances of (*PRUNE) or (*THEN) without a name do not
4474: obliterate a previous (*MARK). In callouts from the DFA matching func-
4475: tions this field always contains NULL.
4476:
4477:
4478: RETURN VALUES
4479:
4480: The external callout function returns an integer to PCRE. If the value
4481: is zero, matching proceeds as normal. If the value is greater than
4482: zero, matching fails at the current point, but the testing of other
4483: matching possibilities goes ahead, just as if a lookahead assertion had
4484: failed. If the value is less than zero, the match is abandoned, the
4485: matching function returns the negative value.
4486:
4487: Negative values should normally be chosen from the set of
4488: PCRE_ERROR_xxx values. In particular, PCRE_ERROR_NOMATCH forces a stan-
4489: dard "no match" failure. The error number PCRE_ERROR_CALLOUT is
4490: reserved for use by callout functions; it will never be used by PCRE
4491: itself.
4492:
4493:
4494: AUTHOR
4495:
4496: Philip Hazel
4497: University Computing Service
4498: Cambridge CB2 3QH, England.
4499:
4500:
4501: REVISION
4502:
4503: Last updated: 12 November 2013
4504: Copyright (c) 1997-2013 University of Cambridge.
4505: ------------------------------------------------------------------------------
4506:
4507:
4508: PCRECOMPAT(3) Library Functions Manual PCRECOMPAT(3)
4509:
4510:
4511:
4512: NAME
4513: PCRE - Perl-compatible regular expressions
4514:
4515: DIFFERENCES BETWEEN PCRE AND PERL
4516:
4517: This document describes the differences in the ways that PCRE and Perl
4518: handle regular expressions. The differences described here are with
4519: respect to Perl versions 5.10 and above.
4520:
4521: 1. PCRE has only a subset of Perl's Unicode support. Details of what it
4522: does have are given in the pcreunicode page.
4523:
4524: 2. PCRE allows repeat quantifiers only on parenthesized assertions, but
4525: they do not mean what you might think. For example, (?!a){3} does not
4526: assert that the next three characters are not "a". It just asserts that
4527: the next character is not "a" three times (in principle: PCRE optimizes
4528: this to run the assertion just once). Perl allows repeat quantifiers on
4529: other assertions such as \b, but these do not seem to have any use.
4530:
4531: 3. Capturing subpatterns that occur inside negative lookahead asser-
4532: tions are counted, but their entries in the offsets vector are never
4533: set. Perl sometimes (but not always) sets its numerical variables from
4534: inside negative assertions.
4535:
4536: 4. Though binary zero characters are supported in the subject string,
4537: they are not allowed in a pattern string because it is passed as a nor-
4538: mal C string, terminated by zero. The escape sequence \0 can be used in
4539: the pattern to represent a binary zero.
4540:
4541: 5. The following Perl escape sequences are not supported: \l, \u, \L,
4542: \U, and \N when followed by a character name or Unicode value. (\N on
4543: its own, matching a non-newline character, is supported.) In fact these
4544: are implemented by Perl's general string-handling and are not part of
4545: its pattern matching engine. If any of these are encountered by PCRE,
4546: an error is generated by default. However, if the PCRE_JAVASCRIPT_COM-
4547: PAT option is set, \U and \u are interpreted as JavaScript interprets
4548: them.
4549:
4550: 6. The Perl escape sequences \p, \P, and \X are supported only if PCRE
4551: is built with Unicode character property support. The properties that
4552: can be tested with \p and \P are limited to the general category prop-
4553: erties such as Lu and Nd, script names such as Greek or Han, and the
4554: derived properties Any and L&. PCRE does support the Cs (surrogate)
4555: property, which Perl does not; the Perl documentation says "Because
4556: Perl hides the need for the user to understand the internal representa-
4557: tion of Unicode characters, there is no need to implement the somewhat
4558: messy concept of surrogates."
4559:
4560: 7. PCRE does support the \Q...\E escape for quoting substrings. Charac-
4561: ters in between are treated as literals. This is slightly different
4562: from Perl in that $ and @ are also handled as literals inside the
4563: quotes. In Perl, they cause variable interpolation (but of course PCRE
4564: does not have variables). Note the following examples:
4565:
4566: Pattern PCRE matches Perl matches
4567:
4568: \Qabc$xyz\E abc$xyz abc followed by the
4569: contents of $xyz
4570: \Qabc\$xyz\E abc\$xyz abc\$xyz
4571: \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4572:
4573: The \Q...\E sequence is recognized both inside and outside character
4574: classes.
4575:
4576: 8. Fairly obviously, PCRE does not support the (?{code}) and (??{code})
4577: constructions. However, there is support for recursive patterns. This
4578: is not available in Perl 5.8, but it is in Perl 5.10. Also, the PCRE
4579: "callout" feature allows an external function to be called during pat-
4580: tern matching. See the pcrecallout documentation for details.
4581:
4582: 9. Subpatterns that are called as subroutines (whether or not recur-
4583: sively) are always treated as atomic groups in PCRE. This is like
4584: Python, but unlike Perl. Captured values that are set outside a sub-
4585: routine call can be reference from inside in PCRE, but not in Perl.
4586: There is a discussion that explains these differences in more detail in
4587: the section on recursion differences from Perl in the pcrepattern page.
4588:
4589: 10. If any of the backtracking control verbs are used in a subpattern
4590: that is called as a subroutine (whether or not recursively), their
4591: effect is confined to that subpattern; it does not extend to the sur-
4592: rounding pattern. This is not always the case in Perl. In particular,
4593: if (*THEN) is present in a group that is called as a subroutine, its
4594: action is limited to that group, even if the group does not contain any
4595: | characters. Note that such subpatterns are processed as anchored at
4596: the point where they are tested.
4597:
4598: 11. If a pattern contains more than one backtracking control verb, the
4599: first one that is backtracked onto acts. For example, in the pattern
4600: A(*COMMIT)B(*PRUNE)C a failure in B triggers (*COMMIT), but a failure
4601: in C triggers (*PRUNE). Perl's behaviour is more complex; in many cases
4602: it is the same as PCRE, but there are examples where it differs.
4603:
4604: 12. Most backtracking verbs in assertions have their normal actions.
4605: They are not confined to the assertion.
4606:
4607: 13. There are some differences that are concerned with the settings of
4608: captured strings when part of a pattern is repeated. For example,
4609: matching "aba" against the pattern /^(a(b)?)+$/ in Perl leaves $2
4610: unset, but in PCRE it is set to "b".
4611:
4612: 14. PCRE's handling of duplicate subpattern numbers and duplicate sub-
4613: pattern names is not as general as Perl's. This is a consequence of the
4614: fact the PCRE works internally just with numbers, using an external ta-
4615: ble to translate between numbers and names. In particular, a pattern
4616: such as (?|(?<a>A)|(?<b)B), where the two capturing parentheses have
4617: the same number but different names, is not supported, and causes an
4618: error at compile time. If it were allowed, it would not be possible to
4619: distinguish which parentheses matched, because both names map to cap-
4620: turing subpattern number 1. To avoid this confusing situation, an error
4621: is given at compile time.
4622:
4623: 15. Perl recognizes comments in some places that PCRE does not, for
4624: example, between the ( and ? at the start of a subpattern. If the /x
4625: modifier is set, Perl allows white space between ( and ? (though cur-
4626: rent Perls warn that this is deprecated) but PCRE never does, even if
4627: the PCRE_EXTENDED option is set.
4628:
4629: 16. Perl, when in warning mode, gives warnings for character classes
4630: such as [A-\d] or [a-[:digit:]]. It then treats the hyphens as liter-
4631: als. PCRE has no warning features, so it gives an error in these cases
4632: because they are almost certainly user mistakes.
4633:
4634: 17. In PCRE, the upper/lower case character properties Lu and Ll are
4635: not affected when case-independent matching is specified. For example,
4636: \p{Lu} always matches an upper case letter. I think Perl has changed in
4637: this respect; in the release at the time of writing (5.16), \p{Lu} and
4638: \p{Ll} match all letters, regardless of case, when case independence is
4639: specified.
4640:
4641: 18. PCRE provides some extensions to the Perl regular expression facil-
4642: ities. Perl 5.10 includes new features that are not in earlier ver-
4643: sions of Perl, some of which (such as named parentheses) have been in
4644: PCRE for some time. This list is with respect to Perl 5.10:
4645:
4646: (a) Although lookbehind assertions in PCRE must match fixed length
4647: strings, each alternative branch of a lookbehind assertion can match a
4648: different length of string. Perl requires them all to have the same
4649: length.
4650:
4651: (b) If PCRE_DOLLAR_ENDONLY is set and PCRE_MULTILINE is not set, the $
4652: meta-character matches only at the very end of the string.
4653:
4654: (c) If PCRE_EXTRA is set, a backslash followed by a letter with no spe-
4655: cial meaning is faulted. Otherwise, like Perl, the backslash is quietly
4656: ignored. (Perl can be made to issue a warning.)
4657:
4658: (d) If PCRE_UNGREEDY is set, the greediness of the repetition quanti-
4659: fiers is inverted, that is, by default they are not greedy, but if fol-
4660: lowed by a question mark they are.
4661:
4662: (e) PCRE_ANCHORED can be used at matching time to force a pattern to be
4663: tried only at the first matching position in the subject string.
4664:
4665: (f) The PCRE_NOTBOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NOTEMPTY_ATSTART,
4666: and PCRE_NO_AUTO_CAPTURE options for pcre_exec() have no Perl equiva-
4667: lents.
4668:
4669: (g) The \R escape sequence can be restricted to match only CR, LF, or
4670: CRLF by the PCRE_BSR_ANYCRLF option.
4671:
4672: (h) The callout facility is PCRE-specific.
4673:
4674: (i) The partial matching facility is PCRE-specific.
4675:
4676: (j) Patterns compiled by PCRE can be saved and re-used at a later time,
4677: even on different hosts that have the other endianness. However, this
4678: does not apply to optimized data created by the just-in-time compiler.
4679:
4680: (k) The alternative matching functions (pcre_dfa_exec(),
4681: pcre16_dfa_exec() and pcre32_dfa_exec(),) match in a different way and
4682: are not Perl-compatible.
4683:
4684: (l) PCRE recognizes some special sequences such as (*CR) at the start
4685: of a pattern that set overall options that cannot be changed within the
4686: pattern.
4687:
4688:
4689: AUTHOR
4690:
4691: Philip Hazel
4692: University Computing Service
4693: Cambridge CB2 3QH, England.
4694:
4695:
4696: REVISION
4697:
4698: Last updated: 10 November 2013
4699: Copyright (c) 1997-2013 University of Cambridge.
4700: ------------------------------------------------------------------------------
4701:
4702:
4703: PCREPATTERN(3) Library Functions Manual PCREPATTERN(3)
4704:
4705:
4706:
4707: NAME
4708: PCRE - Perl-compatible regular expressions
4709:
4710: PCRE REGULAR EXPRESSION DETAILS
4711:
4712: The syntax and semantics of the regular expressions that are supported
4713: by PCRE are described in detail below. There is a quick-reference syn-
4714: tax summary in the pcresyntax page. PCRE tries to match Perl syntax and
4715: semantics as closely as it can. PCRE also supports some alternative
4716: regular expression syntax (which does not conflict with the Perl syn-
4717: tax) in order to provide some compatibility with regular expressions in
4718: Python, .NET, and Oniguruma.
4719:
4720: Perl's regular expressions are described in its own documentation, and
4721: regular expressions in general are covered in a number of books, some
4722: of which have copious examples. Jeffrey Friedl's "Mastering Regular
4723: Expressions", published by O'Reilly, covers regular expressions in
4724: great detail. This description of PCRE's regular expressions is
4725: intended as reference material.
4726:
4727: This document discusses the patterns that are supported by PCRE when
4728: one its main matching functions, pcre_exec() (8-bit) or
4729: pcre[16|32]_exec() (16- or 32-bit), is used. PCRE also has alternative
4730: matching functions, pcre_dfa_exec() and pcre[16|32_dfa_exec(), which
4731: match using a different algorithm that is not Perl-compatible. Some of
4732: the features discussed below are not available when DFA matching is
4733: used. The advantages and disadvantages of the alternative functions,
4734: and how they differ from the normal functions, are discussed in the
4735: pcrematching page.
4736:
4737:
4738: SPECIAL START-OF-PATTERN ITEMS
4739:
4740: A number of options that can be passed to pcre_compile() can also be
4741: set by special items at the start of a pattern. These are not Perl-com-
4742: patible, but are provided to make these options accessible to pattern
4743: writers who are not able to change the program that processes the pat-
4744: tern. Any number of these items may appear, but they must all be
4745: together right at the start of the pattern string, and the letters must
4746: be in upper case.
4747:
4748: UTF support
4749:
4750: The original operation of PCRE was on strings of one-byte characters.
4751: However, there is now also support for UTF-8 strings in the original
4752: library, an extra library that supports 16-bit and UTF-16 character
4753: strings, and a third library that supports 32-bit and UTF-32 character
4754: strings. To use these features, PCRE must be built to include appropri-
4755: ate support. When using UTF strings you must either call the compiling
4756: function with the PCRE_UTF8, PCRE_UTF16, or PCRE_UTF32 option, or the
4757: pattern must start with one of these special sequences:
4758:
4759: (*UTF8)
4760: (*UTF16)
4761: (*UTF32)
4762: (*UTF)
4763:
4764: (*UTF) is a generic sequence that can be used with any of the
4765: libraries. Starting a pattern with such a sequence is equivalent to
4766: setting the relevant option. How setting a UTF mode affects pattern
4767: matching is mentioned in several places below. There is also a summary
4768: of features in the pcreunicode page.
4769:
4770: Some applications that allow their users to supply patterns may wish to
4771: restrict them to non-UTF data for security reasons. If the
4772: PCRE_NEVER_UTF option is set at compile time, (*UTF) etc. are not
4773: allowed, and their appearance causes an error.
4774:
4775: Unicode property support
4776:
4777: Another special sequence that may appear at the start of a pattern is
4778: (*UCP). This has the same effect as setting the PCRE_UCP option: it
4779: causes sequences such as \d and \w to use Unicode properties to deter-
4780: mine character types, instead of recognizing only characters with codes
4781: less than 128 via a lookup table.
4782:
4783: Disabling auto-possessification
4784:
4785: If a pattern starts with (*NO_AUTO_POSSESS), it has the same effect as
4786: setting the PCRE_NO_AUTO_POSSESS option at compile time. This stops
4787: PCRE from making quantifiers possessive when what follows cannot match
4788: the repeated item. For example, by default a+b is treated as a++b. For
4789: more details, see the pcreapi documentation.
4790:
4791: Disabling start-up optimizations
4792:
4793: If a pattern starts with (*NO_START_OPT), it has the same effect as
4794: setting the PCRE_NO_START_OPTIMIZE option either at compile or matching
4795: time. This disables several optimizations for quickly reaching "no
4796: match" results. For more details, see the pcreapi documentation.
4797:
4798: Newline conventions
4799:
4800: PCRE supports five different conventions for indicating line breaks in
4801: strings: a single CR (carriage return) character, a single LF (line-
4802: feed) character, the two-character sequence CRLF, any of the three pre-
4803: ceding, or any Unicode newline sequence. The pcreapi page has further
4804: discussion about newlines, and shows how to set the newline convention
4805: in the options arguments for the compiling and matching functions.
4806:
4807: It is also possible to specify a newline convention by starting a pat-
4808: tern string with one of the following five sequences:
4809:
4810: (*CR) carriage return
4811: (*LF) linefeed
4812: (*CRLF) carriage return, followed by linefeed
4813: (*ANYCRLF) any of the three above
4814: (*ANY) all Unicode newline sequences
4815:
4816: These override the default and the options given to the compiling func-
4817: tion. For example, on a Unix system where LF is the default newline
4818: sequence, the pattern
4819:
4820: (*CR)a.b
4821:
4822: changes the convention to CR. That pattern matches "a\nb" because LF is
4823: no longer a newline. If more than one of these settings is present, the
4824: last one is used.
4825:
4826: The newline convention affects where the circumflex and dollar asser-
4827: tions are true. It also affects the interpretation of the dot metachar-
4828: acter when PCRE_DOTALL is not set, and the behaviour of \N. However, it
4829: does not affect what the \R escape sequence matches. By default, this
4830: is any Unicode newline sequence, for Perl compatibility. However, this
4831: can be changed; see the description of \R in the section entitled "New-
4832: line sequences" below. A change of \R setting can be combined with a
4833: change of newline convention.
4834:
4835: Setting match and recursion limits
4836:
4837: The caller of pcre_exec() can set a limit on the number of times the
4838: internal match() function is called and on the maximum depth of recur-
4839: sive calls. These facilities are provided to catch runaway matches that
4840: are provoked by patterns with huge matching trees (a typical example is
4841: a pattern with nested unlimited repeats) and to avoid running out of
4842: system stack by too much recursion. When one of these limits is
4843: reached, pcre_exec() gives an error return. The limits can also be set
4844: by items at the start of the pattern of the form
4845:
4846: (*LIMIT_MATCH=d)
4847: (*LIMIT_RECURSION=d)
4848:
4849: where d is any number of decimal digits. However, the value of the set-
4850: ting must be less than the value set (or defaulted) by the caller of
4851: pcre_exec() for it to have any effect. In other words, the pattern
4852: writer can lower the limits set by the programmer, but not raise them.
4853: If there is more than one setting of one of these limits, the lower
4854: value is used.
4855:
4856:
4857: EBCDIC CHARACTER CODES
4858:
4859: PCRE can be compiled to run in an environment that uses EBCDIC as its
4860: character code rather than ASCII or Unicode (typically a mainframe sys-
4861: tem). In the sections below, character code values are ASCII or Uni-
4862: code; in an EBCDIC environment these characters may have different code
4863: values, and there are no code points greater than 255.
4864:
4865:
4866: CHARACTERS AND METACHARACTERS
4867:
4868: A regular expression is a pattern that is matched against a subject
4869: string from left to right. Most characters stand for themselves in a
4870: pattern, and match the corresponding characters in the subject. As a
4871: trivial example, the pattern
4872:
4873: The quick brown fox
4874:
4875: matches a portion of a subject string that is identical to itself. When
4876: caseless matching is specified (the PCRE_CASELESS option), letters are
4877: matched independently of case. In a UTF mode, PCRE always understands
4878: the concept of case for characters whose values are less than 128, so
4879: caseless matching is always possible. For characters with higher val-
4880: ues, the concept of case is supported if PCRE is compiled with Unicode
4881: property support, but not otherwise. If you want to use caseless
4882: matching for characters 128 and above, you must ensure that PCRE is
4883: compiled with Unicode property support as well as with UTF support.
4884:
4885: The power of regular expressions comes from the ability to include
4886: alternatives and repetitions in the pattern. These are encoded in the
4887: pattern by the use of metacharacters, which do not stand for themselves
4888: but instead are interpreted in some special way.
4889:
4890: There are two different sets of metacharacters: those that are recog-
4891: nized anywhere in the pattern except within square brackets, and those
4892: that are recognized within square brackets. Outside square brackets,
4893: the metacharacters are as follows:
4894:
4895: \ general escape character with several uses
4896: ^ assert start of string (or line, in multiline mode)
4897: $ assert end of string (or line, in multiline mode)
4898: . match any character except newline (by default)
4899: [ start character class definition
4900: | start of alternative branch
4901: ( start subpattern
4902: ) end subpattern
4903: ? extends the meaning of (
4904: also 0 or 1 quantifier
4905: also quantifier minimizer
4906: * 0 or more quantifier
4907: + 1 or more quantifier
4908: also "possessive quantifier"
4909: { start min/max quantifier
4910:
4911: Part of a pattern that is in square brackets is called a "character
4912: class". In a character class the only metacharacters are:
4913:
4914: \ general escape character
4915: ^ negate the class, but only if the first character
4916: - indicates character range
4917: [ POSIX character class (only if followed by POSIX
4918: syntax)
4919: ] terminates the character class
4920:
4921: The following sections describe the use of each of the metacharacters.
4922:
4923:
4924: BACKSLASH
4925:
4926: The backslash character has several uses. Firstly, if it is followed by
4927: a character that is not a number or a letter, it takes away any special
4928: meaning that character may have. This use of backslash as an escape
4929: character applies both inside and outside character classes.
4930:
4931: For example, if you want to match a * character, you write \* in the
4932: pattern. This escaping action applies whether or not the following
4933: character would otherwise be interpreted as a metacharacter, so it is
4934: always safe to precede a non-alphanumeric with backslash to specify
4935: that it stands for itself. In particular, if you want to match a back-
4936: slash, you write \\.
4937:
4938: In a UTF mode, only ASCII numbers and letters have any special meaning
4939: after a backslash. All other characters (in particular, those whose
4940: codepoints are greater than 127) are treated as literals.
4941:
4942: If a pattern is compiled with the PCRE_EXTENDED option, most white
4943: space in the pattern (other than in a character class), and characters
4944: between a # outside a character class and the next newline, inclusive,
4945: are ignored. An escaping backslash can be used to include a white space
4946: or # character as part of the pattern.
4947:
4948: If you want to remove the special meaning from a sequence of charac-
4949: ters, you can do so by putting them between \Q and \E. This is differ-
4950: ent from Perl in that $ and @ are handled as literals in \Q...\E
4951: sequences in PCRE, whereas in Perl, $ and @ cause variable interpola-
4952: tion. Note the following examples:
4953:
4954: Pattern PCRE matches Perl matches
4955:
4956: \Qabc$xyz\E abc$xyz abc followed by the
4957: contents of $xyz
4958: \Qabc\$xyz\E abc\$xyz abc\$xyz
4959: \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
4960:
4961: The \Q...\E sequence is recognized both inside and outside character
4962: classes. An isolated \E that is not preceded by \Q is ignored. If \Q
4963: is not followed by \E later in the pattern, the literal interpretation
4964: continues to the end of the pattern (that is, \E is assumed at the
4965: end). If the isolated \Q is inside a character class, this causes an
4966: error, because the character class is not terminated.
4967:
4968: Non-printing characters
4969:
4970: A second use of backslash provides a way of encoding non-printing char-
4971: acters in patterns in a visible manner. There is no restriction on the
4972: appearance of non-printing characters, apart from the binary zero that
4973: terminates a pattern, but when a pattern is being prepared by text
4974: editing, it is often easier to use one of the following escape
4975: sequences than the binary character it represents:
4976:
4977: \a alarm, that is, the BEL character (hex 07)
4978: \cx "control-x", where x is any ASCII character
4979: \e escape (hex 1B)
4980: \f form feed (hex 0C)
4981: \n linefeed (hex 0A)
4982: \r carriage return (hex 0D)
4983: \t tab (hex 09)
4984: \0dd character with octal code 0dd
4985: \ddd character with octal code ddd, or back reference
4986: \o{ddd..} character with octal code ddd..
4987: \xhh character with hex code hh
4988: \x{hhh..} character with hex code hhh.. (non-JavaScript mode)
4989: \uhhhh character with hex code hhhh (JavaScript mode only)
4990:
4991: The precise effect of \cx on ASCII characters is as follows: if x is a
4992: lower case letter, it is converted to upper case. Then bit 6 of the
4993: character (hex 40) is inverted. Thus \cA to \cZ become hex 01 to hex 1A
4994: (A is 41, Z is 5A), but \c{ becomes hex 3B ({ is 7B), and \c; becomes
4995: hex 7B (; is 3B). If the data item (byte or 16-bit value) following \c
4996: has a value greater than 127, a compile-time error occurs. This locks
4997: out non-ASCII characters in all modes.
4998:
4999: The \c facility was designed for use with ASCII characters, but with
5000: the extension to Unicode it is even less useful than it once was. It
5001: is, however, recognized when PCRE is compiled in EBCDIC mode, where
5002: data items are always bytes. In this mode, all values are valid after
5003: \c. If the next character is a lower case letter, it is converted to
5004: upper case. Then the 0xc0 bits of the byte are inverted. Thus \cA
5005: becomes hex 01, as in ASCII (A is C1), but because the EBCDIC letters
5006: are disjoint, \cZ becomes hex 29 (Z is E9), and other characters also
5007: generate different values.
5008:
5009: After \0 up to two further octal digits are read. If there are fewer
5010: than two digits, just those that are present are used. Thus the
5011: sequence \0\x\07 specifies two binary zeros followed by a BEL character
5012: (code value 7). Make sure you supply two digits after the initial zero
5013: if the pattern character that follows is itself an octal digit.
5014:
5015: The escape \o must be followed by a sequence of octal digits, enclosed
5016: in braces. An error occurs if this is not the case. This escape is a
5017: recent addition to Perl; it provides way of specifying character code
5018: points as octal numbers greater than 0777, and it also allows octal
5019: numbers and back references to be unambiguously specified.
5020:
5021: For greater clarity and unambiguity, it is best to avoid following \ by
5022: a digit greater than zero. Instead, use \o{} or \x{} to specify charac-
5023: ter numbers, and \g{} to specify back references. The following para-
5024: graphs describe the old, ambiguous syntax.
5025:
5026: The handling of a backslash followed by a digit other than 0 is compli-
5027: cated, and Perl has changed in recent releases, causing PCRE also to
5028: change. Outside a character class, PCRE reads the digit and any follow-
5029: ing digits as a decimal number. If the number is less than 8, or if
5030: there have been at least that many previous capturing left parentheses
5031: in the expression, the entire sequence is taken as a back reference. A
5032: description of how this works is given later, following the discussion
5033: of parenthesized subpatterns.
5034:
5035: Inside a character class, or if the decimal number following \ is
5036: greater than 7 and there have not been that many capturing subpatterns,
5037: PCRE handles \8 and \9 as the literal characters "8" and "9", and oth-
5038: erwise re-reads up to three octal digits following the backslash, using
5039: them to generate a data character. Any subsequent digits stand for
5040: themselves. For example:
5041:
5042: \040 is another way of writing an ASCII space
5043: \40 is the same, provided there are fewer than 40
5044: previous capturing subpatterns
5045: \7 is always a back reference
5046: \11 might be a back reference, or another way of
5047: writing a tab
5048: \011 is always a tab
5049: \0113 is a tab followed by the character "3"
5050: \113 might be a back reference, otherwise the
5051: character with octal code 113
5052: \377 might be a back reference, otherwise
5053: the value 255 (decimal)
5054: \81 is either a back reference, or the two
5055: characters "8" and "1"
5056:
5057: Note that octal values of 100 or greater that are specified using this
5058: syntax must not be introduced by a leading zero, because no more than
5059: three octal digits are ever read.
5060:
5061: By default, after \x that is not followed by {, from zero to two hexa-
5062: decimal digits are read (letters can be in upper or lower case). Any
5063: number of hexadecimal digits may appear between \x{ and }. If a charac-
5064: ter other than a hexadecimal digit appears between \x{ and }, or if
5065: there is no terminating }, an error occurs.
5066:
5067: If the PCRE_JAVASCRIPT_COMPAT option is set, the interpretation of \x
5068: is as just described only when it is followed by two hexadecimal dig-
5069: its. Otherwise, it matches a literal "x" character. In JavaScript
5070: mode, support for code points greater than 256 is provided by \u, which
5071: must be followed by four hexadecimal digits; otherwise it matches a
5072: literal "u" character.
5073:
5074: Characters whose value is less than 256 can be defined by either of the
5075: two syntaxes for \x (or by \u in JavaScript mode). There is no differ-
5076: ence in the way they are handled. For example, \xdc is exactly the same
5077: as \x{dc} (or \u00dc in JavaScript mode).
5078:
5079: Constraints on character values
5080:
5081: Characters that are specified using octal or hexadecimal numbers are
5082: limited to certain values, as follows:
5083:
5084: 8-bit non-UTF mode less than 0x100
5085: 8-bit UTF-8 mode less than 0x10ffff and a valid codepoint
5086: 16-bit non-UTF mode less than 0x10000
5087: 16-bit UTF-16 mode less than 0x10ffff and a valid codepoint
5088: 32-bit non-UTF mode less than 0x100000000
5089: 32-bit UTF-32 mode less than 0x10ffff and a valid codepoint
5090:
5091: Invalid Unicode codepoints are the range 0xd800 to 0xdfff (the so-
5092: called "surrogate" codepoints), and 0xffef.
5093:
5094: Escape sequences in character classes
5095:
5096: All the sequences that define a single character value can be used both
5097: inside and outside character classes. In addition, inside a character
5098: class, \b is interpreted as the backspace character (hex 08).
5099:
5100: \N is not allowed in a character class. \B, \R, and \X are not special
5101: inside a character class. Like other unrecognized escape sequences,
5102: they are treated as the literal characters "B", "R", and "X" by
5103: default, but cause an error if the PCRE_EXTRA option is set. Outside a
5104: character class, these sequences have different meanings.
5105:
5106: Unsupported escape sequences
5107:
5108: In Perl, the sequences \l, \L, \u, and \U are recognized by its string
5109: handler and used to modify the case of following characters. By
5110: default, PCRE does not support these escape sequences. However, if the
5111: PCRE_JAVASCRIPT_COMPAT option is set, \U matches a "U" character, and
5112: \u can be used to define a character by code point, as described in the
5113: previous section.
5114:
5115: Absolute and relative back references
5116:
5117: The sequence \g followed by an unsigned or a negative number, option-
5118: ally enclosed in braces, is an absolute or relative back reference. A
5119: named back reference can be coded as \g{name}. Back references are dis-
5120: cussed later, following the discussion of parenthesized subpatterns.
5121:
5122: Absolute and relative subroutine calls
5123:
5124: For compatibility with Oniguruma, the non-Perl syntax \g followed by a
5125: name or a number enclosed either in angle brackets or single quotes, is
5126: an alternative syntax for referencing a subpattern as a "subroutine".
5127: Details are discussed later. Note that \g{...} (Perl syntax) and
5128: \g<...> (Oniguruma syntax) are not synonymous. The former is a back
5129: reference; the latter is a subroutine call.
5130:
5131: Generic character types
5132:
5133: Another use of backslash is for specifying generic character types:
5134:
5135: \d any decimal digit
5136: \D any character that is not a decimal digit
5137: \h any horizontal white space character
5138: \H any character that is not a horizontal white space character
5139: \s any white space character
5140: \S any character that is not a white space character
5141: \v any vertical white space character
5142: \V any character that is not a vertical white space character
5143: \w any "word" character
5144: \W any "non-word" character
5145:
5146: There is also the single sequence \N, which matches a non-newline char-
5147: acter. This is the same as the "." metacharacter when PCRE_DOTALL is
5148: not set. Perl also uses \N to match characters by name; PCRE does not
5149: support this.
5150:
5151: Each pair of lower and upper case escape sequences partitions the com-
5152: plete set of characters into two disjoint sets. Any given character
5153: matches one, and only one, of each pair. The sequences can appear both
5154: inside and outside character classes. They each match one character of
5155: the appropriate type. If the current matching point is at the end of
5156: the subject string, all of them fail, because there is no character to
5157: match.
5158:
5159: For compatibility with Perl, \s did not used to match the VT character
5160: (code 11), which made it different from the the POSIX "space" class.
5161: However, Perl added VT at release 5.18, and PCRE followed suit at
5162: release 8.34. The default \s characters are now HT (9), LF (10), VT
5163: (11), FF (12), CR (13), and space (32), which are defined as white
5164: space in the "C" locale. This list may vary if locale-specific matching
5165: is taking place. For example, in some locales the "non-breaking space"
5166: character (\xA0) is recognized as white space, and in others the VT
5167: character is not.
5168:
5169: A "word" character is an underscore or any character that is a letter
5170: or digit. By default, the definition of letters and digits is con-
5171: trolled by PCRE's low-valued character tables, and may vary if locale-
5172: specific matching is taking place (see "Locale support" in the pcreapi
5173: page). For example, in a French locale such as "fr_FR" in Unix-like
5174: systems, or "french" in Windows, some character codes greater than 127
5175: are used for accented letters, and these are then matched by \w. The
5176: use of locales with Unicode is discouraged.
5177:
5178: By default, characters whose code points are greater than 127 never
5179: match \d, \s, or \w, and always match \D, \S, and \W, although this may
5180: vary for characters in the range 128-255 when locale-specific matching
5181: is happening. These escape sequences retain their original meanings
5182: from before Unicode support was available, mainly for efficiency rea-
5183: sons. If PCRE is compiled with Unicode property support, and the
5184: PCRE_UCP option is set, the behaviour is changed so that Unicode prop-
5185: erties are used to determine character types, as follows:
5186:
5187: \d any character that matches \p{Nd} (decimal digit)
5188: \s any character that matches \p{Z} or \h or \v
5189: \w any character that matches \p{L} or \p{N}, plus underscore
5190:
5191: The upper case escapes match the inverse sets of characters. Note that
5192: \d matches only decimal digits, whereas \w matches any Unicode digit,
5193: as well as any Unicode letter, and underscore. Note also that PCRE_UCP
5194: affects \b, and \B because they are defined in terms of \w and \W.
5195: Matching these sequences is noticeably slower when PCRE_UCP is set.
5196:
5197: The sequences \h, \H, \v, and \V are features that were added to Perl
5198: at release 5.10. In contrast to the other sequences, which match only
5199: ASCII characters by default, these always match certain high-valued
5200: code points, whether or not PCRE_UCP is set. The horizontal space char-
5201: acters are:
5202:
5203: U+0009 Horizontal tab (HT)
5204: U+0020 Space
5205: U+00A0 Non-break space
5206: U+1680 Ogham space mark
5207: U+180E Mongolian vowel separator
5208: U+2000 En quad
5209: U+2001 Em quad
5210: U+2002 En space
5211: U+2003 Em space
5212: U+2004 Three-per-em space
5213: U+2005 Four-per-em space
5214: U+2006 Six-per-em space
5215: U+2007 Figure space
5216: U+2008 Punctuation space
5217: U+2009 Thin space
5218: U+200A Hair space
5219: U+202F Narrow no-break space
5220: U+205F Medium mathematical space
5221: U+3000 Ideographic space
5222:
5223: The vertical space characters are:
5224:
5225: U+000A Linefeed (LF)
5226: U+000B Vertical tab (VT)
5227: U+000C Form feed (FF)
5228: U+000D Carriage return (CR)
5229: U+0085 Next line (NEL)
5230: U+2028 Line separator
5231: U+2029 Paragraph separator
5232:
5233: In 8-bit, non-UTF-8 mode, only the characters with codepoints less than
5234: 256 are relevant.
5235:
5236: Newline sequences
5237:
5238: Outside a character class, by default, the escape sequence \R matches
5239: any Unicode newline sequence. In 8-bit non-UTF-8 mode \R is equivalent
5240: to the following:
5241:
5242: (?>\r\n|\n|\x0b|\f|\r|\x85)
5243:
5244: This is an example of an "atomic group", details of which are given
5245: below. This particular group matches either the two-character sequence
5246: CR followed by LF, or one of the single characters LF (linefeed,
5247: U+000A), VT (vertical tab, U+000B), FF (form feed, U+000C), CR (car-
5248: riage return, U+000D), or NEL (next line, U+0085). The two-character
5249: sequence is treated as a single unit that cannot be split.
5250:
5251: In other modes, two additional characters whose codepoints are greater
5252: than 255 are added: LS (line separator, U+2028) and PS (paragraph sepa-
5253: rator, U+2029). Unicode character property support is not needed for
5254: these characters to be recognized.
5255:
5256: It is possible to restrict \R to match only CR, LF, or CRLF (instead of
5257: the complete set of Unicode line endings) by setting the option
5258: PCRE_BSR_ANYCRLF either at compile time or when the pattern is matched.
5259: (BSR is an abbrevation for "backslash R".) This can be made the default
5260: when PCRE is built; if this is the case, the other behaviour can be
5261: requested via the PCRE_BSR_UNICODE option. It is also possible to
5262: specify these settings by starting a pattern string with one of the
5263: following sequences:
5264:
5265: (*BSR_ANYCRLF) CR, LF, or CRLF only
5266: (*BSR_UNICODE) any Unicode newline sequence
5267:
5268: These override the default and the options given to the compiling func-
5269: tion, but they can themselves be overridden by options given to a
5270: matching function. Note that these special settings, which are not
5271: Perl-compatible, are recognized only at the very start of a pattern,
5272: and that they must be in upper case. If more than one of them is
5273: present, the last one is used. They can be combined with a change of
5274: newline convention; for example, a pattern can start with:
5275:
5276: (*ANY)(*BSR_ANYCRLF)
5277:
5278: They can also be combined with the (*UTF8), (*UTF16), (*UTF32), (*UTF)
5279: or (*UCP) special sequences. Inside a character class, \R is treated as
5280: an unrecognized escape sequence, and so matches the letter "R" by
5281: default, but causes an error if PCRE_EXTRA is set.
5282:
5283: Unicode character properties
5284:
5285: When PCRE is built with Unicode character property support, three addi-
5286: tional escape sequences that match characters with specific properties
5287: are available. When in 8-bit non-UTF-8 mode, these sequences are of
5288: course limited to testing characters whose codepoints are less than
5289: 256, but they do work in this mode. The extra escape sequences are:
5290:
5291: \p{xx} a character with the xx property
5292: \P{xx} a character without the xx property
5293: \X a Unicode extended grapheme cluster
5294:
5295: The property names represented by xx above are limited to the Unicode
5296: script names, the general category properties, "Any", which matches any
5297: character (including newline), and some special PCRE properties
5298: (described in the next section). Other Perl properties such as "InMu-
5299: sicalSymbols" are not currently supported by PCRE. Note that \P{Any}
5300: does not match any characters, so always causes a match failure.
5301:
5302: Sets of Unicode characters are defined as belonging to certain scripts.
5303: A character from one of these sets can be matched using a script name.
5304: For example:
5305:
5306: \p{Greek}
5307: \P{Han}
5308:
5309: Those that are not part of an identified script are lumped together as
5310: "Common". The current list of scripts is:
5311:
5312: Arabic, Armenian, Avestan, Balinese, Bamum, Batak, Bengali, Bopomofo,
5313: Brahmi, Braille, Buginese, Buhid, Canadian_Aboriginal, Carian, Chakma,
5314: Cham, Cherokee, Common, Coptic, Cuneiform, Cypriot, Cyrillic, Deseret,
5315: Devanagari, Egyptian_Hieroglyphs, Ethiopic, Georgian, Glagolitic,
5316: Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
5317: gana, Imperial_Aramaic, Inherited, Inscriptional_Pahlavi, Inscrip-
5318: tional_Parthian, Javanese, Kaithi, Kannada, Katakana, Kayah_Li,
5319: Kharoshthi, Khmer, Lao, Latin, Lepcha, Limbu, Linear_B, Lisu, Lycian,
5320: Lydian, Malayalam, Mandaic, Meetei_Mayek, Meroitic_Cursive,
5321: Meroitic_Hieroglyphs, Miao, Mongolian, Myanmar, New_Tai_Lue, Nko,
5322: Ogham, Old_Italic, Old_Persian, Old_South_Arabian, Old_Turkic,
5323: Ol_Chiki, Oriya, Osmanya, Phags_Pa, Phoenician, Rejang, Runic, Samari-
5324: tan, Saurashtra, Sharada, Shavian, Sinhala, Sora_Sompeng, Sundanese,
5325: Syloti_Nagri, Syriac, Tagalog, Tagbanwa, Tai_Le, Tai_Tham, Tai_Viet,
5326: Takri, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Vai,
5327: Yi.
5328:
5329: Each character has exactly one Unicode general category property, spec-
5330: ified by a two-letter abbreviation. For compatibility with Perl, nega-
5331: tion can be specified by including a circumflex between the opening
5332: brace and the property name. For example, \p{^Lu} is the same as
5333: \P{Lu}.
5334:
5335: If only one letter is specified with \p or \P, it includes all the gen-
5336: eral category properties that start with that letter. In this case, in
5337: the absence of negation, the curly brackets in the escape sequence are
5338: optional; these two examples have the same effect:
5339:
5340: \p{L}
5341: \pL
5342:
5343: The following general category property codes are supported:
5344:
5345: C Other
5346: Cc Control
5347: Cf Format
5348: Cn Unassigned
5349: Co Private use
5350: Cs Surrogate
5351:
5352: L Letter
5353: Ll Lower case letter
5354: Lm Modifier letter
5355: Lo Other letter
5356: Lt Title case letter
5357: Lu Upper case letter
5358:
5359: M Mark
5360: Mc Spacing mark
5361: Me Enclosing mark
5362: Mn Non-spacing mark
5363:
5364: N Number
5365: Nd Decimal number
5366: Nl Letter number
5367: No Other number
5368:
5369: P Punctuation
5370: Pc Connector punctuation
5371: Pd Dash punctuation
5372: Pe Close punctuation
5373: Pf Final punctuation
5374: Pi Initial punctuation
5375: Po Other punctuation
5376: Ps Open punctuation
5377:
5378: S Symbol
5379: Sc Currency symbol
5380: Sk Modifier symbol
5381: Sm Mathematical symbol
5382: So Other symbol
5383:
5384: Z Separator
5385: Zl Line separator
5386: Zp Paragraph separator
5387: Zs Space separator
5388:
5389: The special property L& is also supported: it matches a character that
5390: has the Lu, Ll, or Lt property, in other words, a letter that is not
5391: classified as a modifier or "other".
5392:
5393: The Cs (Surrogate) property applies only to characters in the range
5394: U+D800 to U+DFFF. Such characters are not valid in Unicode strings and
5395: so cannot be tested by PCRE, unless UTF validity checking has been
5396: turned off (see the discussion of PCRE_NO_UTF8_CHECK,
5397: PCRE_NO_UTF16_CHECK and PCRE_NO_UTF32_CHECK in the pcreapi page). Perl
5398: does not support the Cs property.
5399:
5400: The long synonyms for property names that Perl supports (such as
5401: \p{Letter}) are not supported by PCRE, nor is it permitted to prefix
5402: any of these properties with "Is".
5403:
5404: No character that is in the Unicode table has the Cn (unassigned) prop-
5405: erty. Instead, this property is assumed for any code point that is not
5406: in the Unicode table.
5407:
5408: Specifying caseless matching does not affect these escape sequences.
5409: For example, \p{Lu} always matches only upper case letters. This is
5410: different from the behaviour of current versions of Perl.
5411:
5412: Matching characters by Unicode property is not fast, because PCRE has
5413: to do a multistage table lookup in order to find a character's prop-
5414: erty. That is why the traditional escape sequences such as \d and \w do
5415: not use Unicode properties in PCRE by default, though you can make them
5416: do so by setting the PCRE_UCP option or by starting the pattern with
5417: (*UCP).
5418:
5419: Extended grapheme clusters
5420:
5421: The \X escape matches any number of Unicode characters that form an
5422: "extended grapheme cluster", and treats the sequence as an atomic group
5423: (see below). Up to and including release 8.31, PCRE matched an ear-
5424: lier, simpler definition that was equivalent to
5425:
5426: (?>\PM\pM*)
5427:
5428: That is, it matched a character without the "mark" property, followed
5429: by zero or more characters with the "mark" property. Characters with
5430: the "mark" property are typically non-spacing accents that affect the
5431: preceding character.
5432:
5433: This simple definition was extended in Unicode to include more compli-
5434: cated kinds of composite character by giving each character a grapheme
5435: breaking property, and creating rules that use these properties to
5436: define the boundaries of extended grapheme clusters. In releases of
5437: PCRE later than 8.31, \X matches one of these clusters.
5438:
5439: \X always matches at least one character. Then it decides whether to
5440: add additional characters according to the following rules for ending a
5441: cluster:
5442:
5443: 1. End at the end of the subject string.
5444:
5445: 2. Do not end between CR and LF; otherwise end after any control char-
5446: acter.
5447:
5448: 3. Do not break Hangul (a Korean script) syllable sequences. Hangul
5449: characters are of five types: L, V, T, LV, and LVT. An L character may
5450: be followed by an L, V, LV, or LVT character; an LV or V character may
5451: be followed by a V or T character; an LVT or T character may be follwed
5452: only by a T character.
5453:
5454: 4. Do not end before extending characters or spacing marks. Characters
5455: with the "mark" property always have the "extend" grapheme breaking
5456: property.
5457:
5458: 5. Do not end after prepend characters.
5459:
5460: 6. Otherwise, end the cluster.
5461:
5462: PCRE's additional properties
5463:
5464: As well as the standard Unicode properties described above, PCRE sup-
5465: ports four more that make it possible to convert traditional escape
5466: sequences such as \w and \s to use Unicode properties. PCRE uses these
5467: non-standard, non-Perl properties internally when PCRE_UCP is set. How-
5468: ever, they may also be used explicitly. These properties are:
5469:
5470: Xan Any alphanumeric character
5471: Xps Any POSIX space character
5472: Xsp Any Perl space character
5473: Xwd Any Perl "word" character
5474:
5475: Xan matches characters that have either the L (letter) or the N (num-
5476: ber) property. Xps matches the characters tab, linefeed, vertical tab,
5477: form feed, or carriage return, and any other character that has the Z
5478: (separator) property. Xsp is the same as Xps; it used to exclude ver-
5479: tical tab, for Perl compatibility, but Perl changed, and so PCRE fol-
5480: lowed at release 8.34. Xwd matches the same characters as Xan, plus
5481: underscore.
5482:
5483: There is another non-standard property, Xuc, which matches any charac-
5484: ter that can be represented by a Universal Character Name in C++ and
5485: other programming languages. These are the characters $, @, ` (grave
5486: accent), and all characters with Unicode code points greater than or
5487: equal to U+00A0, except for the surrogates U+D800 to U+DFFF. Note that
5488: most base (ASCII) characters are excluded. (Universal Character Names
5489: are of the form \uHHHH or \UHHHHHHHH where H is a hexadecimal digit.
5490: Note that the Xuc property does not match these sequences but the char-
5491: acters that they represent.)
5492:
5493: Resetting the match start
5494:
5495: The escape sequence \K causes any previously matched characters not to
5496: be included in the final matched sequence. For example, the pattern:
5497:
5498: foo\Kbar
5499:
5500: matches "foobar", but reports that it has matched "bar". This feature
5501: is similar to a lookbehind assertion (described below). However, in
5502: this case, the part of the subject before the real match does not have
5503: to be of fixed length, as lookbehind assertions do. The use of \K does
5504: not interfere with the setting of captured substrings. For example,
5505: when the pattern
5506:
5507: (foo)\Kbar
5508:
5509: matches "foobar", the first substring is still set to "foo".
5510:
5511: Perl documents that the use of \K within assertions is "not well
5512: defined". In PCRE, \K is acted upon when it occurs inside positive
5513: assertions, but is ignored in negative assertions.
5514:
5515: Simple assertions
5516:
5517: The final use of backslash is for certain simple assertions. An asser-
5518: tion specifies a condition that has to be met at a particular point in
5519: a match, without consuming any characters from the subject string. The
5520: use of subpatterns for more complicated assertions is described below.
5521: The backslashed assertions are:
5522:
5523: \b matches at a word boundary
5524: \B matches when not at a word boundary
5525: \A matches at the start of the subject
5526: \Z matches at the end of the subject
5527: also matches before a newline at the end of the subject
5528: \z matches only at the end of the subject
5529: \G matches at the first matching position in the subject
5530:
5531: Inside a character class, \b has a different meaning; it matches the
5532: backspace character. If any other of these assertions appears in a
5533: character class, by default it matches the corresponding literal char-
5534: acter (for example, \B matches the letter B). However, if the
5535: PCRE_EXTRA option is set, an "invalid escape sequence" error is gener-
5536: ated instead.
5537:
5538: A word boundary is a position in the subject string where the current
5539: character and the previous character do not both match \w or \W (i.e.
5540: one matches \w and the other matches \W), or the start or end of the
5541: string if the first or last character matches \w, respectively. In a
5542: UTF mode, the meanings of \w and \W can be changed by setting the
5543: PCRE_UCP option. When this is done, it also affects \b and \B. Neither
5544: PCRE nor Perl has a separate "start of word" or "end of word" metase-
5545: quence. However, whatever follows \b normally determines which it is.
5546: For example, the fragment \ba matches "a" at the start of a word.
5547:
5548: The \A, \Z, and \z assertions differ from the traditional circumflex
5549: and dollar (described in the next section) in that they only ever match
5550: at the very start and end of the subject string, whatever options are
5551: set. Thus, they are independent of multiline mode. These three asser-
5552: tions are not affected by the PCRE_NOTBOL or PCRE_NOTEOL options, which
5553: affect only the behaviour of the circumflex and dollar metacharacters.
5554: However, if the startoffset argument of pcre_exec() is non-zero, indi-
5555: cating that matching is to start at a point other than the beginning of
5556: the subject, \A can never match. The difference between \Z and \z is
5557: that \Z matches before a newline at the end of the string as well as at
5558: the very end, whereas \z matches only at the end.
5559:
5560: The \G assertion is true only when the current matching position is at
5561: the start point of the match, as specified by the startoffset argument
5562: of pcre_exec(). It differs from \A when the value of startoffset is
5563: non-zero. By calling pcre_exec() multiple times with appropriate argu-
5564: ments, you can mimic Perl's /g option, and it is in this kind of imple-
5565: mentation where \G can be useful.
5566:
5567: Note, however, that PCRE's interpretation of \G, as the start of the
5568: current match, is subtly different from Perl's, which defines it as the
5569: end of the previous match. In Perl, these can be different when the
5570: previously matched string was empty. Because PCRE does just one match
5571: at a time, it cannot reproduce this behaviour.
5572:
5573: If all the alternatives of a pattern begin with \G, the expression is
5574: anchored to the starting match position, and the "anchored" flag is set
5575: in the compiled regular expression.
5576:
5577:
5578: CIRCUMFLEX AND DOLLAR
5579:
5580: The circumflex and dollar metacharacters are zero-width assertions.
5581: That is, they test for a particular condition being true without con-
5582: suming any characters from the subject string.
5583:
5584: Outside a character class, in the default matching mode, the circumflex
5585: character is an assertion that is true only if the current matching
5586: point is at the start of the subject string. If the startoffset argu-
5587: ment of pcre_exec() is non-zero, circumflex can never match if the
5588: PCRE_MULTILINE option is unset. Inside a character class, circumflex
5589: has an entirely different meaning (see below).
5590:
5591: Circumflex need not be the first character of the pattern if a number
5592: of alternatives are involved, but it should be the first thing in each
5593: alternative in which it appears if the pattern is ever to match that
5594: branch. If all possible alternatives start with a circumflex, that is,
5595: if the pattern is constrained to match only at the start of the sub-
5596: ject, it is said to be an "anchored" pattern. (There are also other
5597: constructs that can cause a pattern to be anchored.)
5598:
5599: The dollar character is an assertion that is true only if the current
5600: matching point is at the end of the subject string, or immediately
5601: before a newline at the end of the string (by default). Note, however,
5602: that it does not actually match the newline. Dollar need not be the
5603: last character of the pattern if a number of alternatives are involved,
5604: but it should be the last item in any branch in which it appears. Dol-
5605: lar has no special meaning in a character class.
5606:
5607: The meaning of dollar can be changed so that it matches only at the
5608: very end of the string, by setting the PCRE_DOLLAR_ENDONLY option at
5609: compile time. This does not affect the \Z assertion.
5610:
5611: The meanings of the circumflex and dollar characters are changed if the
5612: PCRE_MULTILINE option is set. When this is the case, a circumflex
5613: matches immediately after internal newlines as well as at the start of
5614: the subject string. It does not match after a newline that ends the
5615: string. A dollar matches before any newlines in the string, as well as
5616: at the very end, when PCRE_MULTILINE is set. When newline is specified
5617: as the two-character sequence CRLF, isolated CR and LF characters do
5618: not indicate newlines.
5619:
5620: For example, the pattern /^abc$/ matches the subject string "def\nabc"
5621: (where \n represents a newline) in multiline mode, but not otherwise.
5622: Consequently, patterns that are anchored in single line mode because
5623: all branches start with ^ are not anchored in multiline mode, and a
5624: match for circumflex is possible when the startoffset argument of
5625: pcre_exec() is non-zero. The PCRE_DOLLAR_ENDONLY option is ignored if
5626: PCRE_MULTILINE is set.
5627:
5628: Note that the sequences \A, \Z, and \z can be used to match the start
5629: and end of the subject in both modes, and if all branches of a pattern
5630: start with \A it is always anchored, whether or not PCRE_MULTILINE is
5631: set.
5632:
5633:
5634: FULL STOP (PERIOD, DOT) AND \N
5635:
5636: Outside a character class, a dot in the pattern matches any one charac-
5637: ter in the subject string except (by default) a character that signi-
5638: fies the end of a line.
5639:
5640: When a line ending is defined as a single character, dot never matches
5641: that character; when the two-character sequence CRLF is used, dot does
5642: not match CR if it is immediately followed by LF, but otherwise it
5643: matches all characters (including isolated CRs and LFs). When any Uni-
5644: code line endings are being recognized, dot does not match CR or LF or
5645: any of the other line ending characters.
5646:
5647: The behaviour of dot with regard to newlines can be changed. If the
5648: PCRE_DOTALL option is set, a dot matches any one character, without
5649: exception. If the two-character sequence CRLF is present in the subject
5650: string, it takes two dots to match it.
5651:
5652: The handling of dot is entirely independent of the handling of circum-
5653: flex and dollar, the only relationship being that they both involve
5654: newlines. Dot has no special meaning in a character class.
5655:
5656: The escape sequence \N behaves like a dot, except that it is not
5657: affected by the PCRE_DOTALL option. In other words, it matches any
5658: character except one that signifies the end of a line. Perl also uses
5659: \N to match characters by name; PCRE does not support this.
5660:
5661:
5662: MATCHING A SINGLE DATA UNIT
5663:
5664: Outside a character class, the escape sequence \C matches any one data
5665: unit, whether or not a UTF mode is set. In the 8-bit library, one data
5666: unit is one byte; in the 16-bit library it is a 16-bit unit; in the
5667: 32-bit library it is a 32-bit unit. Unlike a dot, \C always matches
5668: line-ending characters. The feature is provided in Perl in order to
5669: match individual bytes in UTF-8 mode, but it is unclear how it can use-
5670: fully be used. Because \C breaks up characters into individual data
5671: units, matching one unit with \C in a UTF mode means that the rest of
5672: the string may start with a malformed UTF character. This has undefined
5673: results, because PCRE assumes that it is dealing with valid UTF strings
5674: (and by default it checks this at the start of processing unless the
5675: PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK or PCRE_NO_UTF32_CHECK option
5676: is used).
5677:
5678: PCRE does not allow \C to appear in lookbehind assertions (described
5679: below) in a UTF mode, because this would make it impossible to calcu-
5680: late the length of the lookbehind.
5681:
5682: In general, the \C escape sequence is best avoided. However, one way of
5683: using it that avoids the problem of malformed UTF characters is to use
5684: a lookahead to check the length of the next character, as in this pat-
5685: tern, which could be used with a UTF-8 string (ignore white space and
5686: line breaks):
5687:
5688: (?| (?=[\x00-\x7f])(\C) |
5689: (?=[\x80-\x{7ff}])(\C)(\C) |
5690: (?=[\x{800}-\x{ffff}])(\C)(\C)(\C) |
5691: (?=[\x{10000}-\x{1fffff}])(\C)(\C)(\C)(\C))
5692:
5693: A group that starts with (?| resets the capturing parentheses numbers
5694: in each alternative (see "Duplicate Subpattern Numbers" below). The
5695: assertions at the start of each branch check the next UTF-8 character
5696: for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The
5697: character's individual bytes are then captured by the appropriate num-
5698: ber of groups.
5699:
5700:
5701: SQUARE BRACKETS AND CHARACTER CLASSES
5702:
5703: An opening square bracket introduces a character class, terminated by a
5704: closing square bracket. A closing square bracket on its own is not spe-
5705: cial by default. However, if the PCRE_JAVASCRIPT_COMPAT option is set,
5706: a lone closing square bracket causes a compile-time error. If a closing
5707: square bracket is required as a member of the class, it should be the
5708: first data character in the class (after an initial circumflex, if
5709: present) or escaped with a backslash.
5710:
5711: A character class matches a single character in the subject. In a UTF
5712: mode, the character may be more than one data unit long. A matched
5713: character must be in the set of characters defined by the class, unless
5714: the first character in the class definition is a circumflex, in which
5715: case the subject character must not be in the set defined by the class.
5716: If a circumflex is actually required as a member of the class, ensure
5717: it is not the first character, or escape it with a backslash.
5718:
5719: For example, the character class [aeiou] matches any lower case vowel,
5720: while [^aeiou] matches any character that is not a lower case vowel.
5721: Note that a circumflex is just a convenient notation for specifying the
5722: characters that are in the class by enumerating those that are not. A
5723: class that starts with a circumflex is not an assertion; it still con-
5724: sumes a character from the subject string, and therefore it fails if
5725: the current pointer is at the end of the string.
5726:
5727: In UTF-8 (UTF-16, UTF-32) mode, characters with values greater than 255
5728: (0xffff) can be included in a class as a literal string of data units,
5729: or by using the \x{ escaping mechanism.
5730:
5731: When caseless matching is set, any letters in a class represent both
5732: their upper case and lower case versions, so for example, a caseless
5733: [aeiou] matches "A" as well as "a", and a caseless [^aeiou] does not
5734: match "A", whereas a caseful version would. In a UTF mode, PCRE always
5735: understands the concept of case for characters whose values are less
5736: than 128, so caseless matching is always possible. For characters with
5737: higher values, the concept of case is supported if PCRE is compiled
5738: with Unicode property support, but not otherwise. If you want to use
5739: caseless matching in a UTF mode for characters 128 and above, you must
5740: ensure that PCRE is compiled with Unicode property support as well as
5741: with UTF support.
5742:
5743: Characters that might indicate line breaks are never treated in any
5744: special way when matching character classes, whatever line-ending
5745: sequence is in use, and whatever setting of the PCRE_DOTALL and
5746: PCRE_MULTILINE options is used. A class such as [^a] always matches one
5747: of these characters.
5748:
5749: The minus (hyphen) character can be used to specify a range of charac-
5750: ters in a character class. For example, [d-m] matches any letter
5751: between d and m, inclusive. If a minus character is required in a
5752: class, it must be escaped with a backslash or appear in a position
5753: where it cannot be interpreted as indicating a range, typically as the
5754: first or last character in the class, or immediately after a range. For
5755: example, [b-d-z] matches letters in the range b to d, a hyphen charac-
5756: ter, or z.
5757:
5758: It is not possible to have the literal character "]" as the end charac-
5759: ter of a range. A pattern such as [W-]46] is interpreted as a class of
5760: two characters ("W" and "-") followed by a literal string "46]", so it
5761: would match "W46]" or "-46]". However, if the "]" is escaped with a
5762: backslash it is interpreted as the end of range, so [W-\]46] is inter-
5763: preted as a class containing a range followed by two other characters.
5764: The octal or hexadecimal representation of "]" can also be used to end
5765: a range.
5766:
5767: An error is generated if a POSIX character class (see below) or an
5768: escape sequence other than one that defines a single character appears
5769: at a point where a range ending character is expected. For example,
5770: [z-\xff] is valid, but [A-\d] and [A-[:digit:]] are not.
5771:
5772: Ranges operate in the collating sequence of character values. They can
5773: also be used for characters specified numerically, for example
5774: [\000-\037]. Ranges can include any characters that are valid for the
5775: current mode.
5776:
5777: If a range that includes letters is used when caseless matching is set,
5778: it matches the letters in either case. For example, [W-c] is equivalent
5779: to [][\\^_`wxyzabc], matched caselessly, and in a non-UTF mode, if
5780: character tables for a French locale are in use, [\xc8-\xcb] matches
5781: accented E characters in both cases. In UTF modes, PCRE supports the
5782: concept of case for characters with values greater than 128 only when
5783: it is compiled with Unicode property support.
5784:
5785: The character escape sequences \d, \D, \h, \H, \p, \P, \s, \S, \v, \V,
5786: \w, and \W may appear in a character class, and add the characters that
5787: they match to the class. For example, [\dABCDEF] matches any hexadeci-
5788: mal digit. In UTF modes, the PCRE_UCP option affects the meanings of
5789: \d, \s, \w and their upper case partners, just as it does when they
5790: appear outside a character class, as described in the section entitled
5791: "Generic character types" above. The escape sequence \b has a different
5792: meaning inside a character class; it matches the backspace character.
5793: The sequences \B, \N, \R, and \X are not special inside a character
5794: class. Like any other unrecognized escape sequences, they are treated
5795: as the literal characters "B", "N", "R", and "X" by default, but cause
5796: an error if the PCRE_EXTRA option is set.
5797:
5798: A circumflex can conveniently be used with the upper case character
5799: types to specify a more restricted set of characters than the matching
5800: lower case type. For example, the class [^\W_] matches any letter or
5801: digit, but not underscore, whereas [\w] includes underscore. A positive
5802: character class should be read as "something OR something OR ..." and a
5803: negative class as "NOT something AND NOT something AND NOT ...".
5804:
5805: The only metacharacters that are recognized in character classes are
5806: backslash, hyphen (only where it can be interpreted as specifying a
5807: range), circumflex (only at the start), opening square bracket (only
5808: when it can be interpreted as introducing a POSIX class name, or for a
5809: special compatibility feature - see the next two sections), and the
5810: terminating closing square bracket. However, escaping other non-
5811: alphanumeric characters does no harm.
5812:
5813:
5814: POSIX CHARACTER CLASSES
5815:
5816: Perl supports the POSIX notation for character classes. This uses names
5817: enclosed by [: and :] within the enclosing square brackets. PCRE also
5818: supports this notation. For example,
5819:
5820: [01[:alpha:]%]
5821:
5822: matches "0", "1", any alphabetic character, or "%". The supported class
5823: names are:
5824:
5825: alnum letters and digits
5826: alpha letters
5827: ascii character codes 0 - 127
5828: blank space or tab only
5829: cntrl control characters
5830: digit decimal digits (same as \d)
5831: graph printing characters, excluding space
5832: lower lower case letters
5833: print printing characters, including space
5834: punct printing characters, excluding letters and digits and space
5835: space white space (the same as \s from PCRE 8.34)
5836: upper upper case letters
5837: word "word" characters (same as \w)
5838: xdigit hexadecimal digits
5839:
5840: The default "space" characters are HT (9), LF (10), VT (11), FF (12),
5841: CR (13), and space (32). If locale-specific matching is taking place,
5842: the list of space characters may be different; there may be fewer or
5843: more of them. "Space" used to be different to \s, which did not include
5844: VT, for Perl compatibility. However, Perl changed at release 5.18, and
5845: PCRE followed at release 8.34. "Space" and \s now match the same set
5846: of characters.
5847:
5848: The name "word" is a Perl extension, and "blank" is a GNU extension
5849: from Perl 5.8. Another Perl extension is negation, which is indicated
5850: by a ^ character after the colon. For example,
5851:
5852: [12[:^digit:]]
5853:
5854: matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the
5855: POSIX syntax [.ch.] and [=ch=] where "ch" is a "collating element", but
5856: these are not supported, and an error is given if they are encountered.
5857:
5858: By default, characters with values greater than 128 do not match any of
5859: the POSIX character classes. However, if the PCRE_UCP option is passed
5860: to pcre_compile(), some of the classes are changed so that Unicode
5861: character properties are used. This is achieved by replacing certain
5862: POSIX classes by other sequences, as follows:
5863:
5864: [:alnum:] becomes \p{Xan}
5865: [:alpha:] becomes \p{L}
5866: [:blank:] becomes \h
5867: [:digit:] becomes \p{Nd}
5868: [:lower:] becomes \p{Ll}
5869: [:space:] becomes \p{Xps}
5870: [:upper:] becomes \p{Lu}
5871: [:word:] becomes \p{Xwd}
5872:
5873: Negated versions, such as [:^alpha:] use \P instead of \p. Three other
5874: POSIX classes are handled specially in UCP mode:
5875:
5876: [:graph:] This matches characters that have glyphs that mark the page
5877: when printed. In Unicode property terms, it matches all char-
5878: acters with the L, M, N, P, S, or Cf properties, except for:
5879:
5880: U+061C Arabic Letter Mark
5881: U+180E Mongolian Vowel Separator
5882: U+2066 - U+2069 Various "isolate"s
5883:
5884:
5885: [:print:] This matches the same characters as [:graph:] plus space
5886: characters that are not controls, that is, characters with
5887: the Zs property.
5888:
5889: [:punct:] This matches all characters that have the Unicode P (punctua-
5890: tion) property, plus those characters whose code points are
5891: less than 128 that have the S (Symbol) property.
5892:
5893: The other POSIX classes are unchanged, and match only characters with
5894: code points less than 128.
5895:
5896:
5897: COMPATIBILITY FEATURE FOR WORD BOUNDARIES
5898:
5899: In the POSIX.2 compliant library that was included in 4.4BSD Unix, the
5900: ugly syntax [[:<:]] and [[:>:]] is used for matching "start of word"
5901: and "end of word". PCRE treats these items as follows:
5902:
5903: [[:<:]] is converted to \b(?=\w)
5904: [[:>:]] is converted to \b(?<=\w)
5905:
5906: Only these exact character sequences are recognized. A sequence such as
5907: [a[:<:]b] provokes error for an unrecognized POSIX class name. This
5908: support is not compatible with Perl. It is provided to help migrations
5909: from other environments, and is best not used in any new patterns. Note
5910: that \b matches at the start and the end of a word (see "Simple asser-
5911: tions" above), and in a Perl-style pattern the preceding or following
5912: character normally shows which is wanted, without the need for the
5913: assertions that are used above in order to give exactly the POSIX be-
5914: haviour.
5915:
5916:
5917: VERTICAL BAR
5918:
5919: Vertical bar characters are used to separate alternative patterns. For
5920: example, the pattern
5921:
5922: gilbert|sullivan
5923:
5924: matches either "gilbert" or "sullivan". Any number of alternatives may
5925: appear, and an empty alternative is permitted (matching the empty
5926: string). The matching process tries each alternative in turn, from left
5927: to right, and the first one that succeeds is used. If the alternatives
5928: are within a subpattern (defined below), "succeeds" means matching the
5929: rest of the main pattern as well as the alternative in the subpattern.
5930:
5931:
5932: INTERNAL OPTION SETTING
5933:
5934: The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
5935: PCRE_EXTENDED options (which are Perl-compatible) can be changed from
5936: within the pattern by a sequence of Perl option letters enclosed
5937: between "(?" and ")". The option letters are
5938:
5939: i for PCRE_CASELESS
5940: m for PCRE_MULTILINE
5941: s for PCRE_DOTALL
5942: x for PCRE_EXTENDED
5943:
5944: For example, (?im) sets caseless, multiline matching. It is also possi-
5945: ble to unset these options by preceding the letter with a hyphen, and a
5946: combined setting and unsetting such as (?im-sx), which sets PCRE_CASE-
5947: LESS and PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED,
5948: is also permitted. If a letter appears both before and after the
5949: hyphen, the option is unset.
5950:
5951: The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA
5952: can be changed in the same way as the Perl-compatible options by using
5953: the characters J, U and X respectively.
5954:
5955: When one of these option changes occurs at top level (that is, not
5956: inside subpattern parentheses), the change applies to the remainder of
5957: the pattern that follows. If the change is placed right at the start of
5958: a pattern, PCRE extracts it into the global options (and it will there-
5959: fore show up in data extracted by the pcre_fullinfo() function).
5960:
5961: An option change within a subpattern (see below for a description of
5962: subpatterns) affects only that part of the subpattern that follows it,
5963: so
5964:
5965: (a(?i)b)c
5966:
5967: matches abc and aBc and no other strings (assuming PCRE_CASELESS is not
5968: used). By this means, options can be made to have different settings
5969: in different parts of the pattern. Any changes made in one alternative
5970: do carry on into subsequent branches within the same subpattern. For
5971: example,
5972:
5973: (a(?i)b|c)
5974:
5975: matches "ab", "aB", "c", and "C", even though when matching "C" the
5976: first branch is abandoned before the option setting. This is because
5977: the effects of option settings happen at compile time. There would be
5978: some very weird behaviour otherwise.
5979:
5980: Note: There are other PCRE-specific options that can be set by the
5981: application when the compiling or matching functions are called. In
5982: some cases the pattern can contain special leading sequences such as
5983: (*CRLF) to override what the application has set or what has been
5984: defaulted. Details are given in the section entitled "Newline
5985: sequences" above. There are also the (*UTF8), (*UTF16),(*UTF32), and
5986: (*UCP) leading sequences that can be used to set UTF and Unicode prop-
5987: erty modes; they are equivalent to setting the PCRE_UTF8, PCRE_UTF16,
5988: PCRE_UTF32 and the PCRE_UCP options, respectively. The (*UTF) sequence
5989: is a generic version that can be used with any of the libraries. How-
5990: ever, the application can set the PCRE_NEVER_UTF option, which locks
5991: out the use of the (*UTF) sequences.
5992:
5993:
5994: SUBPATTERNS
5995:
5996: Subpatterns are delimited by parentheses (round brackets), which can be
5997: nested. Turning part of a pattern into a subpattern does two things:
5998:
5999: 1. It localizes a set of alternatives. For example, the pattern
6000:
6001: cat(aract|erpillar|)
6002:
6003: matches "cataract", "caterpillar", or "cat". Without the parentheses,
6004: it would match "cataract", "erpillar" or an empty string.
6005:
6006: 2. It sets up the subpattern as a capturing subpattern. This means
6007: that, when the whole pattern matches, that portion of the subject
6008: string that matched the subpattern is passed back to the caller via the
6009: ovector argument of the matching function. (This applies only to the
6010: traditional matching functions; the DFA matching functions do not sup-
6011: port capturing.)
6012:
6013: Opening parentheses are counted from left to right (starting from 1) to
6014: obtain numbers for the capturing subpatterns. For example, if the
6015: string "the red king" is matched against the pattern
6016:
6017: the ((red|white) (king|queen))
6018:
6019: the captured substrings are "red king", "red", and "king", and are num-
6020: bered 1, 2, and 3, respectively.
6021:
6022: The fact that plain parentheses fulfil two functions is not always
6023: helpful. There are often times when a grouping subpattern is required
6024: without a capturing requirement. If an opening parenthesis is followed
6025: by a question mark and a colon, the subpattern does not do any captur-
6026: ing, and is not counted when computing the number of any subsequent
6027: capturing subpatterns. For example, if the string "the white queen" is
6028: matched against the pattern
6029:
6030: the ((?:red|white) (king|queen))
6031:
6032: the captured substrings are "white queen" and "queen", and are numbered
6033: 1 and 2. The maximum number of capturing subpatterns is 65535.
6034:
6035: As a convenient shorthand, if any option settings are required at the
6036: start of a non-capturing subpattern, the option letters may appear
6037: between the "?" and the ":". Thus the two patterns
6038:
6039: (?i:saturday|sunday)
6040: (?:(?i)saturday|sunday)
6041:
6042: match exactly the same set of strings. Because alternative branches are
6043: tried from left to right, and options are not reset until the end of
6044: the subpattern is reached, an option setting in one branch does affect
6045: subsequent branches, so the above patterns match "SUNDAY" as well as
6046: "Saturday".
6047:
6048:
6049: DUPLICATE SUBPATTERN NUMBERS
6050:
6051: Perl 5.10 introduced a feature whereby each alternative in a subpattern
6052: uses the same numbers for its capturing parentheses. Such a subpattern
6053: starts with (?| and is itself a non-capturing subpattern. For example,
6054: consider this pattern:
6055:
6056: (?|(Sat)ur|(Sun))day
6057:
6058: Because the two alternatives are inside a (?| group, both sets of cap-
6059: turing parentheses are numbered one. Thus, when the pattern matches,
6060: you can look at captured substring number one, whichever alternative
6061: matched. This construct is useful when you want to capture part, but
6062: not all, of one of a number of alternatives. Inside a (?| group, paren-
6063: theses are numbered as usual, but the number is reset at the start of
6064: each branch. The numbers of any capturing parentheses that follow the
6065: subpattern start after the highest number used in any branch. The fol-
6066: lowing example is taken from the Perl documentation. The numbers under-
6067: neath show in which buffer the captured content will be stored.
6068:
6069: # before ---------------branch-reset----------- after
6070: / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
6071: # 1 2 2 3 2 3 4
6072:
6073: A back reference to a numbered subpattern uses the most recent value
6074: that is set for that number by any subpattern. The following pattern
6075: matches "abcabc" or "defdef":
6076:
6077: /(?|(abc)|(def))\1/
6078:
6079: In contrast, a subroutine call to a numbered subpattern always refers
6080: to the first one in the pattern with the given number. The following
6081: pattern matches "abcabc" or "defabc":
6082:
6083: /(?|(abc)|(def))(?1)/
6084:
6085: If a condition test for a subpattern's having matched refers to a non-
6086: unique number, the test is true if any of the subpatterns of that num-
6087: ber have matched.
6088:
6089: An alternative approach to using this "branch reset" feature is to use
6090: duplicate named subpatterns, as described in the next section.
6091:
6092:
6093: NAMED SUBPATTERNS
6094:
6095: Identifying capturing parentheses by number is simple, but it can be
6096: very hard to keep track of the numbers in complicated regular expres-
6097: sions. Furthermore, if an expression is modified, the numbers may
6098: change. To help with this difficulty, PCRE supports the naming of sub-
6099: patterns. This feature was not added to Perl until release 5.10. Python
6100: had the feature earlier, and PCRE introduced it at release 4.0, using
6101: the Python syntax. PCRE now supports both the Perl and the Python syn-
6102: tax. Perl allows identically numbered subpatterns to have different
6103: names, but PCRE does not.
6104:
6105: In PCRE, a subpattern can be named in one of three ways: (?<name>...)
6106: or (?'name'...) as in Perl, or (?P<name>...) as in Python. References
6107: to capturing parentheses from other parts of the pattern, such as back
6108: references, recursion, and conditions, can be made by name as well as
6109: by number.
6110:
6111: Names consist of up to 32 alphanumeric characters and underscores, but
6112: must start with a non-digit. Named capturing parentheses are still
6113: allocated numbers as well as names, exactly as if the names were not
6114: present. The PCRE API provides function calls for extracting the name-
6115: to-number translation table from a compiled pattern. There is also a
6116: convenience function for extracting a captured substring by name.
6117:
6118: By default, a name must be unique within a pattern, but it is possible
6119: to relax this constraint by setting the PCRE_DUPNAMES option at compile
6120: time. (Duplicate names are also always permitted for subpatterns with
6121: the same number, set up as described in the previous section.) Dupli-
6122: cate names can be useful for patterns where only one instance of the
6123: named parentheses can match. Suppose you want to match the name of a
6124: weekday, either as a 3-letter abbreviation or as the full name, and in
6125: both cases you want to extract the abbreviation. This pattern (ignoring
6126: the line breaks) does the job:
6127:
6128: (?<DN>Mon|Fri|Sun)(?:day)?|
6129: (?<DN>Tue)(?:sday)?|
6130: (?<DN>Wed)(?:nesday)?|
6131: (?<DN>Thu)(?:rsday)?|
6132: (?<DN>Sat)(?:urday)?
6133:
6134: There are five capturing substrings, but only one is ever set after a
6135: match. (An alternative way of solving this problem is to use a "branch
6136: reset" subpattern, as described in the previous section.)
6137:
6138: The convenience function for extracting the data by name returns the
6139: substring for the first (and in this example, the only) subpattern of
6140: that name that matched. This saves searching to find which numbered
6141: subpattern it was.
6142:
6143: If you make a back reference to a non-unique named subpattern from
6144: elsewhere in the pattern, the subpatterns to which the name refers are
6145: checked in the order in which they appear in the overall pattern. The
6146: first one that is set is used for the reference. For example, this pat-
6147: tern matches both "foofoo" and "barbar" but not "foobar" or "barfoo":
6148:
6149: (?:(?<n>foo)|(?<n>bar))\k<n>
6150:
6151:
6152: If you make a subroutine call to a non-unique named subpattern, the one
6153: that corresponds to the first occurrence of the name is used. In the
6154: absence of duplicate numbers (see the previous section) this is the one
6155: with the lowest number.
6156:
6157: If you use a named reference in a condition test (see the section about
6158: conditions below), either to check whether a subpattern has matched, or
6159: to check for recursion, all subpatterns with the same name are tested.
6160: If the condition is true for any one of them, the overall condition is
6161: true. This is the same behaviour as testing by number. For further
6162: details of the interfaces for handling named subpatterns, see the
6163: pcreapi documentation.
6164:
6165: Warning: You cannot use different names to distinguish between two sub-
6166: patterns with the same number because PCRE uses only the numbers when
6167: matching. For this reason, an error is given at compile time if differ-
6168: ent names are given to subpatterns with the same number. However, you
6169: can always give the same name to subpatterns with the same number, even
6170: when PCRE_DUPNAMES is not set.
6171:
6172:
6173: REPETITION
6174:
6175: Repetition is specified by quantifiers, which can follow any of the
6176: following items:
6177:
6178: a literal data character
6179: the dot metacharacter
6180: the \C escape sequence
6181: the \X escape sequence
6182: the \R escape sequence
6183: an escape such as \d or \pL that matches a single character
6184: a character class
6185: a back reference (see next section)
6186: a parenthesized subpattern (including assertions)
6187: a subroutine call to a subpattern (recursive or otherwise)
6188:
6189: The general repetition quantifier specifies a minimum and maximum num-
6190: ber of permitted matches, by giving the two numbers in curly brackets
6191: (braces), separated by a comma. The numbers must be less than 65536,
6192: and the first must be less than or equal to the second. For example:
6193:
6194: z{2,4}
6195:
6196: matches "zz", "zzz", or "zzzz". A closing brace on its own is not a
6197: special character. If the second number is omitted, but the comma is
6198: present, there is no upper limit; if the second number and the comma
6199: are both omitted, the quantifier specifies an exact number of required
6200: matches. Thus
6201:
6202: [aeiou]{3,}
6203:
6204: matches at least 3 successive vowels, but may match many more, while
6205:
6206: \d{8}
6207:
6208: matches exactly 8 digits. An opening curly bracket that appears in a
6209: position where a quantifier is not allowed, or one that does not match
6210: the syntax of a quantifier, is taken as a literal character. For exam-
6211: ple, {,6} is not a quantifier, but a literal string of four characters.
6212:
6213: In UTF modes, quantifiers apply to characters rather than to individual
6214: data units. Thus, for example, \x{100}{2} matches two characters, each
6215: of which is represented by a two-byte sequence in a UTF-8 string. Simi-
6216: larly, \X{3} matches three Unicode extended grapheme clusters, each of
6217: which may be several data units long (and they may be of different
6218: lengths).
6219:
6220: The quantifier {0} is permitted, causing the expression to behave as if
6221: the previous item and the quantifier were not present. This may be use-
6222: ful for subpatterns that are referenced as subroutines from elsewhere
6223: in the pattern (but see also the section entitled "Defining subpatterns
6224: for use by reference only" below). Items other than subpatterns that
6225: have a {0} quantifier are omitted from the compiled pattern.
6226:
6227: For convenience, the three most common quantifiers have single-charac-
6228: ter abbreviations:
6229:
6230: * is equivalent to {0,}
6231: + is equivalent to {1,}
6232: ? is equivalent to {0,1}
6233:
6234: It is possible to construct infinite loops by following a subpattern
6235: that can match no characters with a quantifier that has no upper limit,
6236: for example:
6237:
6238: (a?)*
6239:
6240: Earlier versions of Perl and PCRE used to give an error at compile time
6241: for such patterns. However, because there are cases where this can be
6242: useful, such patterns are now accepted, but if any repetition of the
6243: subpattern does in fact match no characters, the loop is forcibly bro-
6244: ken.
6245:
6246: By default, the quantifiers are "greedy", that is, they match as much
6247: as possible (up to the maximum number of permitted times), without
6248: causing the rest of the pattern to fail. The classic example of where
6249: this gives problems is in trying to match comments in C programs. These
6250: appear between /* and */ and within the comment, individual * and /
6251: characters may appear. An attempt to match C comments by applying the
6252: pattern
6253:
6254: /\*.*\*/
6255:
6256: to the string
6257:
6258: /* first comment */ not comment /* second comment */
6259:
6260: fails, because it matches the entire string owing to the greediness of
6261: the .* item.
6262:
6263: However, if a quantifier is followed by a question mark, it ceases to
6264: be greedy, and instead matches the minimum number of times possible, so
6265: the pattern
6266:
6267: /\*.*?\*/
6268:
6269: does the right thing with the C comments. The meaning of the various
6270: quantifiers is not otherwise changed, just the preferred number of
6271: matches. Do not confuse this use of question mark with its use as a
6272: quantifier in its own right. Because it has two uses, it can sometimes
6273: appear doubled, as in
6274:
6275: \d??\d
6276:
6277: which matches one digit by preference, but can match two if that is the
6278: only way the rest of the pattern matches.
6279:
6280: If the PCRE_UNGREEDY option is set (an option that is not available in
6281: Perl), the quantifiers are not greedy by default, but individual ones
6282: can be made greedy by following them with a question mark. In other
6283: words, it inverts the default behaviour.
6284:
6285: When a parenthesized subpattern is quantified with a minimum repeat
6286: count that is greater than 1 or with a limited maximum, more memory is
6287: required for the compiled pattern, in proportion to the size of the
6288: minimum or maximum.
6289:
6290: If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equiv-
6291: alent to Perl's /s) is set, thus allowing the dot to match newlines,
6292: the pattern is implicitly anchored, because whatever follows will be
6293: tried against every character position in the subject string, so there
6294: is no point in retrying the overall match at any position after the
6295: first. PCRE normally treats such a pattern as though it were preceded
6296: by \A.
6297:
6298: In cases where it is known that the subject string contains no new-
6299: lines, it is worth setting PCRE_DOTALL in order to obtain this opti-
6300: mization, or alternatively using ^ to indicate anchoring explicitly.
6301:
6302: However, there are some cases where the optimization cannot be used.
6303: When .* is inside capturing parentheses that are the subject of a back
6304: reference elsewhere in the pattern, a match at the start may fail where
6305: a later one succeeds. Consider, for example:
6306:
6307: (.*)abc\1
6308:
6309: If the subject is "xyz123abc123" the match point is the fourth charac-
6310: ter. For this reason, such a pattern is not implicitly anchored.
6311:
6312: Another case where implicit anchoring is not applied is when the lead-
6313: ing .* is inside an atomic group. Once again, a match at the start may
6314: fail where a later one succeeds. Consider this pattern:
6315:
6316: (?>.*?a)b
6317:
6318: It matches "ab" in the subject "aab". The use of the backtracking con-
6319: trol verbs (*PRUNE) and (*SKIP) also disable this optimization.
6320:
6321: When a capturing subpattern is repeated, the value captured is the sub-
6322: string that matched the final iteration. For example, after
6323:
6324: (tweedle[dume]{3}\s*)+
6325:
6326: has matched "tweedledum tweedledee" the value of the captured substring
6327: is "tweedledee". However, if there are nested capturing subpatterns,
6328: the corresponding captured values may have been set in previous itera-
6329: tions. For example, after
6330:
6331: /(a|(b))+/
6332:
6333: matches "aba" the value of the second captured substring is "b".
6334:
6335:
6336: ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS
6337:
6338: With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
6339: repetition, failure of what follows normally causes the repeated item
6340: to be re-evaluated to see if a different number of repeats allows the
6341: rest of the pattern to match. Sometimes it is useful to prevent this,
6342: either to change the nature of the match, or to cause it fail earlier
6343: than it otherwise might, when the author of the pattern knows there is
6344: no point in carrying on.
6345:
6346: Consider, for example, the pattern \d+foo when applied to the subject
6347: line
6348:
6349: 123456bar
6350:
6351: After matching all 6 digits and then failing to match "foo", the normal
6352: action of the matcher is to try again with only 5 digits matching the
6353: \d+ item, and then with 4, and so on, before ultimately failing.
6354: "Atomic grouping" (a term taken from Jeffrey Friedl's book) provides
6355: the means for specifying that once a subpattern has matched, it is not
6356: to be re-evaluated in this way.
6357:
6358: If we use atomic grouping for the previous example, the matcher gives
6359: up immediately on failing to match "foo" the first time. The notation
6360: is a kind of special parenthesis, starting with (?> as in this example:
6361:
6362: (?>\d+)foo
6363:
6364: This kind of parenthesis "locks up" the part of the pattern it con-
6365: tains once it has matched, and a failure further into the pattern is
6366: prevented from backtracking into it. Backtracking past it to previous
6367: items, however, works as normal.
6368:
6369: An alternative description is that a subpattern of this type matches
6370: the string of characters that an identical standalone pattern would
6371: match, if anchored at the current point in the subject string.
6372:
6373: Atomic grouping subpatterns are not capturing subpatterns. Simple cases
6374: such as the above example can be thought of as a maximizing repeat that
6375: must swallow everything it can. So, while both \d+ and \d+? are pre-
6376: pared to adjust the number of digits they match in order to make the
6377: rest of the pattern match, (?>\d+) can only match an entire sequence of
6378: digits.
6379:
6380: Atomic groups in general can of course contain arbitrarily complicated
6381: subpatterns, and can be nested. However, when the subpattern for an
6382: atomic group is just a single repeated item, as in the example above, a
6383: simpler notation, called a "possessive quantifier" can be used. This
6384: consists of an additional + character following a quantifier. Using
6385: this notation, the previous example can be rewritten as
6386:
6387: \d++foo
6388:
6389: Note that a possessive quantifier can be used with an entire group, for
6390: example:
6391:
6392: (abc|xyz){2,3}+
6393:
6394: Possessive quantifiers are always greedy; the setting of the
6395: PCRE_UNGREEDY option is ignored. They are a convenient notation for the
6396: simpler forms of atomic group. However, there is no difference in the
6397: meaning of a possessive quantifier and the equivalent atomic group,
6398: though there may be a performance difference; possessive quantifiers
6399: should be slightly faster.
6400:
6401: The possessive quantifier syntax is an extension to the Perl 5.8 syn-
6402: tax. Jeffrey Friedl originated the idea (and the name) in the first
6403: edition of his book. Mike McCloskey liked it, so implemented it when he
6404: built Sun's Java package, and PCRE copied it from there. It ultimately
6405: found its way into Perl at release 5.10.
6406:
6407: PCRE has an optimization that automatically "possessifies" certain sim-
6408: ple pattern constructs. For example, the sequence A+B is treated as
6409: A++B because there is no point in backtracking into a sequence of A's
6410: when B must follow.
6411:
6412: When a pattern contains an unlimited repeat inside a subpattern that
6413: can itself be repeated an unlimited number of times, the use of an
6414: atomic group is the only way to avoid some failing matches taking a
6415: very long time indeed. The pattern
6416:
6417: (\D+|<\d+>)*[!?]
6418:
6419: matches an unlimited number of substrings that either consist of non-
6420: digits, or digits enclosed in <>, followed by either ! or ?. When it
6421: matches, it runs quickly. However, if it is applied to
6422:
6423: aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
6424:
6425: it takes a long time before reporting failure. This is because the
6426: string can be divided between the internal \D+ repeat and the external
6427: * repeat in a large number of ways, and all have to be tried. (The
6428: example uses [!?] rather than a single character at the end, because
6429: both PCRE and Perl have an optimization that allows for fast failure
6430: when a single character is used. They remember the last single charac-
6431: ter that is required for a match, and fail early if it is not present
6432: in the string.) If the pattern is changed so that it uses an atomic
6433: group, like this:
6434:
6435: ((?>\D+)|<\d+>)*[!?]
6436:
6437: sequences of non-digits cannot be broken, and failure happens quickly.
6438:
6439:
6440: BACK REFERENCES
6441:
6442: Outside a character class, a backslash followed by a digit greater than
6443: 0 (and possibly further digits) is a back reference to a capturing sub-
6444: pattern earlier (that is, to its left) in the pattern, provided there
6445: have been that many previous capturing left parentheses.
6446:
6447: However, if the decimal number following the backslash is less than 10,
6448: it is always taken as a back reference, and causes an error only if
6449: there are not that many capturing left parentheses in the entire pat-
6450: tern. In other words, the parentheses that are referenced need not be
6451: to the left of the reference for numbers less than 10. A "forward back
6452: reference" of this type can make sense when a repetition is involved
6453: and the subpattern to the right has participated in an earlier itera-
6454: tion.
6455:
6456: It is not possible to have a numerical "forward back reference" to a
6457: subpattern whose number is 10 or more using this syntax because a
6458: sequence such as \50 is interpreted as a character defined in octal.
6459: See the subsection entitled "Non-printing characters" above for further
6460: details of the handling of digits following a backslash. There is no
6461: such problem when named parentheses are used. A back reference to any
6462: subpattern is possible using named parentheses (see below).
6463:
6464: Another way of avoiding the ambiguity inherent in the use of digits
6465: following a backslash is to use the \g escape sequence. This escape
6466: must be followed by an unsigned number or a negative number, optionally
6467: enclosed in braces. These examples are all identical:
6468:
6469: (ring), \1
6470: (ring), \g1
6471: (ring), \g{1}
6472:
6473: An unsigned number specifies an absolute reference without the ambigu-
6474: ity that is present in the older syntax. It is also useful when literal
6475: digits follow the reference. A negative number is a relative reference.
6476: Consider this example:
6477:
6478: (abc(def)ghi)\g{-1}
6479:
6480: The sequence \g{-1} is a reference to the most recently started captur-
6481: ing subpattern before \g, that is, is it equivalent to \2 in this exam-
6482: ple. Similarly, \g{-2} would be equivalent to \1. The use of relative
6483: references can be helpful in long patterns, and also in patterns that
6484: are created by joining together fragments that contain references
6485: within themselves.
6486:
6487: A back reference matches whatever actually matched the capturing sub-
6488: pattern in the current subject string, rather than anything matching
6489: the subpattern itself (see "Subpatterns as subroutines" below for a way
6490: of doing that). So the pattern
6491:
6492: (sens|respons)e and \1ibility
6493:
6494: matches "sense and sensibility" and "response and responsibility", but
6495: not "sense and responsibility". If caseful matching is in force at the
6496: time of the back reference, the case of letters is relevant. For exam-
6497: ple,
6498:
6499: ((?i)rah)\s+\1
6500:
6501: matches "rah rah" and "RAH RAH", but not "RAH rah", even though the
6502: original capturing subpattern is matched caselessly.
6503:
6504: There are several different ways of writing back references to named
6505: subpatterns. The .NET syntax \k{name} and the Perl syntax \k<name> or
6506: \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's
6507: unified back reference syntax, in which \g can be used for both numeric
6508: and named references, is also supported. We could rewrite the above
6509: example in any of the following ways:
6510:
6511: (?<p1>(?i)rah)\s+\k<p1>
6512: (?'p1'(?i)rah)\s+\k{p1}
6513: (?P<p1>(?i)rah)\s+(?P=p1)
6514: (?<p1>(?i)rah)\s+\g{p1}
6515:
6516: A subpattern that is referenced by name may appear in the pattern
6517: before or after the reference.
6518:
6519: There may be more than one back reference to the same subpattern. If a
6520: subpattern has not actually been used in a particular match, any back
6521: references to it always fail by default. For example, the pattern
6522:
6523: (a|(bc))\2
6524:
6525: always fails if it starts to match "a" rather than "bc". However, if
6526: the PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back refer-
6527: ence to an unset value matches an empty string.
6528:
6529: Because there may be many capturing parentheses in a pattern, all dig-
6530: its following a backslash are taken as part of a potential back refer-
6531: ence number. If the pattern continues with a digit character, some
6532: delimiter must be used to terminate the back reference. If the
6533: PCRE_EXTENDED option is set, this can be white space. Otherwise, the
6534: \g{ syntax or an empty comment (see "Comments" below) can be used.
6535:
6536: Recursive back references
6537:
6538: A back reference that occurs inside the parentheses to which it refers
6539: fails when the subpattern is first used, so, for example, (a\1) never
6540: matches. However, such references can be useful inside repeated sub-
6541: patterns. For example, the pattern
6542:
6543: (a|b\1)+
6544:
6545: matches any number of "a"s and also "aba", "ababbaa" etc. At each iter-
6546: ation of the subpattern, the back reference matches the character
6547: string corresponding to the previous iteration. In order for this to
6548: work, the pattern must be such that the first iteration does not need
6549: to match the back reference. This can be done using alternation, as in
6550: the example above, or by a quantifier with a minimum of zero.
6551:
6552: Back references of this type cause the group that they reference to be
6553: treated as an atomic group. Once the whole group has been matched, a
6554: subsequent matching failure cannot cause backtracking into the middle
6555: of the group.
6556:
6557:
6558: ASSERTIONS
6559:
6560: An assertion is a test on the characters following or preceding the
6561: current matching point that does not actually consume any characters.
6562: The simple assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are
6563: described above.
6564:
6565: More complicated assertions are coded as subpatterns. There are two
6566: kinds: those that look ahead of the current position in the subject
6567: string, and those that look behind it. An assertion subpattern is
6568: matched in the normal way, except that it does not cause the current
6569: matching position to be changed.
6570:
6571: Assertion subpatterns are not capturing subpatterns. If such an asser-
6572: tion contains capturing subpatterns within it, these are counted for
6573: the purposes of numbering the capturing subpatterns in the whole pat-
6574: tern. However, substring capturing is carried out only for positive
6575: assertions. (Perl sometimes, but not always, does do capturing in nega-
6576: tive assertions.)
6577:
6578: For compatibility with Perl, assertion subpatterns may be repeated;
6579: though it makes no sense to assert the same thing several times, the
6580: side effect of capturing parentheses may occasionally be useful. In
6581: practice, there only three cases:
6582:
6583: (1) If the quantifier is {0}, the assertion is never obeyed during
6584: matching. However, it may contain internal capturing parenthesized
6585: groups that are called from elsewhere via the subroutine mechanism.
6586:
6587: (2) If quantifier is {0,n} where n is greater than zero, it is treated
6588: as if it were {0,1}. At run time, the rest of the pattern match is
6589: tried with and without the assertion, the order depending on the greed-
6590: iness of the quantifier.
6591:
6592: (3) If the minimum repetition is greater than zero, the quantifier is
6593: ignored. The assertion is obeyed just once when encountered during
6594: matching.
6595:
6596: Lookahead assertions
6597:
6598: Lookahead assertions start with (?= for positive assertions and (?! for
6599: negative assertions. For example,
6600:
6601: \w+(?=;)
6602:
6603: matches a word followed by a semicolon, but does not include the semi-
6604: colon in the match, and
6605:
6606: foo(?!bar)
6607:
6608: matches any occurrence of "foo" that is not followed by "bar". Note
6609: that the apparently similar pattern
6610:
6611: (?!foo)bar
6612:
6613: does not find an occurrence of "bar" that is preceded by something
6614: other than "foo"; it finds any occurrence of "bar" whatsoever, because
6615: the assertion (?!foo) is always true when the next three characters are
6616: "bar". A lookbehind assertion is needed to achieve the other effect.
6617:
6618: If you want to force a matching failure at some point in a pattern, the
6619: most convenient way to do it is with (?!) because an empty string
6620: always matches, so an assertion that requires there not to be an empty
6621: string must always fail. The backtracking control verb (*FAIL) or (*F)
6622: is a synonym for (?!).
6623:
6624: Lookbehind assertions
6625:
6626: Lookbehind assertions start with (?<= for positive assertions and (?<!
6627: for negative assertions. For example,
6628:
6629: (?<!foo)bar
6630:
6631: does find an occurrence of "bar" that is not preceded by "foo". The
6632: contents of a lookbehind assertion are restricted such that all the
6633: strings it matches must have a fixed length. However, if there are sev-
6634: eral top-level alternatives, they do not all have to have the same
6635: fixed length. Thus
6636:
6637: (?<=bullock|donkey)
6638:
6639: is permitted, but
6640:
6641: (?<!dogs?|cats?)
6642:
6643: causes an error at compile time. Branches that match different length
6644: strings are permitted only at the top level of a lookbehind assertion.
6645: This is an extension compared with Perl, which requires all branches to
6646: match the same length of string. An assertion such as
6647:
6648: (?<=ab(c|de))
6649:
6650: is not permitted, because its single top-level branch can match two
6651: different lengths, but it is acceptable to PCRE if rewritten to use two
6652: top-level branches:
6653:
6654: (?<=abc|abde)
6655:
6656: In some cases, the escape sequence \K (see above) can be used instead
6657: of a lookbehind assertion to get round the fixed-length restriction.
6658:
6659: The implementation of lookbehind assertions is, for each alternative,
6660: to temporarily move the current position back by the fixed length and
6661: then try to match. If there are insufficient characters before the cur-
6662: rent position, the assertion fails.
6663:
6664: In a UTF mode, PCRE does not allow the \C escape (which matches a sin-
6665: gle data unit even in a UTF mode) to appear in lookbehind assertions,
6666: because it makes it impossible to calculate the length of the lookbe-
6667: hind. The \X and \R escapes, which can match different numbers of data
6668: units, are also not permitted.
6669:
6670: "Subroutine" calls (see below) such as (?2) or (?&X) are permitted in
6671: lookbehinds, as long as the subpattern matches a fixed-length string.
6672: Recursion, however, is not supported.
6673:
6674: Possessive quantifiers can be used in conjunction with lookbehind
6675: assertions to specify efficient matching of fixed-length strings at the
6676: end of subject strings. Consider a simple pattern such as
6677:
6678: abcd$
6679:
6680: when applied to a long string that does not match. Because matching
6681: proceeds from left to right, PCRE will look for each "a" in the subject
6682: and then see if what follows matches the rest of the pattern. If the
6683: pattern is specified as
6684:
6685: ^.*abcd$
6686:
6687: the initial .* matches the entire string at first, but when this fails
6688: (because there is no following "a"), it backtracks to match all but the
6689: last character, then all but the last two characters, and so on. Once
6690: again the search for "a" covers the entire string, from right to left,
6691: so we are no better off. However, if the pattern is written as
6692:
6693: ^.*+(?<=abcd)
6694:
6695: there can be no backtracking for the .*+ item; it can match only the
6696: entire string. The subsequent lookbehind assertion does a single test
6697: on the last four characters. If it fails, the match fails immediately.
6698: For long strings, this approach makes a significant difference to the
6699: processing time.
6700:
6701: Using multiple assertions
6702:
6703: Several assertions (of any sort) may occur in succession. For example,
6704:
6705: (?<=\d{3})(?<!999)foo
6706:
6707: matches "foo" preceded by three digits that are not "999". Notice that
6708: each of the assertions is applied independently at the same point in
6709: the subject string. First there is a check that the previous three
6710: characters are all digits, and then there is a check that the same
6711: three characters are not "999". This pattern does not match "foo" pre-
6712: ceded by six characters, the first of which are digits and the last
6713: three of which are not "999". For example, it doesn't match "123abc-
6714: foo". A pattern to do that is
6715:
6716: (?<=\d{3}...)(?<!999)foo
6717:
6718: This time the first assertion looks at the preceding six characters,
6719: checking that the first three are digits, and then the second assertion
6720: checks that the preceding three characters are not "999".
6721:
6722: Assertions can be nested in any combination. For example,
6723:
6724: (?<=(?<!foo)bar)baz
6725:
6726: matches an occurrence of "baz" that is preceded by "bar" which in turn
6727: is not preceded by "foo", while
6728:
6729: (?<=\d{3}(?!999)...)foo
6730:
6731: is another pattern that matches "foo" preceded by three digits and any
6732: three characters that are not "999".
6733:
6734:
6735: CONDITIONAL SUBPATTERNS
6736:
6737: It is possible to cause the matching process to obey a subpattern con-
6738: ditionally or to choose between two alternative subpatterns, depending
6739: on the result of an assertion, or whether a specific capturing subpat-
6740: tern has already been matched. The two possible forms of conditional
6741: subpattern are:
6742:
6743: (?(condition)yes-pattern)
6744: (?(condition)yes-pattern|no-pattern)
6745:
6746: If the condition is satisfied, the yes-pattern is used; otherwise the
6747: no-pattern (if present) is used. If there are more than two alterna-
6748: tives in the subpattern, a compile-time error occurs. Each of the two
6749: alternatives may itself contain nested subpatterns of any form, includ-
6750: ing conditional subpatterns; the restriction to two alternatives
6751: applies only at the level of the condition. This pattern fragment is an
6752: example where the alternatives are complex:
6753:
6754: (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
6755:
6756:
6757: There are four kinds of condition: references to subpatterns, refer-
6758: ences to recursion, a pseudo-condition called DEFINE, and assertions.
6759:
6760: Checking for a used subpattern by number
6761:
6762: If the text between the parentheses consists of a sequence of digits,
6763: the condition is true if a capturing subpattern of that number has pre-
6764: viously matched. If there is more than one capturing subpattern with
6765: the same number (see the earlier section about duplicate subpattern
6766: numbers), the condition is true if any of them have matched. An alter-
6767: native notation is to precede the digits with a plus or minus sign. In
6768: this case, the subpattern number is relative rather than absolute. The
6769: most recently opened parentheses can be referenced by (?(-1), the next
6770: most recent by (?(-2), and so on. Inside loops it can also make sense
6771: to refer to subsequent groups. The next parentheses to be opened can be
6772: referenced as (?(+1), and so on. (The value zero in any of these forms
6773: is not used; it provokes a compile-time error.)
6774:
6775: Consider the following pattern, which contains non-significant white
6776: space to make it more readable (assume the PCRE_EXTENDED option) and to
6777: divide it into three parts for ease of discussion:
6778:
6779: ( \( )? [^()]+ (?(1) \) )
6780:
6781: The first part matches an optional opening parenthesis, and if that
6782: character is present, sets it as the first captured substring. The sec-
6783: ond part matches one or more characters that are not parentheses. The
6784: third part is a conditional subpattern that tests whether or not the
6785: first set of parentheses matched. If they did, that is, if subject
6786: started with an opening parenthesis, the condition is true, and so the
6787: yes-pattern is executed and a closing parenthesis is required. Other-
6788: wise, since no-pattern is not present, the subpattern matches nothing.
6789: In other words, this pattern matches a sequence of non-parentheses,
6790: optionally enclosed in parentheses.
6791:
6792: If you were embedding this pattern in a larger one, you could use a
6793: relative reference:
6794:
6795: ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
6796:
6797: This makes the fragment independent of the parentheses in the larger
6798: pattern.
6799:
6800: Checking for a used subpattern by name
6801:
6802: Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a
6803: used subpattern by name. For compatibility with earlier versions of
6804: PCRE, which had this facility before Perl, the syntax (?(name)...) is
6805: also recognized.
6806:
6807: Rewriting the above example to use a named subpattern gives this:
6808:
6809: (?<OPEN> \( )? [^()]+ (?(<OPEN>) \) )
6810:
6811: If the name used in a condition of this kind is a duplicate, the test
6812: is applied to all subpatterns of the same name, and is true if any one
6813: of them has matched.
6814:
6815: Checking for pattern recursion
6816:
6817: If the condition is the string (R), and there is no subpattern with the
6818: name R, the condition is true if a recursive call to the whole pattern
6819: or any subpattern has been made. If digits or a name preceded by amper-
6820: sand follow the letter R, for example:
6821:
6822: (?(R3)...) or (?(R&name)...)
6823:
6824: the condition is true if the most recent recursion is into a subpattern
6825: whose number or name is given. This condition does not check the entire
6826: recursion stack. If the name used in a condition of this kind is a
6827: duplicate, the test is applied to all subpatterns of the same name, and
6828: is true if any one of them is the most recent recursion.
6829:
6830: At "top level", all these recursion test conditions are false. The
6831: syntax for recursive patterns is described below.
6832:
6833: Defining subpatterns for use by reference only
6834:
6835: If the condition is the string (DEFINE), and there is no subpattern
6836: with the name DEFINE, the condition is always false. In this case,
6837: there may be only one alternative in the subpattern. It is always
6838: skipped if control reaches this point in the pattern; the idea of
6839: DEFINE is that it can be used to define subroutines that can be refer-
6840: enced from elsewhere. (The use of subroutines is described below.) For
6841: example, a pattern to match an IPv4 address such as "192.168.23.245"
6842: could be written like this (ignore white space and line breaks):
6843:
6844: (?(DEFINE) (?<byte> 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
6845: \b (?&byte) (\.(?&byte)){3} \b
6846:
6847: The first part of the pattern is a DEFINE group inside which a another
6848: group named "byte" is defined. This matches an individual component of
6849: an IPv4 address (a number less than 256). When matching takes place,
6850: this part of the pattern is skipped because DEFINE acts like a false
6851: condition. The rest of the pattern uses references to the named group
6852: to match the four dot-separated components of an IPv4 address, insist-
6853: ing on a word boundary at each end.
6854:
6855: Assertion conditions
6856:
6857: If the condition is not in any of the above formats, it must be an
6858: assertion. This may be a positive or negative lookahead or lookbehind
6859: assertion. Consider this pattern, again containing non-significant
6860: white space, and with the two alternatives on the second line:
6861:
6862: (?(?=[^a-z]*[a-z])
6863: \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
6864:
6865: The condition is a positive lookahead assertion that matches an
6866: optional sequence of non-letters followed by a letter. In other words,
6867: it tests for the presence of at least one letter in the subject. If a
6868: letter is found, the subject is matched against the first alternative;
6869: otherwise it is matched against the second. This pattern matches
6870: strings in one of the two forms dd-aaa-dd or dd-dd-dd, where aaa are
6871: letters and dd are digits.
6872:
6873:
6874: COMMENTS
6875:
6876: There are two ways of including comments in patterns that are processed
6877: by PCRE. In both cases, the start of the comment must not be in a char-
6878: acter class, nor in the middle of any other sequence of related charac-
6879: ters such as (?: or a subpattern name or number. The characters that
6880: make up a comment play no part in the pattern matching.
6881:
6882: The sequence (?# marks the start of a comment that continues up to the
6883: next closing parenthesis. Nested parentheses are not permitted. If the
6884: PCRE_EXTENDED option is set, an unescaped # character also introduces a
6885: comment, which in this case continues to immediately after the next
6886: newline character or character sequence in the pattern. Which charac-
6887: ters are interpreted as newlines is controlled by the options passed to
6888: a compiling function or by a special sequence at the start of the pat-
6889: tern, as described in the section entitled "Newline conventions" above.
6890: Note that the end of this type of comment is a literal newline sequence
6891: in the pattern; escape sequences that happen to represent a newline do
6892: not count. For example, consider this pattern when PCRE_EXTENDED is
6893: set, and the default newline convention is in force:
6894:
6895: abc #comment \n still comment
6896:
6897: On encountering the # character, pcre_compile() skips along, looking
6898: for a newline in the pattern. The sequence \n is still literal at this
6899: stage, so it does not terminate the comment. Only an actual character
6900: with the code value 0x0a (the default newline) does so.
6901:
6902:
6903: RECURSIVE PATTERNS
6904:
6905: Consider the problem of matching a string in parentheses, allowing for
6906: unlimited nested parentheses. Without the use of recursion, the best
6907: that can be done is to use a pattern that matches up to some fixed
6908: depth of nesting. It is not possible to handle an arbitrary nesting
6909: depth.
6910:
6911: For some time, Perl has provided a facility that allows regular expres-
6912: sions to recurse (amongst other things). It does this by interpolating
6913: Perl code in the expression at run time, and the code can refer to the
6914: expression itself. A Perl pattern using code interpolation to solve the
6915: parentheses problem can be created like this:
6916:
6917: $re = qr{\( (?: (?>[^()]+) | (?p{$re}) )* \)}x;
6918:
6919: The (?p{...}) item interpolates Perl code at run time, and in this case
6920: refers recursively to the pattern in which it appears.
6921:
6922: Obviously, PCRE cannot support the interpolation of Perl code. Instead,
6923: it supports special syntax for recursion of the entire pattern, and
6924: also for individual subpattern recursion. After its introduction in
6925: PCRE and Python, this kind of recursion was subsequently introduced
6926: into Perl at release 5.10.
6927:
6928: A special item that consists of (? followed by a number greater than
6929: zero and a closing parenthesis is a recursive subroutine call of the
6930: subpattern of the given number, provided that it occurs inside that
6931: subpattern. (If not, it is a non-recursive subroutine call, which is
6932: described in the next section.) The special item (?R) or (?0) is a
6933: recursive call of the entire regular expression.
6934:
6935: This PCRE pattern solves the nested parentheses problem (assume the
6936: PCRE_EXTENDED option is set so that white space is ignored):
6937:
6938: \( ( [^()]++ | (?R) )* \)
6939:
6940: First it matches an opening parenthesis. Then it matches any number of
6941: substrings which can either be a sequence of non-parentheses, or a
6942: recursive match of the pattern itself (that is, a correctly parenthe-
6943: sized substring). Finally there is a closing parenthesis. Note the use
6944: of a possessive quantifier to avoid backtracking into sequences of non-
6945: parentheses.
6946:
6947: If this were part of a larger pattern, you would not want to recurse
6948: the entire pattern, so instead you could use this:
6949:
6950: ( \( ( [^()]++ | (?1) )* \) )
6951:
6952: We have put the pattern into parentheses, and caused the recursion to
6953: refer to them instead of the whole pattern.
6954:
6955: In a larger pattern, keeping track of parenthesis numbers can be
6956: tricky. This is made easier by the use of relative references. Instead
6957: of (?1) in the pattern above you can write (?-2) to refer to the second
6958: most recently opened parentheses preceding the recursion. In other
6959: words, a negative number counts capturing parentheses leftwards from
6960: the point at which it is encountered.
6961:
6962: It is also possible to refer to subsequently opened parentheses, by
6963: writing references such as (?+2). However, these cannot be recursive
6964: because the reference is not inside the parentheses that are refer-
6965: enced. They are always non-recursive subroutine calls, as described in
6966: the next section.
6967:
6968: An alternative approach is to use named parentheses instead. The Perl
6969: syntax for this is (?&name); PCRE's earlier syntax (?P>name) is also
6970: supported. We could rewrite the above example as follows:
6971:
6972: (?<pn> \( ( [^()]++ | (?&pn) )* \) )
6973:
6974: If there is more than one subpattern with the same name, the earliest
6975: one is used.
6976:
6977: This particular example pattern that we have been looking at contains
6978: nested unlimited repeats, and so the use of a possessive quantifier for
6979: matching strings of non-parentheses is important when applying the pat-
6980: tern to strings that do not match. For example, when this pattern is
6981: applied to
6982:
6983: (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
6984:
6985: it yields "no match" quickly. However, if a possessive quantifier is
6986: not used, the match runs for a very long time indeed because there are
6987: so many different ways the + and * repeats can carve up the subject,
6988: and all have to be tested before failure can be reported.
6989:
6990: At the end of a match, the values of capturing parentheses are those
6991: from the outermost level. If you want to obtain intermediate values, a
6992: callout function can be used (see below and the pcrecallout documenta-
6993: tion). If the pattern above is matched against
6994:
6995: (ab(cd)ef)
6996:
6997: the value for the inner capturing parentheses (numbered 2) is "ef",
6998: which is the last value taken on at the top level. If a capturing sub-
6999: pattern is not matched at the top level, its final captured value is
7000: unset, even if it was (temporarily) set at a deeper level during the
7001: matching process.
7002:
7003: If there are more than 15 capturing parentheses in a pattern, PCRE has
7004: to obtain extra memory to store data during a recursion, which it does
7005: by using pcre_malloc, freeing it via pcre_free afterwards. If no memory
7006: can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
7007:
7008: Do not confuse the (?R) item with the condition (R), which tests for
7009: recursion. Consider this pattern, which matches text in angle brack-
7010: ets, allowing for arbitrary nesting. Only digits are allowed in nested
7011: brackets (that is, when recursing), whereas any characters are permit-
7012: ted at the outer level.
7013:
7014: < (?: (?(R) \d++ | [^<>]*+) | (?R)) * >
7015:
7016: In this pattern, (?(R) is the start of a conditional subpattern, with
7017: two different alternatives for the recursive and non-recursive cases.
7018: The (?R) item is the actual recursive call.
7019:
7020: Differences in recursion processing between PCRE and Perl
7021:
7022: Recursion processing in PCRE differs from Perl in two important ways.
7023: In PCRE (like Python, but unlike Perl), a recursive subpattern call is
7024: always treated as an atomic group. That is, once it has matched some of
7025: the subject string, it is never re-entered, even if it contains untried
7026: alternatives and there is a subsequent matching failure. This can be
7027: illustrated by the following pattern, which purports to match a palin-
7028: dromic string that contains an odd number of characters (for example,
7029: "a", "aba", "abcba", "abcdcba"):
7030:
7031: ^(.|(.)(?1)\2)$
7032:
7033: The idea is that it either matches a single character, or two identical
7034: characters surrounding a sub-palindrome. In Perl, this pattern works;
7035: in PCRE it does not if the pattern is longer than three characters.
7036: Consider the subject string "abcba":
7037:
7038: At the top level, the first character is matched, but as it is not at
7039: the end of the string, the first alternative fails; the second alterna-
7040: tive is taken and the recursion kicks in. The recursive call to subpat-
7041: tern 1 successfully matches the next character ("b"). (Note that the
7042: beginning and end of line tests are not part of the recursion).
7043:
7044: Back at the top level, the next character ("c") is compared with what
7045: subpattern 2 matched, which was "a". This fails. Because the recursion
7046: is treated as an atomic group, there are now no backtracking points,
7047: and so the entire match fails. (Perl is able, at this point, to re-
7048: enter the recursion and try the second alternative.) However, if the
7049: pattern is written with the alternatives in the other order, things are
7050: different:
7051:
7052: ^((.)(?1)\2|.)$
7053:
7054: This time, the recursing alternative is tried first, and continues to
7055: recurse until it runs out of characters, at which point the recursion
7056: fails. But this time we do have another alternative to try at the
7057: higher level. That is the big difference: in the previous case the
7058: remaining alternative is at a deeper recursion level, which PCRE cannot
7059: use.
7060:
7061: To change the pattern so that it matches all palindromic strings, not
7062: just those with an odd number of characters, it is tempting to change
7063: the pattern to this:
7064:
7065: ^((.)(?1)\2|.?)$
7066:
7067: Again, this works in Perl, but not in PCRE, and for the same reason.
7068: When a deeper recursion has matched a single character, it cannot be
7069: entered again in order to match an empty string. The solution is to
7070: separate the two cases, and write out the odd and even cases as alter-
7071: natives at the higher level:
7072:
7073: ^(?:((.)(?1)\2|)|((.)(?3)\4|.))
7074:
7075: If you want to match typical palindromic phrases, the pattern has to
7076: ignore all non-word characters, which can be done like this:
7077:
7078: ^\W*+(?:((.)\W*+(?1)\W*+\2|)|((.)\W*+(?3)\W*+\4|\W*+.\W*+))\W*+$
7079:
7080: If run with the PCRE_CASELESS option, this pattern matches phrases such
7081: as "A man, a plan, a canal: Panama!" and it works well in both PCRE and
7082: Perl. Note the use of the possessive quantifier *+ to avoid backtrack-
7083: ing into sequences of non-word characters. Without this, PCRE takes a
7084: great deal longer (ten times or more) to match typical phrases, and
7085: Perl takes so long that you think it has gone into a loop.
7086:
7087: WARNING: The palindrome-matching patterns above work only if the sub-
7088: ject string does not start with a palindrome that is shorter than the
7089: entire string. For example, although "abcba" is correctly matched, if
7090: the subject is "ababa", PCRE finds the palindrome "aba" at the start,
7091: then fails at top level because the end of the string does not follow.
7092: Once again, it cannot jump back into the recursion to try other alter-
7093: natives, so the entire match fails.
7094:
7095: The second way in which PCRE and Perl differ in their recursion pro-
7096: cessing is in the handling of captured values. In Perl, when a subpat-
7097: tern is called recursively or as a subpattern (see the next section),
7098: it has no access to any values that were captured outside the recur-
7099: sion, whereas in PCRE these values can be referenced. Consider this
7100: pattern:
7101:
7102: ^(.)(\1|a(?2))
7103:
7104: In PCRE, this pattern matches "bab". The first capturing parentheses
7105: match "b", then in the second group, when the back reference \1 fails
7106: to match "b", the second alternative matches "a" and then recurses. In
7107: the recursion, \1 does now match "b" and so the whole match succeeds.
7108: In Perl, the pattern fails to match because inside the recursive call
7109: \1 cannot access the externally set value.
7110:
7111:
7112: SUBPATTERNS AS SUBROUTINES
7113:
7114: If the syntax for a recursive subpattern call (either by number or by
7115: name) is used outside the parentheses to which it refers, it operates
7116: like a subroutine in a programming language. The called subpattern may
7117: be defined before or after the reference. A numbered reference can be
7118: absolute or relative, as in these examples:
7119:
7120: (...(absolute)...)...(?2)...
7121: (...(relative)...)...(?-1)...
7122: (...(?+1)...(relative)...
7123:
7124: An earlier example pointed out that the pattern
7125:
7126: (sens|respons)e and \1ibility
7127:
7128: matches "sense and sensibility" and "response and responsibility", but
7129: not "sense and responsibility". If instead the pattern
7130:
7131: (sens|respons)e and (?1)ibility
7132:
7133: is used, it does match "sense and responsibility" as well as the other
7134: two strings. Another example is given in the discussion of DEFINE
7135: above.
7136:
7137: All subroutine calls, whether recursive or not, are always treated as
7138: atomic groups. That is, once a subroutine has matched some of the sub-
7139: ject string, it is never re-entered, even if it contains untried alter-
7140: natives and there is a subsequent matching failure. Any capturing
7141: parentheses that are set during the subroutine call revert to their
7142: previous values afterwards.
7143:
7144: Processing options such as case-independence are fixed when a subpat-
7145: tern is defined, so if it is used as a subroutine, such options cannot
7146: be changed for different calls. For example, consider this pattern:
7147:
7148: (abc)(?i:(?-1))
7149:
7150: It matches "abcabc". It does not match "abcABC" because the change of
7151: processing option does not affect the called subpattern.
7152:
7153:
7154: ONIGURUMA SUBROUTINE SYNTAX
7155:
7156: For compatibility with Oniguruma, the non-Perl syntax \g followed by a
7157: name or a number enclosed either in angle brackets or single quotes, is
7158: an alternative syntax for referencing a subpattern as a subroutine,
7159: possibly recursively. Here are two of the examples used above, rewrit-
7160: ten using this syntax:
7161:
7162: (?<pn> \( ( (?>[^()]+) | \g<pn> )* \) )
7163: (sens|respons)e and \g'1'ibility
7164:
7165: PCRE supports an extension to Oniguruma: if a number is preceded by a
7166: plus or a minus sign it is taken as a relative reference. For example:
7167:
7168: (abc)(?i:\g<-1>)
7169:
7170: Note that \g{...} (Perl syntax) and \g<...> (Oniguruma syntax) are not
7171: synonymous. The former is a back reference; the latter is a subroutine
7172: call.
7173:
7174:
7175: CALLOUTS
7176:
7177: Perl has a feature whereby using the sequence (?{...}) causes arbitrary
7178: Perl code to be obeyed in the middle of matching a regular expression.
7179: This makes it possible, amongst other things, to extract different sub-
7180: strings that match the same pair of parentheses when there is a repeti-
7181: tion.
7182:
7183: PCRE provides a similar feature, but of course it cannot obey arbitrary
7184: Perl code. The feature is called "callout". The caller of PCRE provides
7185: an external function by putting its entry point in the global variable
7186: pcre_callout (8-bit library) or pcre[16|32]_callout (16-bit or 32-bit
7187: library). By default, this variable contains NULL, which disables all
7188: calling out.
7189:
7190: Within a regular expression, (?C) indicates the points at which the
7191: external function is to be called. If you want to identify different
7192: callout points, you can put a number less than 256 after the letter C.
7193: The default value is zero. For example, this pattern has two callout
7194: points:
7195:
7196: (?C1)abc(?C2)def
7197:
7198: If the PCRE_AUTO_CALLOUT flag is passed to a compiling function, call-
7199: outs are automatically installed before each item in the pattern. They
7200: are all numbered 255. If there is a conditional group in the pattern
7201: whose condition is an assertion, an additional callout is inserted just
7202: before the condition. An explicit callout may also be set at this posi-
7203: tion, as in this example:
7204:
7205: (?(?C9)(?=a)abc|def)
7206:
7207: Note that this applies only to assertion conditions, not to other types
7208: of condition.
7209:
7210: During matching, when PCRE reaches a callout point, the external func-
7211: tion is called. It is provided with the number of the callout, the
7212: position in the pattern, and, optionally, one item of data originally
7213: supplied by the caller of the matching function. The callout function
7214: may cause matching to proceed, to backtrack, or to fail altogether.
7215:
7216: By default, PCRE implements a number of optimizations at compile time
7217: and matching time, and one side-effect is that sometimes callouts are
7218: skipped. If you need all possible callouts to happen, you need to set
7219: options that disable the relevant optimizations. More details, and a
7220: complete description of the interface to the callout function, are
7221: given in the pcrecallout documentation.
7222:
7223:
7224: BACKTRACKING CONTROL
7225:
7226: Perl 5.10 introduced a number of "Special Backtracking Control Verbs",
7227: which are still described in the Perl documentation as "experimental
7228: and subject to change or removal in a future version of Perl". It goes
7229: on to say: "Their usage in production code should be noted to avoid
7230: problems during upgrades." The same remarks apply to the PCRE features
7231: described in this section.
7232:
7233: The new verbs make use of what was previously invalid syntax: an open-
7234: ing parenthesis followed by an asterisk. They are generally of the form
7235: (*VERB) or (*VERB:NAME). Some may take either form, possibly behaving
7236: differently depending on whether or not a name is present. A name is
7237: any sequence of characters that does not include a closing parenthesis.
7238: The maximum length of name is 255 in the 8-bit library and 65535 in the
7239: 16-bit and 32-bit libraries. If the name is empty, that is, if the
7240: closing parenthesis immediately follows the colon, the effect is as if
7241: the colon were not there. Any number of these verbs may occur in a
7242: pattern.
7243:
7244: Since these verbs are specifically related to backtracking, most of
7245: them can be used only when the pattern is to be matched using one of
7246: the traditional matching functions, because these use a backtracking
7247: algorithm. With the exception of (*FAIL), which behaves like a failing
7248: negative assertion, the backtracking control verbs cause an error if
7249: encountered by a DFA matching function.
7250:
7251: The behaviour of these verbs in repeated groups, assertions, and in
7252: subpatterns called as subroutines (whether or not recursively) is docu-
7253: mented below.
7254:
7255: Optimizations that affect backtracking verbs
7256:
7257: PCRE contains some optimizations that are used to speed up matching by
7258: running some checks at the start of each match attempt. For example, it
7259: may know the minimum length of matching subject, or that a particular
7260: character must be present. When one of these optimizations bypasses the
7261: running of a match, any included backtracking verbs will not, of
7262: course, be processed. You can suppress the start-of-match optimizations
7263: by setting the PCRE_NO_START_OPTIMIZE option when calling pcre_com-
7264: pile() or pcre_exec(), or by starting the pattern with (*NO_START_OPT).
7265: There is more discussion of this option in the section entitled "Option
7266: bits for pcre_exec()" in the pcreapi documentation.
7267:
7268: Experiments with Perl suggest that it too has similar optimizations,
7269: sometimes leading to anomalous results.
7270:
7271: Verbs that act immediately
7272:
7273: The following verbs act as soon as they are encountered. They may not
7274: be followed by a name.
7275:
7276: (*ACCEPT)
7277:
7278: This verb causes the match to end successfully, skipping the remainder
7279: of the pattern. However, when it is inside a subpattern that is called
7280: as a subroutine, only that subpattern is ended successfully. Matching
7281: then continues at the outer level. If (*ACCEPT) in triggered in a posi-
7282: tive assertion, the assertion succeeds; in a negative assertion, the
7283: assertion fails.
7284:
7285: If (*ACCEPT) is inside capturing parentheses, the data so far is cap-
7286: tured. For example:
7287:
7288: A((?:A|B(*ACCEPT)|C)D)
7289:
7290: This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is cap-
7291: tured by the outer parentheses.
7292:
7293: (*FAIL) or (*F)
7294:
7295: This verb causes a matching failure, forcing backtracking to occur. It
7296: is equivalent to (?!) but easier to read. The Perl documentation notes
7297: that it is probably useful only when combined with (?{}) or (??{}).
7298: Those are, of course, Perl features that are not present in PCRE. The
7299: nearest equivalent is the callout feature, as for example in this pat-
7300: tern:
7301:
7302: a+(?C)(*FAIL)
7303:
7304: A match with the string "aaaa" always fails, but the callout is taken
7305: before each backtrack happens (in this example, 10 times).
7306:
7307: Recording which path was taken
7308:
7309: There is one verb whose main purpose is to track how a match was
7310: arrived at, though it also has a secondary use in conjunction with
7311: advancing the match starting point (see (*SKIP) below).
7312:
7313: (*MARK:NAME) or (*:NAME)
7314:
7315: A name is always required with this verb. There may be as many
7316: instances of (*MARK) as you like in a pattern, and their names do not
7317: have to be unique.
7318:
7319: When a match succeeds, the name of the last-encountered (*MARK:NAME),
7320: (*PRUNE:NAME), or (*THEN:NAME) on the matching path is passed back to
7321: the caller as described in the section entitled "Extra data for
7322: pcre_exec()" in the pcreapi documentation. Here is an example of
7323: pcretest output, where the /K modifier requests the retrieval and out-
7324: putting of (*MARK) data:
7325:
7326: re> /X(*MARK:A)Y|X(*MARK:B)Z/K
7327: data> XY
7328: 0: XY
7329: MK: A
7330: XZ
7331: 0: XZ
7332: MK: B
7333:
7334: The (*MARK) name is tagged with "MK:" in this output, and in this exam-
7335: ple it indicates which of the two alternatives matched. This is a more
7336: efficient way of obtaining this information than putting each alterna-
7337: tive in its own capturing parentheses.
7338:
7339: If a verb with a name is encountered in a positive assertion that is
7340: true, the name is recorded and passed back if it is the last-encoun-
7341: tered. This does not happen for negative assertions or failing positive
7342: assertions.
7343:
7344: After a partial match or a failed match, the last encountered name in
7345: the entire match process is returned. For example:
7346:
7347: re> /X(*MARK:A)Y|X(*MARK:B)Z/K
7348: data> XP
7349: No match, mark = B
7350:
7351: Note that in this unanchored example the mark is retained from the
7352: match attempt that started at the letter "X" in the subject. Subsequent
7353: match attempts starting at "P" and then with an empty string do not get
7354: as far as the (*MARK) item, but nevertheless do not reset it.
7355:
7356: If you are interested in (*MARK) values after failed matches, you
7357: should probably set the PCRE_NO_START_OPTIMIZE option (see above) to
7358: ensure that the match is always attempted.
7359:
7360: Verbs that act after backtracking
7361:
7362: The following verbs do nothing when they are encountered. Matching con-
7363: tinues with what follows, but if there is no subsequent match, causing
7364: a backtrack to the verb, a failure is forced. That is, backtracking
7365: cannot pass to the left of the verb. However, when one of these verbs
7366: appears inside an atomic group or an assertion that is true, its effect
7367: is confined to that group, because once the group has been matched,
7368: there is never any backtracking into it. In this situation, backtrack-
7369: ing can "jump back" to the left of the entire atomic group or asser-
7370: tion. (Remember also, as stated above, that this localization also
7371: applies in subroutine calls.)
7372:
7373: These verbs differ in exactly what kind of failure occurs when back-
7374: tracking reaches them. The behaviour described below is what happens
7375: when the verb is not in a subroutine or an assertion. Subsequent sec-
7376: tions cover these special cases.
7377:
7378: (*COMMIT)
7379:
7380: This verb, which may not be followed by a name, causes the whole match
7381: to fail outright if there is a later matching failure that causes back-
7382: tracking to reach it. Even if the pattern is unanchored, no further
7383: attempts to find a match by advancing the starting point take place. If
7384: (*COMMIT) is the only backtracking verb that is encountered, once it
7385: has been passed pcre_exec() is committed to finding a match at the cur-
7386: rent starting point, or not at all. For example:
7387:
7388: a+(*COMMIT)b
7389:
7390: This matches "xxaab" but not "aacaab". It can be thought of as a kind
7391: of dynamic anchor, or "I've started, so I must finish." The name of the
7392: most recently passed (*MARK) in the path is passed back when (*COMMIT)
7393: forces a match failure.
7394:
7395: If there is more than one backtracking verb in a pattern, a different
7396: one that follows (*COMMIT) may be triggered first, so merely passing
7397: (*COMMIT) during a match does not always guarantee that a match must be
7398: at this starting point.
7399:
7400: Note that (*COMMIT) at the start of a pattern is not the same as an
7401: anchor, unless PCRE's start-of-match optimizations are turned off, as
7402: shown in this pcretest example:
7403:
7404: re> /(*COMMIT)abc/
7405: data> xyzabc
7406: 0: abc
7407: xyzabc\Y
7408: No match
7409:
7410: PCRE knows that any match must start with "a", so the optimization
7411: skips along the subject to "a" before running the first match attempt,
7412: which succeeds. When the optimization is disabled by the \Y escape in
7413: the second subject, the match starts at "x" and so the (*COMMIT) causes
7414: it to fail without trying any other starting points.
7415:
7416: (*PRUNE) or (*PRUNE:NAME)
7417:
7418: This verb causes the match to fail at the current starting position in
7419: the subject if there is a later matching failure that causes backtrack-
7420: ing to reach it. If the pattern is unanchored, the normal "bumpalong"
7421: advance to the next starting character then happens. Backtracking can
7422: occur as usual to the left of (*PRUNE), before it is reached, or when
7423: matching to the right of (*PRUNE), but if there is no match to the
7424: right, backtracking cannot cross (*PRUNE). In simple cases, the use of
7425: (*PRUNE) is just an alternative to an atomic group or possessive quan-
7426: tifier, but there are some uses of (*PRUNE) that cannot be expressed in
7427: any other way. In an anchored pattern (*PRUNE) has the same effect as
7428: (*COMMIT).
7429:
7430: The behaviour of (*PRUNE:NAME) is the not the same as
7431: (*MARK:NAME)(*PRUNE). It is like (*MARK:NAME) in that the name is
7432: remembered for passing back to the caller. However, (*SKIP:NAME)
7433: searches only for names set with (*MARK).
7434:
7435: (*SKIP)
7436:
7437: This verb, when given without a name, is like (*PRUNE), except that if
7438: the pattern is unanchored, the "bumpalong" advance is not to the next
7439: character, but to the position in the subject where (*SKIP) was encoun-
7440: tered. (*SKIP) signifies that whatever text was matched leading up to
7441: it cannot be part of a successful match. Consider:
7442:
7443: a+(*SKIP)b
7444:
7445: If the subject is "aaaac...", after the first match attempt fails
7446: (starting at the first character in the string), the starting point
7447: skips on to start the next attempt at "c". Note that a possessive quan-
7448: tifer does not have the same effect as this example; although it would
7449: suppress backtracking during the first match attempt, the second
7450: attempt would start at the second character instead of skipping on to
7451: "c".
7452:
7453: (*SKIP:NAME)
7454:
7455: When (*SKIP) has an associated name, its behaviour is modified. When it
7456: is triggered, the previous path through the pattern is searched for the
7457: most recent (*MARK) that has the same name. If one is found, the
7458: "bumpalong" advance is to the subject position that corresponds to that
7459: (*MARK) instead of to where (*SKIP) was encountered. If no (*MARK) with
7460: a matching name is found, the (*SKIP) is ignored.
7461:
7462: Note that (*SKIP:NAME) searches only for names set by (*MARK:NAME). It
7463: ignores names that are set by (*PRUNE:NAME) or (*THEN:NAME).
7464:
7465: (*THEN) or (*THEN:NAME)
7466:
7467: This verb causes a skip to the next innermost alternative when back-
7468: tracking reaches it. That is, it cancels any further backtracking
7469: within the current alternative. Its name comes from the observation
7470: that it can be used for a pattern-based if-then-else block:
7471:
7472: ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
7473:
7474: If the COND1 pattern matches, FOO is tried (and possibly further items
7475: after the end of the group if FOO succeeds); on failure, the matcher
7476: skips to the second alternative and tries COND2, without backtracking
7477: into COND1. If that succeeds and BAR fails, COND3 is tried. If subse-
7478: quently BAZ fails, there are no more alternatives, so there is a back-
7479: track to whatever came before the entire group. If (*THEN) is not
7480: inside an alternation, it acts like (*PRUNE).
7481:
7482: The behaviour of (*THEN:NAME) is the not the same as
7483: (*MARK:NAME)(*THEN). It is like (*MARK:NAME) in that the name is
7484: remembered for passing back to the caller. However, (*SKIP:NAME)
7485: searches only for names set with (*MARK).
7486:
7487: A subpattern that does not contain a | character is just a part of the
7488: enclosing alternative; it is not a nested alternation with only one
7489: alternative. The effect of (*THEN) extends beyond such a subpattern to
7490: the enclosing alternative. Consider this pattern, where A, B, etc. are
7491: complex pattern fragments that do not contain any | characters at this
7492: level:
7493:
7494: A (B(*THEN)C) | D
7495:
7496: If A and B are matched, but there is a failure in C, matching does not
7497: backtrack into A; instead it moves to the next alternative, that is, D.
7498: However, if the subpattern containing (*THEN) is given an alternative,
7499: it behaves differently:
7500:
7501: A (B(*THEN)C | (*FAIL)) | D
7502:
7503: The effect of (*THEN) is now confined to the inner subpattern. After a
7504: failure in C, matching moves to (*FAIL), which causes the whole subpat-
7505: tern to fail because there are no more alternatives to try. In this
7506: case, matching does now backtrack into A.
7507:
7508: Note that a conditional subpattern is not considered as having two
7509: alternatives, because only one is ever used. In other words, the |
7510: character in a conditional subpattern has a different meaning. Ignoring
7511: white space, consider:
7512:
7513: ^.*? (?(?=a) a | b(*THEN)c )
7514:
7515: If the subject is "ba", this pattern does not match. Because .*? is
7516: ungreedy, it initially matches zero characters. The condition (?=a)
7517: then fails, the character "b" is matched, but "c" is not. At this
7518: point, matching does not backtrack to .*? as might perhaps be expected
7519: from the presence of the | character. The conditional subpattern is
7520: part of the single alternative that comprises the whole pattern, and so
7521: the match fails. (If there was a backtrack into .*?, allowing it to
7522: match "b", the match would succeed.)
7523:
7524: The verbs just described provide four different "strengths" of control
7525: when subsequent matching fails. (*THEN) is the weakest, carrying on the
7526: match at the next alternative. (*PRUNE) comes next, failing the match
7527: at the current starting position, but allowing an advance to the next
7528: character (for an unanchored pattern). (*SKIP) is similar, except that
7529: the advance may be more than one character. (*COMMIT) is the strongest,
7530: causing the entire match to fail.
7531:
7532: More than one backtracking verb
7533:
7534: If more than one backtracking verb is present in a pattern, the one
7535: that is backtracked onto first acts. For example, consider this pat-
7536: tern, where A, B, etc. are complex pattern fragments:
7537:
7538: (A(*COMMIT)B(*THEN)C|ABD)
7539:
7540: If A matches but B fails, the backtrack to (*COMMIT) causes the entire
7541: match to fail. However, if A and B match, but C fails, the backtrack to
7542: (*THEN) causes the next alternative (ABD) to be tried. This behaviour
7543: is consistent, but is not always the same as Perl's. It means that if
7544: two or more backtracking verbs appear in succession, all the the last
7545: of them has no effect. Consider this example:
7546:
7547: ...(*COMMIT)(*PRUNE)...
7548:
7549: If there is a matching failure to the right, backtracking onto (*PRUNE)
7550: causes it to be triggered, and its action is taken. There can never be
7551: a backtrack onto (*COMMIT).
7552:
7553: Backtracking verbs in repeated groups
7554:
7555: PCRE differs from Perl in its handling of backtracking verbs in
7556: repeated groups. For example, consider:
7557:
7558: /(a(*COMMIT)b)+ac/
7559:
7560: If the subject is "abac", Perl matches, but PCRE fails because the
7561: (*COMMIT) in the second repeat of the group acts.
7562:
7563: Backtracking verbs in assertions
7564:
7565: (*FAIL) in an assertion has its normal effect: it forces an immediate
7566: backtrack.
7567:
7568: (*ACCEPT) in a positive assertion causes the assertion to succeed with-
7569: out any further processing. In a negative assertion, (*ACCEPT) causes
7570: the assertion to fail without any further processing.
7571:
7572: The other backtracking verbs are not treated specially if they appear
7573: in a positive assertion. In particular, (*THEN) skips to the next
7574: alternative in the innermost enclosing group that has alternations,
7575: whether or not this is within the assertion.
7576:
7577: Negative assertions are, however, different, in order to ensure that
7578: changing a positive assertion into a negative assertion changes its
7579: result. Backtracking into (*COMMIT), (*SKIP), or (*PRUNE) causes a neg-
7580: ative assertion to be true, without considering any further alternative
7581: branches in the assertion. Backtracking into (*THEN) causes it to skip
7582: to the next enclosing alternative within the assertion (the normal be-
7583: haviour), but if the assertion does not have such an alternative,
7584: (*THEN) behaves like (*PRUNE).
7585:
7586: Backtracking verbs in subroutines
7587:
7588: These behaviours occur whether or not the subpattern is called recur-
7589: sively. Perl's treatment of subroutines is different in some cases.
7590:
7591: (*FAIL) in a subpattern called as a subroutine has its normal effect:
7592: it forces an immediate backtrack.
7593:
7594: (*ACCEPT) in a subpattern called as a subroutine causes the subroutine
7595: match to succeed without any further processing. Matching then contin-
7596: ues after the subroutine call.
7597:
7598: (*COMMIT), (*SKIP), and (*PRUNE) in a subpattern called as a subroutine
7599: cause the subroutine match to fail.
7600:
7601: (*THEN) skips to the next alternative in the innermost enclosing group
7602: within the subpattern that has alternatives. If there is no such group
7603: within the subpattern, (*THEN) causes the subroutine match to fail.
7604:
7605:
7606: SEE ALSO
7607:
7608: pcreapi(3), pcrecallout(3), pcrematching(3), pcresyntax(3), pcre(3),
7609: pcre16(3), pcre32(3).
7610:
7611:
7612: AUTHOR
7613:
7614: Philip Hazel
7615: University Computing Service
7616: Cambridge CB2 3QH, England.
7617:
7618:
7619: REVISION
7620:
7621: Last updated: 03 December 2013
7622: Copyright (c) 1997-2013 University of Cambridge.
7623: ------------------------------------------------------------------------------
7624:
7625:
7626: PCRESYNTAX(3) Library Functions Manual PCRESYNTAX(3)
7627:
7628:
7629:
7630: NAME
7631: PCRE - Perl-compatible regular expressions
7632:
7633: PCRE REGULAR EXPRESSION SYNTAX SUMMARY
7634:
7635: The full syntax and semantics of the regular expressions that are sup-
7636: ported by PCRE are described in the pcrepattern documentation. This
7637: document contains a quick-reference summary of the syntax.
7638:
7639:
7640: QUOTING
7641:
7642: \x where x is non-alphanumeric is a literal x
7643: \Q...\E treat enclosed characters as literal
7644:
7645:
7646: CHARACTERS
7647:
7648: \a alarm, that is, the BEL character (hex 07)
7649: \cx "control-x", where x is any ASCII character
7650: \e escape (hex 1B)
7651: \f form feed (hex 0C)
7652: \n newline (hex 0A)
7653: \r carriage return (hex 0D)
7654: \t tab (hex 09)
7655: \0dd character with octal code 0dd
7656: \ddd character with octal code ddd, or backreference
7657: \o{ddd..} character with octal code ddd..
7658: \xhh character with hex code hh
7659: \x{hhh..} character with hex code hhh..
7660:
7661: Note that \0dd is always an octal code, and that \8 and \9 are the lit-
7662: eral characters "8" and "9".
7663:
7664:
7665: CHARACTER TYPES
7666:
7667: . any character except newline;
7668: in dotall mode, any character whatsoever
7669: \C one data unit, even in UTF mode (best avoided)
7670: \d a decimal digit
7671: \D a character that is not a decimal digit
7672: \h a horizontal white space character
7673: \H a character that is not a horizontal white space character
7674: \N a character that is not a newline
7675: \p{xx} a character with the xx property
7676: \P{xx} a character without the xx property
7677: \R a newline sequence
7678: \s a white space character
7679: \S a character that is not a white space character
7680: \v a vertical white space character
7681: \V a character that is not a vertical white space character
7682: \w a "word" character
7683: \W a "non-word" character
7684: \X a Unicode extended grapheme cluster
7685:
7686: By default, \d, \s, and \w match only ASCII characters, even in UTF-8
7687: mode or in the 16- bit and 32-bit libraries. However, if locale-spe-
7688: cific matching is happening, \s and \w may also match characters with
7689: code points in the range 128-255. If the PCRE_UCP option is set, the
7690: behaviour of these escape sequences is changed to use Unicode proper-
7691: ties and they match many more characters.
7692:
7693:
7694: GENERAL CATEGORY PROPERTIES FOR \p and \P
7695:
7696: C Other
7697: Cc Control
7698: Cf Format
7699: Cn Unassigned
7700: Co Private use
7701: Cs Surrogate
7702:
7703: L Letter
7704: Ll Lower case letter
7705: Lm Modifier letter
7706: Lo Other letter
7707: Lt Title case letter
7708: Lu Upper case letter
7709: L& Ll, Lu, or Lt
7710:
7711: M Mark
7712: Mc Spacing mark
7713: Me Enclosing mark
7714: Mn Non-spacing mark
7715:
7716: N Number
7717: Nd Decimal number
7718: Nl Letter number
7719: No Other number
7720:
7721: P Punctuation
7722: Pc Connector punctuation
7723: Pd Dash punctuation
7724: Pe Close punctuation
7725: Pf Final punctuation
7726: Pi Initial punctuation
7727: Po Other punctuation
7728: Ps Open punctuation
7729:
7730: S Symbol
7731: Sc Currency symbol
7732: Sk Modifier symbol
7733: Sm Mathematical symbol
7734: So Other symbol
7735:
7736: Z Separator
7737: Zl Line separator
7738: Zp Paragraph separator
7739: Zs Space separator
7740:
7741:
7742: PCRE SPECIAL CATEGORY PROPERTIES FOR \p and \P
7743:
7744: Xan Alphanumeric: union of properties L and N
7745: Xps POSIX space: property Z or tab, NL, VT, FF, CR
7746: Xsp Perl space: property Z or tab, NL, VT, FF, CR
7747: Xuc Univerally-named character: one that can be
7748: represented by a Universal Character Name
7749: Xwd Perl word: property Xan or underscore
7750:
7751: Perl and POSIX space are now the same. Perl added VT to its space char-
7752: acter set at release 5.18 and PCRE changed at release 8.34.
7753:
7754:
7755: SCRIPT NAMES FOR \p AND \P
7756:
7757: Arabic, Armenian, Avestan, Balinese, Bamum, Batak, Bengali, Bopomofo,
7758: Brahmi, Braille, Buginese, Buhid, Canadian_Aboriginal, Carian, Chakma,
7759: Cham, Cherokee, Common, Coptic, Cuneiform, Cypriot, Cyrillic, Deseret,
7760: Devanagari, Egyptian_Hieroglyphs, Ethiopic, Georgian, Glagolitic,
7761: Gothic, Greek, Gujarati, Gurmukhi, Han, Hangul, Hanunoo, Hebrew, Hira-
7762: gana, Imperial_Aramaic, Inherited, Inscriptional_Pahlavi, Inscrip-
7763: tional_Parthian, Javanese, Kaithi, Kannada, Katakana, Kayah_Li,
7764: Kharoshthi, Khmer, Lao, Latin, Lepcha, Limbu, Linear_B, Lisu, Lycian,
7765: Lydian, Malayalam, Mandaic, Meetei_Mayek, Meroitic_Cursive,
7766: Meroitic_Hieroglyphs, Miao, Mongolian, Myanmar, New_Tai_Lue, Nko,
7767: Ogham, Old_Italic, Old_Persian, Old_South_Arabian, Old_Turkic,
7768: Ol_Chiki, Oriya, Osmanya, Phags_Pa, Phoenician, Rejang, Runic, Samari-
7769: tan, Saurashtra, Sharada, Shavian, Sinhala, Sora_Sompeng, Sundanese,
7770: Syloti_Nagri, Syriac, Tagalog, Tagbanwa, Tai_Le, Tai_Tham, Tai_Viet,
7771: Takri, Tamil, Telugu, Thaana, Thai, Tibetan, Tifinagh, Ugaritic, Vai,
7772: Yi.
7773:
7774:
7775: CHARACTER CLASSES
7776:
7777: [...] positive character class
7778: [^...] negative character class
7779: [x-y] range (can be used for hex characters)
7780: [[:xxx:]] positive POSIX named set
7781: [[:^xxx:]] negative POSIX named set
7782:
7783: alnum alphanumeric
7784: alpha alphabetic
7785: ascii 0-127
7786: blank space or tab
7787: cntrl control character
7788: digit decimal digit
7789: graph printing, excluding space
7790: lower lower case letter
7791: print printing, including space
7792: punct printing, excluding alphanumeric
7793: space white space
7794: upper upper case letter
7795: word same as \w
7796: xdigit hexadecimal digit
7797:
7798: In PCRE, POSIX character set names recognize only ASCII characters by
7799: default, but some of them use Unicode properties if PCRE_UCP is set.
7800: You can use \Q...\E inside a character class.
7801:
7802:
7803: QUANTIFIERS
7804:
7805: ? 0 or 1, greedy
7806: ?+ 0 or 1, possessive
7807: ?? 0 or 1, lazy
7808: * 0 or more, greedy
7809: *+ 0 or more, possessive
7810: *? 0 or more, lazy
7811: + 1 or more, greedy
7812: ++ 1 or more, possessive
7813: +? 1 or more, lazy
7814: {n} exactly n
7815: {n,m} at least n, no more than m, greedy
7816: {n,m}+ at least n, no more than m, possessive
7817: {n,m}? at least n, no more than m, lazy
7818: {n,} n or more, greedy
7819: {n,}+ n or more, possessive
7820: {n,}? n or more, lazy
7821:
7822:
7823: ANCHORS AND SIMPLE ASSERTIONS
7824:
7825: \b word boundary
7826: \B not a word boundary
7827: ^ start of subject
7828: also after internal newline in multiline mode
7829: \A start of subject
7830: $ end of subject
7831: also before newline at end of subject
7832: also before internal newline in multiline mode
7833: \Z end of subject
7834: also before newline at end of subject
7835: \z end of subject
7836: \G first matching position in subject
7837:
7838:
7839: MATCH POINT RESET
7840:
7841: \K reset start of match
7842:
7843:
7844: ALTERNATION
7845:
7846: expr|expr|expr...
7847:
7848:
7849: CAPTURING
7850:
7851: (...) capturing group
7852: (?<name>...) named capturing group (Perl)
7853: (?'name'...) named capturing group (Perl)
7854: (?P<name>...) named capturing group (Python)
7855: (?:...) non-capturing group
7856: (?|...) non-capturing group; reset group numbers for
7857: capturing groups in each alternative
7858:
7859:
7860: ATOMIC GROUPS
7861:
7862: (?>...) atomic, non-capturing group
7863:
7864:
7865: COMMENT
7866:
7867: (?#....) comment (not nestable)
7868:
7869:
7870: OPTION SETTING
7871:
7872: (?i) caseless
7873: (?J) allow duplicate names
7874: (?m) multiline
7875: (?s) single line (dotall)
7876: (?U) default ungreedy (lazy)
7877: (?x) extended (ignore white space)
7878: (?-...) unset option(s)
7879:
7880: The following are recognized only at the start of a pattern or after
7881: one of the newline-setting options with similar syntax:
7882:
7883: (*LIMIT_MATCH=d) set the match limit to d (decimal number)
7884: (*LIMIT_RECURSION=d) set the recursion limit to d (decimal number)
7885: (*NO_START_OPT) no start-match optimization (PCRE_NO_START_OPTIMIZE)
7886: (*UTF8) set UTF-8 mode: 8-bit library (PCRE_UTF8)
7887: (*UTF16) set UTF-16 mode: 16-bit library (PCRE_UTF16)
7888: (*UTF32) set UTF-32 mode: 32-bit library (PCRE_UTF32)
7889: (*UTF) set appropriate UTF mode for the library in use
7890: (*UCP) set PCRE_UCP (use Unicode properties for \d etc)
7891:
7892: Note that LIMIT_MATCH and LIMIT_RECURSION can only reduce the value of
7893: the limits set by the caller of pcre_exec(), not increase them.
7894:
7895:
7896: LOOKAHEAD AND LOOKBEHIND ASSERTIONS
7897:
7898: (?=...) positive look ahead
7899: (?!...) negative look ahead
7900: (?<=...) positive look behind
7901: (?<!...) negative look behind
7902:
7903: Each top-level branch of a look behind must be of a fixed length.
7904:
7905:
7906: BACKREFERENCES
7907:
7908: \n reference by number (can be ambiguous)
7909: \gn reference by number
7910: \g{n} reference by number
7911: \g{-n} relative reference by number
7912: \k<name> reference by name (Perl)
7913: \k'name' reference by name (Perl)
7914: \g{name} reference by name (Perl)
7915: \k{name} reference by name (.NET)
7916: (?P=name) reference by name (Python)
7917:
7918:
7919: SUBROUTINE REFERENCES (POSSIBLY RECURSIVE)
7920:
7921: (?R) recurse whole pattern
7922: (?n) call subpattern by absolute number
7923: (?+n) call subpattern by relative number
7924: (?-n) call subpattern by relative number
7925: (?&name) call subpattern by name (Perl)
7926: (?P>name) call subpattern by name (Python)
7927: \g<name> call subpattern by name (Oniguruma)
7928: \g'name' call subpattern by name (Oniguruma)
7929: \g<n> call subpattern by absolute number (Oniguruma)
7930: \g'n' call subpattern by absolute number (Oniguruma)
7931: \g<+n> call subpattern by relative number (PCRE extension)
7932: \g'+n' call subpattern by relative number (PCRE extension)
7933: \g<-n> call subpattern by relative number (PCRE extension)
7934: \g'-n' call subpattern by relative number (PCRE extension)
7935:
7936:
7937: CONDITIONAL PATTERNS
7938:
7939: (?(condition)yes-pattern)
7940: (?(condition)yes-pattern|no-pattern)
7941:
7942: (?(n)... absolute reference condition
7943: (?(+n)... relative reference condition
7944: (?(-n)... relative reference condition
7945: (?(<name>)... named reference condition (Perl)
7946: (?('name')... named reference condition (Perl)
7947: (?(name)... named reference condition (PCRE)
7948: (?(R)... overall recursion condition
7949: (?(Rn)... specific group recursion condition
7950: (?(R&name)... specific recursion condition
7951: (?(DEFINE)... define subpattern for reference
7952: (?(assert)... assertion condition
7953:
7954:
7955: BACKTRACKING CONTROL
7956:
7957: The following act immediately they are reached:
7958:
7959: (*ACCEPT) force successful match
7960: (*FAIL) force backtrack; synonym (*F)
7961: (*MARK:NAME) set name to be passed back; synonym (*:NAME)
7962:
7963: The following act only when a subsequent match failure causes a back-
7964: track to reach them. They all force a match failure, but they differ in
7965: what happens afterwards. Those that advance the start-of-match point do
7966: so only if the pattern is not anchored.
7967:
7968: (*COMMIT) overall failure, no advance of starting point
7969: (*PRUNE) advance to next starting character
7970: (*PRUNE:NAME) equivalent to (*MARK:NAME)(*PRUNE)
7971: (*SKIP) advance to current matching position
7972: (*SKIP:NAME) advance to position corresponding to an earlier
7973: (*MARK:NAME); if not found, the (*SKIP) is ignored
7974: (*THEN) local failure, backtrack to next alternation
7975: (*THEN:NAME) equivalent to (*MARK:NAME)(*THEN)
7976:
7977:
7978: NEWLINE CONVENTIONS
7979:
7980: These are recognized only at the very start of the pattern or after a
7981: (*BSR_...), (*UTF8), (*UTF16), (*UTF32) or (*UCP) option.
7982:
7983: (*CR) carriage return only
7984: (*LF) linefeed only
7985: (*CRLF) carriage return followed by linefeed
7986: (*ANYCRLF) all three of the above
7987: (*ANY) any Unicode newline sequence
7988:
7989:
7990: WHAT \R MATCHES
7991:
7992: These are recognized only at the very start of the pattern or after a
7993: (*...) option that sets the newline convention or a UTF or UCP mode.
7994:
7995: (*BSR_ANYCRLF) CR, LF, or CRLF
7996: (*BSR_UNICODE) any Unicode newline sequence
7997:
7998:
7999: CALLOUTS
8000:
8001: (?C) callout
8002: (?Cn) callout with data n
8003:
8004:
8005: SEE ALSO
8006:
8007: pcrepattern(3), pcreapi(3), pcrecallout(3), pcrematching(3), pcre(3).
8008:
8009:
8010: AUTHOR
8011:
8012: Philip Hazel
8013: University Computing Service
8014: Cambridge CB2 3QH, England.
8015:
8016:
8017: REVISION
8018:
8019: Last updated: 12 November 2013
8020: Copyright (c) 1997-2013 University of Cambridge.
8021: ------------------------------------------------------------------------------
8022:
8023:
8024: PCREUNICODE(3) Library Functions Manual PCREUNICODE(3)
8025:
8026:
8027:
8028: NAME
8029: PCRE - Perl-compatible regular expressions
8030:
8031: UTF-8, UTF-16, UTF-32, AND UNICODE PROPERTY SUPPORT
8032:
8033: As well as UTF-8 support, PCRE also supports UTF-16 (from release 8.30)
8034: and UTF-32 (from release 8.32), by means of two additional libraries.
8035: They can be built as well as, or instead of, the 8-bit library.
8036:
8037:
8038: UTF-8 SUPPORT
8039:
8040: In order process UTF-8 strings, you must build PCRE's 8-bit library
8041: with UTF support, and, in addition, you must call pcre_compile() with
8042: the PCRE_UTF8 option flag, or the pattern must start with the sequence
8043: (*UTF8) or (*UTF). When either of these is the case, both the pattern
8044: and any subject strings that are matched against it are treated as
8045: UTF-8 strings instead of strings of individual 1-byte characters.
8046:
8047:
8048: UTF-16 AND UTF-32 SUPPORT
8049:
8050: In order process UTF-16 or UTF-32 strings, you must build PCRE's 16-bit
8051: or 32-bit library with UTF support, and, in addition, you must call
8052: pcre16_compile() or pcre32_compile() with the PCRE_UTF16 or PCRE_UTF32
8053: option flag, as appropriate. Alternatively, the pattern must start with
8054: the sequence (*UTF16), (*UTF32), as appropriate, or (*UTF), which can
8055: be used with either library. When UTF mode is set, both the pattern and
8056: any subject strings that are matched against it are treated as UTF-16
8057: or UTF-32 strings instead of strings of individual 16-bit or 32-bit
8058: characters.
8059:
8060:
8061: UTF SUPPORT OVERHEAD
8062:
8063: If you compile PCRE with UTF support, but do not use it at run time,
8064: the library will be a bit bigger, but the additional run time overhead
8065: is limited to testing the PCRE_UTF[8|16|32] flag occasionally, so
8066: should not be very big.
8067:
8068:
8069: UNICODE PROPERTY SUPPORT
8070:
8071: If PCRE is built with Unicode character property support (which implies
8072: UTF support), the escape sequences \p{..}, \P{..}, and \X can be used.
8073: The available properties that can be tested are limited to the general
8074: category properties such as Lu for an upper case letter or Nd for a
8075: decimal number, the Unicode script names such as Arabic or Han, and the
8076: derived properties Any and L&. Full lists is given in the pcrepattern
8077: and pcresyntax documentation. Only the short names for properties are
8078: supported. For example, \p{L} matches a letter. Its Perl synonym,
8079: \p{Letter}, is not supported. Furthermore, in Perl, many properties
8080: may optionally be prefixed by "Is", for compatibility with Perl 5.6.
8081: PCRE does not support this.
8082:
8083: Validity of UTF-8 strings
8084:
8085: When you set the PCRE_UTF8 flag, the byte strings passed as patterns
8086: and subjects are (by default) checked for validity on entry to the rel-
8087: evant functions. The entire string is checked before any other process-
8088: ing takes place. From release 7.3 of PCRE, the check is according the
8089: rules of RFC 3629, which are themselves derived from the Unicode speci-
8090: fication. Earlier releases of PCRE followed the rules of RFC 2279,
8091: which allows the full range of 31-bit values (0 to 0x7FFFFFFF). The
8092: current check allows only values in the range U+0 to U+10FFFF, exclud-
8093: ing the surrogate area. (From release 8.33 the so-called "non-charac-
8094: ter" code points are no longer excluded because Unicode corrigendum #9
8095: makes it clear that they should not be.)
8096:
8097: Characters in the "Surrogate Area" of Unicode are reserved for use by
8098: UTF-16, where they are used in pairs to encode codepoints with values
8099: greater than 0xFFFF. The code points that are encoded by UTF-16 pairs
8100: are available independently in the UTF-8 and UTF-32 encodings. (In
8101: other words, the whole surrogate thing is a fudge for UTF-16 which
8102: unfortunately messes up UTF-8 and UTF-32.)
8103:
8104: If an invalid UTF-8 string is passed to PCRE, an error return is given.
8105: At compile time, the only additional information is the offset to the
8106: first byte of the failing character. The run-time functions pcre_exec()
8107: and pcre_dfa_exec() also pass back this information, as well as a more
8108: detailed reason code if the caller has provided memory in which to do
8109: this.
8110:
8111: In some situations, you may already know that your strings are valid,
8112: and therefore want to skip these checks in order to improve perfor-
8113: mance, for example in the case of a long subject string that is being
8114: scanned repeatedly. If you set the PCRE_NO_UTF8_CHECK flag at compile
8115: time or at run time, PCRE assumes that the pattern or subject it is
8116: given (respectively) contains only valid UTF-8 codes. In this case, it
8117: does not diagnose an invalid UTF-8 string.
8118:
8119: Note that passing PCRE_NO_UTF8_CHECK to pcre_compile() just disables
8120: the check for the pattern; it does not also apply to subject strings.
8121: If you want to disable the check for a subject string you must pass
8122: this option to pcre_exec() or pcre_dfa_exec().
8123:
8124: If you pass an invalid UTF-8 string when PCRE_NO_UTF8_CHECK is set, the
8125: result is undefined and your program may crash.
8126:
8127: Validity of UTF-16 strings
8128:
8129: When you set the PCRE_UTF16 flag, the strings of 16-bit data units that
8130: are passed as patterns and subjects are (by default) checked for valid-
8131: ity on entry to the relevant functions. Values other than those in the
8132: surrogate range U+D800 to U+DFFF are independent code points. Values in
8133: the surrogate range must be used in pairs in the correct manner.
8134:
8135: If an invalid UTF-16 string is passed to PCRE, an error return is
8136: given. At compile time, the only additional information is the offset
8137: to the first data unit of the failing character. The run-time functions
8138: pcre16_exec() and pcre16_dfa_exec() also pass back this information, as
8139: well as a more detailed reason code if the caller has provided memory
8140: in which to do this.
8141:
8142: In some situations, you may already know that your strings are valid,
8143: and therefore want to skip these checks in order to improve perfor-
8144: mance. If you set the PCRE_NO_UTF16_CHECK flag at compile time or at
8145: run time, PCRE assumes that the pattern or subject it is given (respec-
8146: tively) contains only valid UTF-16 sequences. In this case, it does not
8147: diagnose an invalid UTF-16 string. However, if an invalid string is
8148: passed, the result is undefined.
8149:
8150: Validity of UTF-32 strings
8151:
8152: When you set the PCRE_UTF32 flag, the strings of 32-bit data units that
8153: are passed as patterns and subjects are (by default) checked for valid-
8154: ity on entry to the relevant functions. This check allows only values
8155: in the range U+0 to U+10FFFF, excluding the surrogate area U+D800 to
8156: U+DFFF.
8157:
8158: If an invalid UTF-32 string is passed to PCRE, an error return is
8159: given. At compile time, the only additional information is the offset
8160: to the first data unit of the failing character. The run-time functions
8161: pcre32_exec() and pcre32_dfa_exec() also pass back this information, as
8162: well as a more detailed reason code if the caller has provided memory
8163: in which to do this.
8164:
8165: In some situations, you may already know that your strings are valid,
8166: and therefore want to skip these checks in order to improve perfor-
8167: mance. If you set the PCRE_NO_UTF32_CHECK flag at compile time or at
8168: run time, PCRE assumes that the pattern or subject it is given (respec-
8169: tively) contains only valid UTF-32 sequences. In this case, it does not
8170: diagnose an invalid UTF-32 string. However, if an invalid string is
8171: passed, the result is undefined.
8172:
8173: General comments about UTF modes
8174:
8175: 1. Codepoints less than 256 can be specified in patterns by either
8176: braced or unbraced hexadecimal escape sequences (for example, \x{b3} or
8177: \xb3). Larger values have to use braced sequences.
8178:
8179: 2. Octal numbers up to \777 are recognized, and in UTF-8 mode they
8180: match two-byte characters for values greater than \177.
8181:
8182: 3. Repeat quantifiers apply to complete UTF characters, not to individ-
8183: ual data units, for example: \x{100}{3}.
8184:
8185: 4. The dot metacharacter matches one UTF character instead of a single
8186: data unit.
8187:
8188: 5. The escape sequence \C can be used to match a single byte in UTF-8
8189: mode, or a single 16-bit data unit in UTF-16 mode, or a single 32-bit
8190: data unit in UTF-32 mode, but its use can lead to some strange effects
8191: because it breaks up multi-unit characters (see the description of \C
8192: in the pcrepattern documentation). The use of \C is not supported in
8193: the alternative matching function pcre[16|32]_dfa_exec(), nor is it
8194: supported in UTF mode by the JIT optimization of pcre[16|32]_exec(). If
8195: JIT optimization is requested for a UTF pattern that contains \C, it
8196: will not succeed, and so the matching will be carried out by the normal
8197: interpretive function.
8198:
8199: 6. The character escapes \b, \B, \d, \D, \s, \S, \w, and \W correctly
8200: test characters of any code value, but, by default, the characters that
8201: PCRE recognizes as digits, spaces, or word characters remain the same
8202: set as in non-UTF mode, all with values less than 256. This remains
8203: true even when PCRE is built to include Unicode property support,
8204: because to do otherwise would slow down PCRE in many common cases. Note
8205: in particular that this applies to \b and \B, because they are defined
8206: in terms of \w and \W. If you really want to test for a wider sense of,
8207: say, "digit", you can use explicit Unicode property tests such as
8208: \p{Nd}. Alternatively, if you set the PCRE_UCP option, the way that the
8209: character escapes work is changed so that Unicode properties are used
8210: to determine which characters match. There are more details in the sec-
8211: tion on generic character types in the pcrepattern documentation.
8212:
8213: 7. Similarly, characters that match the POSIX named character classes
8214: are all low-valued characters, unless the PCRE_UCP option is set.
8215:
8216: 8. However, the horizontal and vertical white space matching escapes
8217: (\h, \H, \v, and \V) do match all the appropriate Unicode characters,
8218: whether or not PCRE_UCP is set.
8219:
8220: 9. Case-insensitive matching applies only to characters whose values
8221: are less than 128, unless PCRE is built with Unicode property support.
8222: A few Unicode characters such as Greek sigma have more than two code-
8223: points that are case-equivalent. Up to and including PCRE release 8.31,
8224: only one-to-one case mappings were supported, but later releases (with
8225: Unicode property support) do treat as case-equivalent all versions of
8226: characters such as Greek sigma.
8227:
8228:
8229: AUTHOR
8230:
8231: Philip Hazel
8232: University Computing Service
8233: Cambridge CB2 3QH, England.
8234:
8235:
8236: REVISION
8237:
8238: Last updated: 27 February 2013
8239: Copyright (c) 1997-2013 University of Cambridge.
8240: ------------------------------------------------------------------------------
8241:
8242:
8243: PCREJIT(3) Library Functions Manual PCREJIT(3)
8244:
8245:
8246:
8247: NAME
8248: PCRE - Perl-compatible regular expressions
8249:
8250: PCRE JUST-IN-TIME COMPILER SUPPORT
8251:
8252: Just-in-time compiling is a heavyweight optimization that can greatly
8253: speed up pattern matching. However, it comes at the cost of extra pro-
8254: cessing before the match is performed. Therefore, it is of most benefit
8255: when the same pattern is going to be matched many times. This does not
8256: necessarily mean many calls of a matching function; if the pattern is
8257: not anchored, matching attempts may take place many times at various
8258: positions in the subject, even for a single call. Therefore, if the
8259: subject string is very long, it may still pay to use JIT for one-off
8260: matches.
8261:
8262: JIT support applies only to the traditional Perl-compatible matching
8263: function. It does not apply when the DFA matching function is being
8264: used. The code for this support was written by Zoltan Herczeg.
8265:
8266:
8267: 8-BIT, 16-BIT AND 32-BIT SUPPORT
8268:
8269: JIT support is available for all of the 8-bit, 16-bit and 32-bit PCRE
8270: libraries. To keep this documentation simple, only the 8-bit interface
8271: is described in what follows. If you are using the 16-bit library, sub-
8272: stitute the 16-bit functions and 16-bit structures (for example,
8273: pcre16_jit_stack instead of pcre_jit_stack). If you are using the
8274: 32-bit library, substitute the 32-bit functions and 32-bit structures
8275: (for example, pcre32_jit_stack instead of pcre_jit_stack).
8276:
8277:
8278: AVAILABILITY OF JIT SUPPORT
8279:
8280: JIT support is an optional feature of PCRE. The "configure" option
8281: --enable-jit (or equivalent CMake option) must be set when PCRE is
8282: built if you want to use JIT. The support is limited to the following
8283: hardware platforms:
8284:
8285: ARM v5, v7, and Thumb2
8286: Intel x86 32-bit and 64-bit
8287: MIPS 32-bit
8288: Power PC 32-bit and 64-bit
8289: SPARC 32-bit (experimental)
8290:
8291: If --enable-jit is set on an unsupported platform, compilation fails.
8292:
8293: A program that is linked with PCRE 8.20 or later can tell if JIT sup-
8294: port is available by calling pcre_config() with the PCRE_CONFIG_JIT
8295: option. The result is 1 when JIT is available, and 0 otherwise. How-
8296: ever, a simple program does not need to check this in order to use JIT.
8297: The normal API is implemented in a way that falls back to the interpre-
8298: tive code if JIT is not available. For programs that need the best pos-
8299: sible performance, there is also a "fast path" API that is JIT-spe-
8300: cific.
8301:
8302: If your program may sometimes be linked with versions of PCRE that are
8303: older than 8.20, but you want to use JIT when it is available, you can
8304: test the values of PCRE_MAJOR and PCRE_MINOR, or the existence of a JIT
8305: macro such as PCRE_CONFIG_JIT, for compile-time control of your code.
8306:
8307:
8308: SIMPLE USE OF JIT
8309:
8310: You have to do two things to make use of the JIT support in the sim-
8311: plest way:
8312:
8313: (1) Call pcre_study() with the PCRE_STUDY_JIT_COMPILE option for
8314: each compiled pattern, and pass the resulting pcre_extra block to
8315: pcre_exec().
8316:
8317: (2) Use pcre_free_study() to free the pcre_extra block when it is
8318: no longer needed, instead of just freeing it yourself. This
8319: ensures that
8320: any JIT data is also freed.
8321:
8322: For a program that may be linked with pre-8.20 versions of PCRE, you
8323: can insert
8324:
8325: #ifndef PCRE_STUDY_JIT_COMPILE
8326: #define PCRE_STUDY_JIT_COMPILE 0
8327: #endif
8328:
8329: so that no option is passed to pcre_study(), and then use something
8330: like this to free the study data:
8331:
8332: #ifdef PCRE_CONFIG_JIT
8333: pcre_free_study(study_ptr);
8334: #else
8335: pcre_free(study_ptr);
8336: #endif
8337:
8338: PCRE_STUDY_JIT_COMPILE requests the JIT compiler to generate code for
8339: complete matches. If you want to run partial matches using the
8340: PCRE_PARTIAL_HARD or PCRE_PARTIAL_SOFT options of pcre_exec(), you
8341: should set one or both of the following options in addition to, or
8342: instead of, PCRE_STUDY_JIT_COMPILE when you call pcre_study():
8343:
8344: PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE
8345: PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE
8346:
8347: The JIT compiler generates different optimized code for each of the
8348: three modes (normal, soft partial, hard partial). When pcre_exec() is
8349: called, the appropriate code is run if it is available. Otherwise, the
8350: pattern is matched using interpretive code.
8351:
8352: In some circumstances you may need to call additional functions. These
8353: are described in the section entitled "Controlling the JIT stack"
8354: below.
8355:
8356: If JIT support is not available, PCRE_STUDY_JIT_COMPILE etc. are
8357: ignored, and no JIT data is created. Otherwise, the compiled pattern is
8358: passed to the JIT compiler, which turns it into machine code that exe-
8359: cutes much faster than the normal interpretive code. When pcre_exec()
8360: is passed a pcre_extra block containing a pointer to JIT code of the
8361: appropriate mode (normal or hard/soft partial), it obeys that code
8362: instead of running the interpreter. The result is identical, but the
8363: compiled JIT code runs much faster.
8364:
8365: There are some pcre_exec() options that are not supported for JIT exe-
8366: cution. There are also some pattern items that JIT cannot handle.
8367: Details are given below. In both cases, execution automatically falls
8368: back to the interpretive code. If you want to know whether JIT was
8369: actually used for a particular match, you should arrange for a JIT
8370: callback function to be set up as described in the section entitled
8371: "Controlling the JIT stack" below, even if you do not need to supply a
8372: non-default JIT stack. Such a callback function is called whenever JIT
8373: code is about to be obeyed. If the execution options are not right for
8374: JIT execution, the callback function is not obeyed.
8375:
8376: If the JIT compiler finds an unsupported item, no JIT data is gener-
8377: ated. You can find out if JIT execution is available after studying a
8378: pattern by calling pcre_fullinfo() with the PCRE_INFO_JIT option. A
8379: result of 1 means that JIT compilation was successful. A result of 0
8380: means that JIT support is not available, or the pattern was not studied
8381: with PCRE_STUDY_JIT_COMPILE etc., or the JIT compiler was not able to
8382: handle the pattern.
8383:
8384: Once a pattern has been studied, with or without JIT, it can be used as
8385: many times as you like for matching different subject strings.
8386:
8387:
8388: UNSUPPORTED OPTIONS AND PATTERN ITEMS
8389:
8390: The only pcre_exec() options that are supported for JIT execution are
8391: PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK, PCRE_NO_UTF32_CHECK, PCRE_NOT-
8392: BOL, PCRE_NOTEOL, PCRE_NOTEMPTY, PCRE_NOTEMPTY_ATSTART, PCRE_PAR-
8393: TIAL_HARD, and PCRE_PARTIAL_SOFT.
8394:
8395: The only unsupported pattern items are \C (match a single data unit)
8396: when running in a UTF mode, and a callout immediately before an asser-
8397: tion condition in a conditional group.
8398:
8399:
8400: RETURN VALUES FROM JIT EXECUTION
8401:
8402: When a pattern is matched using JIT execution, the return values are
8403: the same as those given by the interpretive pcre_exec() code, with the
8404: addition of one new error code: PCRE_ERROR_JIT_STACKLIMIT. This means
8405: that the memory used for the JIT stack was insufficient. See "Control-
8406: ling the JIT stack" below for a discussion of JIT stack usage. For com-
8407: patibility with the interpretive pcre_exec() code, no more than two-
8408: thirds of the ovector argument is used for passing back captured sub-
8409: strings.
8410:
8411: The error code PCRE_ERROR_MATCHLIMIT is returned by the JIT code if
8412: searching a very large pattern tree goes on for too long, as it is in
8413: the same circumstance when JIT is not used, but the details of exactly
8414: what is counted are not the same. The PCRE_ERROR_RECURSIONLIMIT error
8415: code is never returned by JIT execution.
8416:
8417:
8418: SAVING AND RESTORING COMPILED PATTERNS
8419:
8420: The code that is generated by the JIT compiler is architecture-spe-
8421: cific, and is also position dependent. For those reasons it cannot be
8422: saved (in a file or database) and restored later like the bytecode and
8423: other data of a compiled pattern. Saving and restoring compiled pat-
8424: terns is not something many people do. More detail about this facility
8425: is given in the pcreprecompile documentation. It should be possible to
8426: run pcre_study() on a saved and restored pattern, and thereby recreate
8427: the JIT data, but because JIT compilation uses significant resources,
8428: it is probably not worth doing this; you might as well recompile the
8429: original pattern.
8430:
8431:
8432: CONTROLLING THE JIT STACK
8433:
8434: When the compiled JIT code runs, it needs a block of memory to use as a
8435: stack. By default, it uses 32K on the machine stack. However, some
8436: large or complicated patterns need more than this. The error
8437: PCRE_ERROR_JIT_STACKLIMIT is given when there is not enough stack.
8438: Three functions are provided for managing blocks of memory for use as
8439: JIT stacks. There is further discussion about the use of JIT stacks in
8440: the section entitled "JIT stack FAQ" below.
8441:
8442: The pcre_jit_stack_alloc() function creates a JIT stack. Its arguments
8443: are a starting size and a maximum size, and it returns a pointer to an
8444: opaque structure of type pcre_jit_stack, or NULL if there is an error.
8445: The pcre_jit_stack_free() function can be used to free a stack that is
8446: no longer needed. (For the technically minded: the address space is
8447: allocated by mmap or VirtualAlloc.)
8448:
8449: JIT uses far less memory for recursion than the interpretive code, and
8450: a maximum stack size of 512K to 1M should be more than enough for any
8451: pattern.
8452:
8453: The pcre_assign_jit_stack() function specifies which stack JIT code
8454: should use. Its arguments are as follows:
8455:
8456: pcre_extra *extra
8457: pcre_jit_callback callback
8458: void *data
8459:
8460: The extra argument must be the result of studying a pattern with
8461: PCRE_STUDY_JIT_COMPILE etc. There are three cases for the values of the
8462: other two options:
8463:
8464: (1) If callback is NULL and data is NULL, an internal 32K block
8465: on the machine stack is used.
8466:
8467: (2) If callback is NULL and data is not NULL, data must be
8468: a valid JIT stack, the result of calling pcre_jit_stack_alloc().
8469:
8470: (3) If callback is not NULL, it must point to a function that is
8471: called with data as an argument at the start of matching, in
8472: order to set up a JIT stack. If the return from the callback
8473: function is NULL, the internal 32K stack is used; otherwise the
8474: return value must be a valid JIT stack, the result of calling
8475: pcre_jit_stack_alloc().
8476:
8477: A callback function is obeyed whenever JIT code is about to be run; it
8478: is not obeyed when pcre_exec() is called with options that are incom-
8479: patible for JIT execution. A callback function can therefore be used to
8480: determine whether a match operation was executed by JIT or by the
8481: interpreter.
8482:
8483: You may safely use the same JIT stack for more than one pattern (either
8484: by assigning directly or by callback), as long as the patterns are all
8485: matched sequentially in the same thread. In a multithread application,
8486: if you do not specify a JIT stack, or if you assign or pass back NULL
8487: from a callback, that is thread-safe, because each thread has its own
8488: machine stack. However, if you assign or pass back a non-NULL JIT
8489: stack, this must be a different stack for each thread so that the
8490: application is thread-safe.
8491:
8492: Strictly speaking, even more is allowed. You can assign the same non-
8493: NULL stack to any number of patterns as long as they are not used for
8494: matching by multiple threads at the same time. For example, you can
8495: assign the same stack to all compiled patterns, and use a global mutex
8496: in the callback to wait until the stack is available for use. However,
8497: this is an inefficient solution, and not recommended.
8498:
8499: This is a suggestion for how a multithreaded program that needs to set
8500: up non-default JIT stacks might operate:
8501:
8502: During thread initalization
8503: thread_local_var = pcre_jit_stack_alloc(...)
8504:
8505: During thread exit
8506: pcre_jit_stack_free(thread_local_var)
8507:
8508: Use a one-line callback function
8509: return thread_local_var
8510:
8511: All the functions described in this section do nothing if JIT is not
8512: available, and pcre_assign_jit_stack() does nothing unless the extra
8513: argument is non-NULL and points to a pcre_extra block that is the
8514: result of a successful study with PCRE_STUDY_JIT_COMPILE etc.
8515:
8516:
8517: JIT STACK FAQ
8518:
8519: (1) Why do we need JIT stacks?
8520:
8521: PCRE (and JIT) is a recursive, depth-first engine, so it needs a stack
8522: where the local data of the current node is pushed before checking its
8523: child nodes. Allocating real machine stack on some platforms is diffi-
8524: cult. For example, the stack chain needs to be updated every time if we
8525: extend the stack on PowerPC. Although it is possible, its updating
8526: time overhead decreases performance. So we do the recursion in memory.
8527:
8528: (2) Why don't we simply allocate blocks of memory with malloc()?
8529:
8530: Modern operating systems have a nice feature: they can reserve an
8531: address space instead of allocating memory. We can safely allocate mem-
8532: ory pages inside this address space, so the stack could grow without
8533: moving memory data (this is important because of pointers). Thus we can
8534: allocate 1M address space, and use only a single memory page (usually
8535: 4K) if that is enough. However, we can still grow up to 1M anytime if
8536: needed.
8537:
8538: (3) Who "owns" a JIT stack?
8539:
8540: The owner of the stack is the user program, not the JIT studied pattern
8541: or anything else. The user program must ensure that if a stack is used
8542: by pcre_exec(), (that is, it is assigned to the pattern currently run-
8543: ning), that stack must not be used by any other threads (to avoid over-
8544: writing the same memory area). The best practice for multithreaded pro-
8545: grams is to allocate a stack for each thread, and return this stack
8546: through the JIT callback function.
8547:
8548: (4) When should a JIT stack be freed?
8549:
8550: You can free a JIT stack at any time, as long as it will not be used by
8551: pcre_exec() again. When you assign the stack to a pattern, only a
8552: pointer is set. There is no reference counting or any other magic. You
8553: can free the patterns and stacks in any order, anytime. Just do not
8554: call pcre_exec() with a pattern pointing to an already freed stack, as
8555: that will cause SEGFAULT. (Also, do not free a stack currently used by
8556: pcre_exec() in another thread). You can also replace the stack for a
8557: pattern at any time. You can even free the previous stack before
8558: assigning a replacement.
8559:
8560: (5) Should I allocate/free a stack every time before/after calling
8561: pcre_exec()?
8562:
8563: No, because this is too costly in terms of resources. However, you
8564: could implement some clever idea which release the stack if it is not
8565: used in let's say two minutes. The JIT callback can help to achieve
8566: this without keeping a list of the currently JIT studied patterns.
8567:
8568: (6) OK, the stack is for long term memory allocation. But what happens
8569: if a pattern causes stack overflow with a stack of 1M? Is that 1M kept
8570: until the stack is freed?
8571:
8572: Especially on embedded sytems, it might be a good idea to release mem-
8573: ory sometimes without freeing the stack. There is no API for this at
8574: the moment. Probably a function call which returns with the currently
8575: allocated memory for any stack and another which allows releasing mem-
8576: ory (shrinking the stack) would be a good idea if someone needs this.
8577:
8578: (7) This is too much of a headache. Isn't there any better solution for
8579: JIT stack handling?
8580:
8581: No, thanks to Windows. If POSIX threads were used everywhere, we could
8582: throw out this complicated API.
8583:
8584:
8585: EXAMPLE CODE
8586:
8587: This is a single-threaded example that specifies a JIT stack without
8588: using a callback.
8589:
8590: int rc;
8591: int ovector[30];
8592: pcre *re;
8593: pcre_extra *extra;
8594: pcre_jit_stack *jit_stack;
8595:
8596: re = pcre_compile(pattern, 0, &error, &erroffset, NULL);
8597: /* Check for errors */
8598: extra = pcre_study(re, PCRE_STUDY_JIT_COMPILE, &error);
8599: jit_stack = pcre_jit_stack_alloc(32*1024, 512*1024);
8600: /* Check for error (NULL) */
8601: pcre_assign_jit_stack(extra, NULL, jit_stack);
8602: rc = pcre_exec(re, extra, subject, length, 0, 0, ovector, 30);
8603: /* Check results */
8604: pcre_free(re);
8605: pcre_free_study(extra);
8606: pcre_jit_stack_free(jit_stack);
8607:
8608:
8609: JIT FAST PATH API
8610:
8611: Because the API described above falls back to interpreted execution
8612: when JIT is not available, it is convenient for programs that are writ-
8613: ten for general use in many environments. However, calling JIT via
8614: pcre_exec() does have a performance impact. Programs that are written
8615: for use where JIT is known to be available, and which need the best
8616: possible performance, can instead use a "fast path" API to call JIT
8617: execution directly instead of calling pcre_exec() (obviously only for
8618: patterns that have been successfully studied by JIT).
8619:
8620: The fast path function is called pcre_jit_exec(), and it takes exactly
8621: the same arguments as pcre_exec(), plus one additional argument that
8622: must point to a JIT stack. The JIT stack arrangements described above
8623: do not apply. The return values are the same as for pcre_exec().
8624:
8625: When you call pcre_exec(), as well as testing for invalid options, a
8626: number of other sanity checks are performed on the arguments. For exam-
8627: ple, if the subject pointer is NULL, or its length is negative, an
8628: immediate error is given. Also, unless PCRE_NO_UTF[8|16|32] is set, a
8629: UTF subject string is tested for validity. In the interests of speed,
8630: these checks do not happen on the JIT fast path, and if invalid data is
8631: passed, the result is undefined.
8632:
8633: Bypassing the sanity checks and the pcre_exec() wrapping can give
8634: speedups of more than 10%.
8635:
8636:
8637: SEE ALSO
8638:
8639: pcreapi(3)
8640:
8641:
8642: AUTHOR
8643:
8644: Philip Hazel (FAQ by Zoltan Herczeg)
8645: University Computing Service
8646: Cambridge CB2 3QH, England.
8647:
8648:
8649: REVISION
8650:
8651: Last updated: 17 March 2013
8652: Copyright (c) 1997-2013 University of Cambridge.
8653: ------------------------------------------------------------------------------
8654:
8655:
8656: PCREPARTIAL(3) Library Functions Manual PCREPARTIAL(3)
8657:
8658:
8659:
8660: NAME
8661: PCRE - Perl-compatible regular expressions
8662:
8663: PARTIAL MATCHING IN PCRE
8664:
8665: In normal use of PCRE, if the subject string that is passed to a match-
8666: ing function matches as far as it goes, but is too short to match the
8667: entire pattern, PCRE_ERROR_NOMATCH is returned. There are circumstances
8668: where it might be helpful to distinguish this case from other cases in
8669: which there is no match.
8670:
8671: Consider, for example, an application where a human is required to type
8672: in data for a field with specific formatting requirements. An example
8673: might be a date in the form ddmmmyy, defined by this pattern:
8674:
8675: ^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$
8676:
8677: If the application sees the user's keystrokes one by one, and can check
8678: that what has been typed so far is potentially valid, it is able to
8679: raise an error as soon as a mistake is made, by beeping and not
8680: reflecting the character that has been typed, for example. This immedi-
8681: ate feedback is likely to be a better user interface than a check that
8682: is delayed until the entire string has been entered. Partial matching
8683: can also be useful when the subject string is very long and is not all
8684: available at once.
8685:
8686: PCRE supports partial matching by means of the PCRE_PARTIAL_SOFT and
8687: PCRE_PARTIAL_HARD options, which can be set when calling any of the
8688: matching functions. For backwards compatibility, PCRE_PARTIAL is a syn-
8689: onym for PCRE_PARTIAL_SOFT. The essential difference between the two
8690: options is whether or not a partial match is preferred to an alterna-
8691: tive complete match, though the details differ between the two types of
8692: matching function. If both options are set, PCRE_PARTIAL_HARD takes
8693: precedence.
8694:
8695: If you want to use partial matching with just-in-time optimized code,
8696: you must call pcre_study(), pcre16_study() or pcre32_study() with one
8697: or both of these options:
8698:
8699: PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE
8700: PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE
8701:
8702: PCRE_STUDY_JIT_COMPILE should also be set if you are going to run non-
8703: partial matches on the same pattern. If the appropriate JIT study mode
8704: has not been set for a match, the interpretive matching code is used.
8705:
8706: Setting a partial matching option disables two of PCRE's standard opti-
8707: mizations. PCRE remembers the last literal data unit in a pattern, and
8708: abandons matching immediately if it is not present in the subject
8709: string. This optimization cannot be used for a subject string that
8710: might match only partially. If the pattern was studied, PCRE knows the
8711: minimum length of a matching string, and does not bother to run the
8712: matching function on shorter strings. This optimization is also dis-
8713: abled for partial matching.
8714:
8715:
8716: PARTIAL MATCHING USING pcre_exec() OR pcre[16|32]_exec()
8717:
8718: A partial match occurs during a call to pcre_exec() or
8719: pcre[16|32]_exec() when the end of the subject string is reached suc-
8720: cessfully, but matching cannot continue because more characters are
8721: needed. However, at least one character in the subject must have been
8722: inspected. This character need not form part of the final matched
8723: string; lookbehind assertions and the \K escape sequence provide ways
8724: of inspecting characters before the start of a matched substring. The
8725: requirement for inspecting at least one character exists because an
8726: empty string can always be matched; without such a restriction there
8727: would always be a partial match of an empty string at the end of the
8728: subject.
8729:
8730: If there are at least two slots in the offsets vector when a partial
8731: match is returned, the first slot is set to the offset of the earliest
8732: character that was inspected. For convenience, the second offset points
8733: to the end of the subject so that a substring can easily be identified.
8734: If there are at least three slots in the offsets vector, the third slot
8735: is set to the offset of the character where matching started.
8736:
8737: For the majority of patterns, the contents of the first and third slots
8738: will be the same. However, for patterns that contain lookbehind asser-
8739: tions, or begin with \b or \B, characters before the one where matching
8740: started may have been inspected while carrying out the match. For exam-
8741: ple, consider this pattern:
8742:
8743: /(?<=abc)123/
8744:
8745: This pattern matches "123", but only if it is preceded by "abc". If the
8746: subject string is "xyzabc12", the first two offsets after a partial
8747: match are for the substring "abc12", because all these characters were
8748: inspected. However, the third offset is set to 6, because that is the
8749: offset where matching began.
8750:
8751: What happens when a partial match is identified depends on which of the
8752: two partial matching options are set.
8753:
8754: PCRE_PARTIAL_SOFT WITH pcre_exec() OR pcre[16|32]_exec()
8755:
8756: If PCRE_PARTIAL_SOFT is set when pcre_exec() or pcre[16|32]_exec()
8757: identifies a partial match, the partial match is remembered, but match-
8758: ing continues as normal, and other alternatives in the pattern are
8759: tried. If no complete match can be found, PCRE_ERROR_PARTIAL is
8760: returned instead of PCRE_ERROR_NOMATCH.
8761:
8762: This option is "soft" because it prefers a complete match over a par-
8763: tial match. All the various matching items in a pattern behave as if
8764: the subject string is potentially complete. For example, \z, \Z, and $
8765: match at the end of the subject, as normal, and for \b and \B the end
8766: of the subject is treated as a non-alphanumeric.
8767:
8768: If there is more than one partial match, the first one that was found
8769: provides the data that is returned. Consider this pattern:
8770:
8771: /123\w+X|dogY/
8772:
8773: If this is matched against the subject string "abc123dog", both alter-
8774: natives fail to match, but the end of the subject is reached during
8775: matching, so PCRE_ERROR_PARTIAL is returned. The offsets are set to 3
8776: and 9, identifying "123dog" as the first partial match that was found.
8777: (In this example, there are two partial matches, because "dog" on its
8778: own partially matches the second alternative.)
8779:
8780: PCRE_PARTIAL_HARD WITH pcre_exec() OR pcre[16|32]_exec()
8781:
8782: If PCRE_PARTIAL_HARD is set for pcre_exec() or pcre[16|32]_exec(),
8783: PCRE_ERROR_PARTIAL is returned as soon as a partial match is found,
8784: without continuing to search for possible complete matches. This option
8785: is "hard" because it prefers an earlier partial match over a later com-
8786: plete match. For this reason, the assumption is made that the end of
8787: the supplied subject string may not be the true end of the available
8788: data, and so, if \z, \Z, \b, \B, or $ are encountered at the end of the
8789: subject, the result is PCRE_ERROR_PARTIAL, provided that at least one
8790: character in the subject has been inspected.
8791:
8792: Setting PCRE_PARTIAL_HARD also affects the way UTF-8 and UTF-16 subject
8793: strings are checked for validity. Normally, an invalid sequence causes
8794: the error PCRE_ERROR_BADUTF8 or PCRE_ERROR_BADUTF16. However, in the
8795: special case of a truncated character at the end of the subject,
8796: PCRE_ERROR_SHORTUTF8 or PCRE_ERROR_SHORTUTF16 is returned when
8797: PCRE_PARTIAL_HARD is set.
8798:
8799: Comparing hard and soft partial matching
8800:
8801: The difference between the two partial matching options can be illus-
8802: trated by a pattern such as:
8803:
8804: /dog(sbody)?/
8805:
8806: This matches either "dog" or "dogsbody", greedily (that is, it prefers
8807: the longer string if possible). If it is matched against the string
8808: "dog" with PCRE_PARTIAL_SOFT, it yields a complete match for "dog".
8809: However, if PCRE_PARTIAL_HARD is set, the result is PCRE_ERROR_PARTIAL.
8810: On the other hand, if the pattern is made ungreedy the result is dif-
8811: ferent:
8812:
8813: /dog(sbody)??/
8814:
8815: In this case the result is always a complete match because that is
8816: found first, and matching never continues after finding a complete
8817: match. It might be easier to follow this explanation by thinking of the
8818: two patterns like this:
8819:
8820: /dog(sbody)?/ is the same as /dogsbody|dog/
8821: /dog(sbody)??/ is the same as /dog|dogsbody/
8822:
8823: The second pattern will never match "dogsbody", because it will always
8824: find the shorter match first.
8825:
8826:
8827: PARTIAL MATCHING USING pcre_dfa_exec() OR pcre[16|32]_dfa_exec()
8828:
8829: The DFA functions move along the subject string character by character,
8830: without backtracking, searching for all possible matches simultane-
8831: ously. If the end of the subject is reached before the end of the pat-
8832: tern, there is the possibility of a partial match, again provided that
8833: at least one character has been inspected.
8834:
8835: When PCRE_PARTIAL_SOFT is set, PCRE_ERROR_PARTIAL is returned only if
8836: there have been no complete matches. Otherwise, the complete matches
8837: are returned. However, if PCRE_PARTIAL_HARD is set, a partial match
8838: takes precedence over any complete matches. The portion of the string
8839: that was inspected when the longest partial match was found is set as
8840: the first matching string, provided there are at least two slots in the
8841: offsets vector.
8842:
8843: Because the DFA functions always search for all possible matches, and
8844: there is no difference between greedy and ungreedy repetition, their
8845: behaviour is different from the standard functions when PCRE_PAR-
8846: TIAL_HARD is set. Consider the string "dog" matched against the
8847: ungreedy pattern shown above:
8848:
8849: /dog(sbody)??/
8850:
8851: Whereas the standard functions stop as soon as they find the complete
8852: match for "dog", the DFA functions also find the partial match for
8853: "dogsbody", and so return that when PCRE_PARTIAL_HARD is set.
8854:
8855:
8856: PARTIAL MATCHING AND WORD BOUNDARIES
8857:
8858: If a pattern ends with one of sequences \b or \B, which test for word
8859: boundaries, partial matching with PCRE_PARTIAL_SOFT can give counter-
8860: intuitive results. Consider this pattern:
8861:
8862: /\bcat\b/
8863:
8864: This matches "cat", provided there is a word boundary at either end. If
8865: the subject string is "the cat", the comparison of the final "t" with a
8866: following character cannot take place, so a partial match is found.
8867: However, normal matching carries on, and \b matches at the end of the
8868: subject when the last character is a letter, so a complete match is
8869: found. The result, therefore, is not PCRE_ERROR_PARTIAL. Using
8870: PCRE_PARTIAL_HARD in this case does yield PCRE_ERROR_PARTIAL, because
8871: then the partial match takes precedence.
8872:
8873:
8874: FORMERLY RESTRICTED PATTERNS
8875:
8876: For releases of PCRE prior to 8.00, because of the way certain internal
8877: optimizations were implemented in the pcre_exec() function, the
8878: PCRE_PARTIAL option (predecessor of PCRE_PARTIAL_SOFT) could not be
8879: used with all patterns. From release 8.00 onwards, the restrictions no
8880: longer apply, and partial matching with can be requested for any pat-
8881: tern.
8882:
8883: Items that were formerly restricted were repeated single characters and
8884: repeated metasequences. If PCRE_PARTIAL was set for a pattern that did
8885: not conform to the restrictions, pcre_exec() returned the error code
8886: PCRE_ERROR_BADPARTIAL (-13). This error code is no longer in use. The
8887: PCRE_INFO_OKPARTIAL call to pcre_fullinfo() to find out if a compiled
8888: pattern can be used for partial matching now always returns 1.
8889:
8890:
8891: EXAMPLE OF PARTIAL MATCHING USING PCRETEST
8892:
8893: If the escape sequence \P is present in a pcretest data line, the
8894: PCRE_PARTIAL_SOFT option is used for the match. Here is a run of
8895: pcretest that uses the date example quoted above:
8896:
8897: re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
8898: data> 25jun04\P
8899: 0: 25jun04
8900: 1: jun
8901: data> 25dec3\P
8902: Partial match: 23dec3
8903: data> 3ju\P
8904: Partial match: 3ju
8905: data> 3juj\P
8906: No match
8907: data> j\P
8908: No match
8909:
8910: The first data string is matched completely, so pcretest shows the
8911: matched substrings. The remaining four strings do not match the com-
8912: plete pattern, but the first two are partial matches. Similar output is
8913: obtained if DFA matching is used.
8914:
8915: If the escape sequence \P is present more than once in a pcretest data
8916: line, the PCRE_PARTIAL_HARD option is set for the match.
8917:
8918:
8919: MULTI-SEGMENT MATCHING WITH pcre_dfa_exec() OR pcre[16|32]_dfa_exec()
8920:
8921: When a partial match has been found using a DFA matching function, it
8922: is possible to continue the match by providing additional subject data
8923: and calling the function again with the same compiled regular expres-
8924: sion, this time setting the PCRE_DFA_RESTART option. You must pass the
8925: same working space as before, because this is where details of the pre-
8926: vious partial match are stored. Here is an example using pcretest,
8927: using the \R escape sequence to set the PCRE_DFA_RESTART option (\D
8928: specifies the use of the DFA matching function):
8929:
8930: re> /^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$/
8931: data> 23ja\P\D
8932: Partial match: 23ja
8933: data> n05\R\D
8934: 0: n05
8935:
8936: The first call has "23ja" as the subject, and requests partial match-
8937: ing; the second call has "n05" as the subject for the continued
8938: (restarted) match. Notice that when the match is complete, only the
8939: last part is shown; PCRE does not retain the previously partially-
8940: matched string. It is up to the calling program to do that if it needs
8941: to.
8942:
8943: That means that, for an unanchored pattern, if a continued match fails,
8944: it is not possible to try again at a new starting point. All this
8945: facility is capable of doing is continuing with the previous match
8946: attempt. In the previous example, if the second set of data is "ug23"
8947: the result is no match, even though there would be a match for "aug23"
8948: if the entire string were given at once. Depending on the application,
8949: this may or may not be what you want. The only way to allow for start-
8950: ing again at the next character is to retain the matched part of the
8951: subject and try a new complete match.
8952:
8953: You can set the PCRE_PARTIAL_SOFT or PCRE_PARTIAL_HARD options with
8954: PCRE_DFA_RESTART to continue partial matching over multiple segments.
8955: This facility can be used to pass very long subject strings to the DFA
8956: matching functions.
8957:
8958:
8959: MULTI-SEGMENT MATCHING WITH pcre_exec() OR pcre[16|32]_exec()
8960:
8961: From release 8.00, the standard matching functions can also be used to
8962: do multi-segment matching. Unlike the DFA functions, it is not possible
8963: to restart the previous match with a new segment of data. Instead, new
8964: data must be added to the previous subject string, and the entire match
8965: re-run, starting from the point where the partial match occurred. Ear-
8966: lier data can be discarded.
8967:
8968: It is best to use PCRE_PARTIAL_HARD in this situation, because it does
8969: not treat the end of a segment as the end of the subject when matching
8970: \z, \Z, \b, \B, and $. Consider an unanchored pattern that matches
8971: dates:
8972:
8973: re> /\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d/
8974: data> The date is 23ja\P\P
8975: Partial match: 23ja
8976:
8977: At this stage, an application could discard the text preceding "23ja",
8978: add on text from the next segment, and call the matching function
8979: again. Unlike the DFA matching functions, the entire matching string
8980: must always be available, and the complete matching process occurs for
8981: each call, so more memory and more processing time is needed.
8982:
8983: Note: If the pattern contains lookbehind assertions, or \K, or starts
8984: with \b or \B, the string that is returned for a partial match includes
8985: characters that precede the start of what would be returned for a com-
8986: plete match, because it contains all the characters that were inspected
8987: during the partial match.
8988:
8989:
8990: ISSUES WITH MULTI-SEGMENT MATCHING
8991:
8992: Certain types of pattern may give problems with multi-segment matching,
8993: whichever matching function is used.
8994:
8995: 1. If the pattern contains a test for the beginning of a line, you need
8996: to pass the PCRE_NOTBOL option when the subject string for any call
8997: does start at the beginning of a line. There is also a PCRE_NOTEOL
8998: option, but in practice when doing multi-segment matching you should be
8999: using PCRE_PARTIAL_HARD, which includes the effect of PCRE_NOTEOL.
9000:
9001: 2. Lookbehind assertions that have already been obeyed are catered for
9002: in the offsets that are returned for a partial match. However a lookbe-
9003: hind assertion later in the pattern could require even earlier charac-
9004: ters to be inspected. You can handle this case by using the
9005: PCRE_INFO_MAXLOOKBEHIND option of the pcre_fullinfo() or
9006: pcre[16|32]_fullinfo() functions to obtain the length of the longest
9007: lookbehind in the pattern. This length is given in characters, not
9008: bytes. If you always retain at least that many characters before the
9009: partially matched string, all should be well. (Of course, near the
9010: start of the subject, fewer characters may be present; in that case all
9011: characters should be retained.)
9012:
9013: From release 8.33, there is a more accurate way of deciding which char-
9014: acters to retain. Instead of subtracting the length of the longest
9015: lookbehind from the earliest inspected character (offsets[0]), the
9016: match start position (offsets[2]) should be used, and the next match
9017: attempt started at the offsets[2] character by setting the startoffset
9018: argument of pcre_exec() or pcre_dfa_exec().
9019:
9020: For example, if the pattern "(?<=123)abc" is partially matched against
9021: the string "xx123a", the three offset values returned are 2, 6, and 5.
9022: This indicates that the matching process that gave a partial match
9023: started at offset 5, but the characters "123a" were all inspected. The
9024: maximum lookbehind for that pattern is 3, so taking that away from 5
9025: shows that we need only keep "123a", and the next match attempt can be
9026: started at offset 3 (that is, at "a") when further characters have been
9027: added. When the match start is not the earliest inspected character,
9028: pcretest shows it explicitly:
9029:
9030: re> "(?<=123)abc"
9031: data> xx123a\P\P
9032: Partial match at offset 5: 123a
9033:
9034: 3. Because a partial match must always contain at least one character,
9035: what might be considered a partial match of an empty string actually
9036: gives a "no match" result. For example:
9037:
9038: re> /c(?<=abc)x/
9039: data> ab\P
9040: No match
9041:
9042: If the next segment begins "cx", a match should be found, but this will
9043: only happen if characters from the previous segment are retained. For
9044: this reason, a "no match" result should be interpreted as "partial
9045: match of an empty string" when the pattern contains lookbehinds.
9046:
9047: 4. Matching a subject string that is split into multiple segments may
9048: not always produce exactly the same result as matching over one single
9049: long string, especially when PCRE_PARTIAL_SOFT is used. The section
9050: "Partial Matching and Word Boundaries" above describes an issue that
9051: arises if the pattern ends with \b or \B. Another kind of difference
9052: may occur when there are multiple matching possibilities, because (for
9053: PCRE_PARTIAL_SOFT) a partial match result is given only when there are
9054: no completed matches. This means that as soon as the shortest match has
9055: been found, continuation to a new subject segment is no longer possi-
9056: ble. Consider again this pcretest example:
9057:
9058: re> /dog(sbody)?/
9059: data> dogsb\P
9060: 0: dog
9061: data> do\P\D
9062: Partial match: do
9063: data> gsb\R\P\D
9064: 0: g
9065: data> dogsbody\D
9066: 0: dogsbody
9067: 1: dog
9068:
9069: The first data line passes the string "dogsb" to a standard matching
9070: function, setting the PCRE_PARTIAL_SOFT option. Although the string is
9071: a partial match for "dogsbody", the result is not PCRE_ERROR_PARTIAL,
9072: because the shorter string "dog" is a complete match. Similarly, when
9073: the subject is presented to a DFA matching function in several parts
9074: ("do" and "gsb" being the first two) the match stops when "dog" has
9075: been found, and it is not possible to continue. On the other hand, if
9076: "dogsbody" is presented as a single string, a DFA matching function
9077: finds both matches.
9078:
9079: Because of these problems, it is best to use PCRE_PARTIAL_HARD when
9080: matching multi-segment data. The example above then behaves differ-
9081: ently:
9082:
9083: re> /dog(sbody)?/
9084: data> dogsb\P\P
9085: Partial match: dogsb
9086: data> do\P\D
9087: Partial match: do
9088: data> gsb\R\P\P\D
9089: Partial match: gsb
9090:
9091: 5. Patterns that contain alternatives at the top level which do not all
9092: start with the same pattern item may not work as expected when
9093: PCRE_DFA_RESTART is used. For example, consider this pattern:
9094:
9095: 1234|3789
9096:
9097: If the first part of the subject is "ABC123", a partial match of the
9098: first alternative is found at offset 3. There is no partial match for
9099: the second alternative, because such a match does not start at the same
9100: point in the subject string. Attempting to continue with the string
9101: "7890" does not yield a match because only those alternatives that
9102: match at one point in the subject are remembered. The problem arises
9103: because the start of the second alternative matches within the first
9104: alternative. There is no problem with anchored patterns or patterns
9105: such as:
9106:
9107: 1234|ABCD
9108:
9109: where no string can be a partial match for both alternatives. This is
9110: not a problem if a standard matching function is used, because the
9111: entire match has to be rerun each time:
9112:
9113: re> /1234|3789/
9114: data> ABC123\P\P
9115: Partial match: 123
9116: data> 1237890
9117: 0: 3789
9118:
9119: Of course, instead of using PCRE_DFA_RESTART, the same technique of re-
9120: running the entire match can also be used with the DFA matching func-
9121: tions. Another possibility is to work with two buffers. If a partial
9122: match at offset n in the first buffer is followed by "no match" when
9123: PCRE_DFA_RESTART is used on the second buffer, you can then try a new
9124: match starting at offset n+1 in the first buffer.
9125:
9126:
9127: AUTHOR
9128:
9129: Philip Hazel
9130: University Computing Service
9131: Cambridge CB2 3QH, England.
9132:
9133:
9134: REVISION
9135:
9136: Last updated: 02 July 2013
9137: Copyright (c) 1997-2013 University of Cambridge.
9138: ------------------------------------------------------------------------------
9139:
9140:
9141: PCREPRECOMPILE(3) Library Functions Manual PCREPRECOMPILE(3)
9142:
9143:
9144:
9145: NAME
9146: PCRE - Perl-compatible regular expressions
9147:
9148: SAVING AND RE-USING PRECOMPILED PCRE PATTERNS
9149:
9150: If you are running an application that uses a large number of regular
9151: expression patterns, it may be useful to store them in a precompiled
9152: form instead of having to compile them every time the application is
9153: run. If you are not using any private character tables (see the
9154: pcre_maketables() documentation), this is relatively straightforward.
9155: If you are using private tables, it is a little bit more complicated.
9156: However, if you are using the just-in-time optimization feature, it is
9157: not possible to save and reload the JIT data.
9158:
9159: If you save compiled patterns to a file, you can copy them to a differ-
9160: ent host and run them there. If the two hosts have different endianness
9161: (byte order), you should run the pcre[16|32]_pat-
9162: tern_to_host_byte_order() function on the new host before trying to
9163: match the pattern. The matching functions return PCRE_ERROR_BADENDIAN-
9164: NESS if they detect a pattern with the wrong endianness.
9165:
9166: Compiling regular expressions with one version of PCRE for use with a
9167: different version is not guaranteed to work and may cause crashes, and
9168: saving and restoring a compiled pattern loses any JIT optimization
9169: data.
9170:
9171:
9172: SAVING A COMPILED PATTERN
9173:
9174: The value returned by pcre[16|32]_compile() points to a single block of
9175: memory that holds the compiled pattern and associated data. You can
9176: find the length of this block in bytes by calling
9177: pcre[16|32]_fullinfo() with an argument of PCRE_INFO_SIZE. You can then
9178: save the data in any appropriate manner. Here is sample code for the
9179: 8-bit library that compiles a pattern and writes it to a file. It
9180: assumes that the variable fd refers to a file that is open for output:
9181:
9182: int erroroffset, rc, size;
9183: char *error;
9184: pcre *re;
9185:
9186: re = pcre_compile("my pattern", 0, &error, &erroroffset, NULL);
9187: if (re == NULL) { ... handle errors ... }
9188: rc = pcre_fullinfo(re, NULL, PCRE_INFO_SIZE, &size);
9189: if (rc < 0) { ... handle errors ... }
9190: rc = fwrite(re, 1, size, fd);
9191: if (rc != size) { ... handle errors ... }
9192:
9193: In this example, the bytes that comprise the compiled pattern are
9194: copied exactly. Note that this is binary data that may contain any of
9195: the 256 possible byte values. On systems that make a distinction
9196: between binary and non-binary data, be sure that the file is opened for
9197: binary output.
9198:
9199: If you want to write more than one pattern to a file, you will have to
9200: devise a way of separating them. For binary data, preceding each pat-
9201: tern with its length is probably the most straightforward approach.
9202: Another possibility is to write out the data in hexadecimal instead of
9203: binary, one pattern to a line.
9204:
9205: Saving compiled patterns in a file is only one possible way of storing
9206: them for later use. They could equally well be saved in a database, or
9207: in the memory of some daemon process that passes them via sockets to
9208: the processes that want them.
9209:
9210: If the pattern has been studied, it is also possible to save the normal
9211: study data in a similar way to the compiled pattern itself. However, if
9212: the PCRE_STUDY_JIT_COMPILE was used, the just-in-time data that is cre-
9213: ated cannot be saved because it is too dependent on the current envi-
9214: ronment. When studying generates additional information,
9215: pcre[16|32]_study() returns a pointer to a pcre[16|32]_extra data
9216: block. Its format is defined in the section on matching a pattern in
9217: the pcreapi documentation. The study_data field points to the binary
9218: study data, and this is what you must save (not the pcre[16|32]_extra
9219: block itself). The length of the study data can be obtained by calling
9220: pcre[16|32]_fullinfo() with an argument of PCRE_INFO_STUDYSIZE. Remem-
9221: ber to check that pcre[16|32]_study() did return a non-NULL value
9222: before trying to save the study data.
9223:
9224:
9225: RE-USING A PRECOMPILED PATTERN
9226:
9227: Re-using a precompiled pattern is straightforward. Having reloaded it
9228: into main memory, called pcre[16|32]_pattern_to_host_byte_order() if
9229: necessary, you pass its pointer to pcre[16|32]_exec() or
9230: pcre[16|32]_dfa_exec() in the usual way.
9231:
9232: However, if you passed a pointer to custom character tables when the
9233: pattern was compiled (the tableptr argument of pcre[16|32]_compile()),
9234: you must now pass a similar pointer to pcre[16|32]_exec() or
9235: pcre[16|32]_dfa_exec(), because the value saved with the compiled pat-
9236: tern will obviously be nonsense. A field in a pcre[16|32]_extra() block
9237: is used to pass this data, as described in the section on matching a
9238: pattern in the pcreapi documentation.
9239:
9240: Warning: The tables that pcre_exec() and pcre_dfa_exec() use must be
9241: the same as those that were used when the pattern was compiled. If this
9242: is not the case, the behaviour is undefined.
9243:
9244: If you did not provide custom character tables when the pattern was
9245: compiled, the pointer in the compiled pattern is NULL, which causes the
9246: matching functions to use PCRE's internal tables. Thus, you do not need
9247: to take any special action at run time in this case.
9248:
9249: If you saved study data with the compiled pattern, you need to create
9250: your own pcre[16|32]_extra data block and set the study_data field to
9251: point to the reloaded study data. You must also set the
9252: PCRE_EXTRA_STUDY_DATA bit in the flags field to indicate that study
9253: data is present. Then pass the pcre[16|32]_extra block to the matching
9254: function in the usual way. If the pattern was studied for just-in-time
9255: optimization, that data cannot be saved, and so is lost by a
9256: save/restore cycle.
9257:
9258:
9259: COMPATIBILITY WITH DIFFERENT PCRE RELEASES
9260:
9261: In general, it is safest to recompile all saved patterns when you
9262: update to a new PCRE release, though not all updates actually require
9263: this.
9264:
9265:
9266: AUTHOR
9267:
9268: Philip Hazel
9269: University Computing Service
9270: Cambridge CB2 3QH, England.
9271:
9272:
9273: REVISION
9274:
9275: Last updated: 12 November 2013
9276: Copyright (c) 1997-2013 University of Cambridge.
9277: ------------------------------------------------------------------------------
9278:
9279:
9280: PCREPERFORM(3) Library Functions Manual PCREPERFORM(3)
9281:
9282:
9283:
9284: NAME
9285: PCRE - Perl-compatible regular expressions
9286:
9287: PCRE PERFORMANCE
9288:
9289: Two aspects of performance are discussed below: memory usage and pro-
9290: cessing time. The way you express your pattern as a regular expression
9291: can affect both of them.
9292:
9293:
9294: COMPILED PATTERN MEMORY USAGE
9295:
9296: Patterns are compiled by PCRE into a reasonably efficient interpretive
9297: code, so that most simple patterns do not use much memory. However,
9298: there is one case where the memory usage of a compiled pattern can be
9299: unexpectedly large. If a parenthesized subpattern has a quantifier with
9300: a minimum greater than 1 and/or a limited maximum, the whole subpattern
9301: is repeated in the compiled code. For example, the pattern
9302:
9303: (abc|def){2,4}
9304:
9305: is compiled as if it were
9306:
9307: (abc|def)(abc|def)((abc|def)(abc|def)?)?
9308:
9309: (Technical aside: It is done this way so that backtrack points within
9310: each of the repetitions can be independently maintained.)
9311:
9312: For regular expressions whose quantifiers use only small numbers, this
9313: is not usually a problem. However, if the numbers are large, and par-
9314: ticularly if such repetitions are nested, the memory usage can become
9315: an embarrassment. For example, the very simple pattern
9316:
9317: ((ab){1,1000}c){1,3}
9318:
9319: uses 51K bytes when compiled using the 8-bit library. When PCRE is com-
9320: piled with its default internal pointer size of two bytes, the size
9321: limit on a compiled pattern is 64K data units, and this is reached with
9322: the above pattern if the outer repetition is increased from 3 to 4.
9323: PCRE can be compiled to use larger internal pointers and thus handle
9324: larger compiled patterns, but it is better to try to rewrite your pat-
9325: tern to use less memory if you can.
9326:
9327: One way of reducing the memory usage for such patterns is to make use
9328: of PCRE's "subroutine" facility. Re-writing the above pattern as
9329:
9330: ((ab)(?2){0,999}c)(?1){0,2}
9331:
9332: reduces the memory requirements to 18K, and indeed it remains under 20K
9333: even with the outer repetition increased to 100. However, this pattern
9334: is not exactly equivalent, because the "subroutine" calls are treated
9335: as atomic groups into which there can be no backtracking if there is a
9336: subsequent matching failure. Therefore, PCRE cannot do this kind of
9337: rewriting automatically. Furthermore, there is a noticeable loss of
9338: speed when executing the modified pattern. Nevertheless, if the atomic
9339: grouping is not a problem and the loss of speed is acceptable, this
9340: kind of rewriting will allow you to process patterns that PCRE cannot
9341: otherwise handle.
9342:
9343:
9344: STACK USAGE AT RUN TIME
9345:
9346: When pcre_exec() or pcre[16|32]_exec() is used for matching, certain
9347: kinds of pattern can cause it to use large amounts of the process
9348: stack. In some environments the default process stack is quite small,
9349: and if it runs out the result is often SIGSEGV. This issue is probably
9350: the most frequently raised problem with PCRE. Rewriting your pattern
9351: can often help. The pcrestack documentation discusses this issue in
9352: detail.
9353:
9354:
9355: PROCESSING TIME
9356:
9357: Certain items in regular expression patterns are processed more effi-
9358: ciently than others. It is more efficient to use a character class like
9359: [aeiou] than a set of single-character alternatives such as
9360: (a|e|i|o|u). In general, the simplest construction that provides the
9361: required behaviour is usually the most efficient. Jeffrey Friedl's book
9362: contains a lot of useful general discussion about optimizing regular
9363: expressions for efficient performance. This document contains a few
9364: observations about PCRE.
9365:
9366: Using Unicode character properties (the \p, \P, and \X escapes) is
9367: slow, because PCRE has to use a multi-stage table lookup whenever it
9368: needs a character's property. If you can find an alternative pattern
9369: that does not use character properties, it will probably be faster.
9370:
9371: By default, the escape sequences \b, \d, \s, and \w, and the POSIX
9372: character classes such as [:alpha:] do not use Unicode properties,
9373: partly for backwards compatibility, and partly for performance reasons.
9374: However, you can set PCRE_UCP if you want Unicode character properties
9375: to be used. This can double the matching time for items such as \d,
9376: when matched with a traditional matching function; the performance loss
9377: is less with a DFA matching function, and in both cases there is not
9378: much difference for \b.
9379:
9380: When a pattern begins with .* not in parentheses, or in parentheses
9381: that are not the subject of a backreference, and the PCRE_DOTALL option
9382: is set, the pattern is implicitly anchored by PCRE, since it can match
9383: only at the start of a subject string. However, if PCRE_DOTALL is not
9384: set, PCRE cannot make this optimization, because the . metacharacter
9385: does not then match a newline, and if the subject string contains new-
9386: lines, the pattern may match from the character immediately following
9387: one of them instead of from the very start. For example, the pattern
9388:
9389: .*second
9390:
9391: matches the subject "first\nand second" (where \n stands for a newline
9392: character), with the match starting at the seventh character. In order
9393: to do this, PCRE has to retry the match starting after every newline in
9394: the subject.
9395:
9396: If you are using such a pattern with subject strings that do not con-
9397: tain newlines, the best performance is obtained by setting PCRE_DOTALL,
9398: or starting the pattern with ^.* or ^.*? to indicate explicit anchor-
9399: ing. That saves PCRE from having to scan along the subject looking for
9400: a newline to restart at.
9401:
9402: Beware of patterns that contain nested indefinite repeats. These can
9403: take a long time to run when applied to a string that does not match.
9404: Consider the pattern fragment
9405:
9406: ^(a+)*
9407:
9408: This can match "aaaa" in 16 different ways, and this number increases
9409: very rapidly as the string gets longer. (The * repeat can match 0, 1,
9410: 2, 3, or 4 times, and for each of those cases other than 0 or 4, the +
9411: repeats can match different numbers of times.) When the remainder of
9412: the pattern is such that the entire match is going to fail, PCRE has in
9413: principle to try every possible variation, and this can take an
9414: extremely long time, even for relatively short strings.
9415:
9416: An optimization catches some of the more simple cases such as
9417:
9418: (a+)*b
9419:
9420: where a literal character follows. Before embarking on the standard
9421: matching procedure, PCRE checks that there is a "b" later in the sub-
9422: ject string, and if there is not, it fails the match immediately. How-
9423: ever, when there is no following literal this optimization cannot be
9424: used. You can see the difference by comparing the behaviour of
9425:
9426: (a+)*\d
9427:
9428: with the pattern above. The former gives a failure almost instantly
9429: when applied to a whole line of "a" characters, whereas the latter
9430: takes an appreciable time with strings longer than about 20 characters.
9431:
9432: In many cases, the solution to this kind of performance issue is to use
9433: an atomic group or a possessive quantifier.
9434:
9435:
9436: AUTHOR
9437:
9438: Philip Hazel
9439: University Computing Service
9440: Cambridge CB2 3QH, England.
9441:
9442:
9443: REVISION
9444:
9445: Last updated: 25 August 2012
9446: Copyright (c) 1997-2012 University of Cambridge.
9447: ------------------------------------------------------------------------------
9448:
9449:
9450: PCREPOSIX(3) Library Functions Manual PCREPOSIX(3)
9451:
9452:
9453:
9454: NAME
9455: PCRE - Perl-compatible regular expressions.
9456:
9457: SYNOPSIS
9458:
9459: #include <pcreposix.h>
9460:
9461: int regcomp(regex_t *preg, const char *pattern,
9462: int cflags);
9463:
9464: int regexec(regex_t *preg, const char *string,
9465: size_t nmatch, regmatch_t pmatch[], int eflags);
9466: size_t regerror(int errcode, const regex_t *preg,
9467: char *errbuf, size_t errbuf_size);
9468:
9469: void regfree(regex_t *preg);
9470:
9471:
9472: DESCRIPTION
9473:
9474: This set of functions provides a POSIX-style API for the PCRE regular
9475: expression 8-bit library. See the pcreapi documentation for a descrip-
9476: tion of PCRE's native API, which contains much additional functional-
9477: ity. There is no POSIX-style wrapper for PCRE's 16-bit and 32-bit
9478: library.
9479:
9480: The functions described here are just wrapper functions that ultimately
9481: call the PCRE native API. Their prototypes are defined in the
9482: pcreposix.h header file, and on Unix systems the library itself is
9483: called pcreposix.a, so can be accessed by adding -lpcreposix to the
9484: command for linking an application that uses them. Because the POSIX
9485: functions call the native ones, it is also necessary to add -lpcre.
9486:
9487: I have implemented only those POSIX option bits that can be reasonably
9488: mapped to PCRE native options. In addition, the option REG_EXTENDED is
9489: defined with the value zero. This has no effect, but since programs
9490: that are written to the POSIX interface often use it, this makes it
9491: easier to slot in PCRE as a replacement library. Other POSIX options
9492: are not even defined.
9493:
9494: There are also some other options that are not defined by POSIX. These
9495: have been added at the request of users who want to make use of certain
9496: PCRE-specific features via the POSIX calling interface.
9497:
9498: When PCRE is called via these functions, it is only the API that is
9499: POSIX-like in style. The syntax and semantics of the regular expres-
9500: sions themselves are still those of Perl, subject to the setting of
9501: various PCRE options, as described below. "POSIX-like in style" means
9502: that the API approximates to the POSIX definition; it is not fully
9503: POSIX-compatible, and in multi-byte encoding domains it is probably
9504: even less compatible.
9505:
9506: The header for these functions is supplied as pcreposix.h to avoid any
9507: potential clash with other POSIX libraries. It can, of course, be
9508: renamed or aliased as regex.h, which is the "correct" name. It provides
9509: two structure types, regex_t for compiled internal forms, and reg-
9510: match_t for returning captured substrings. It also defines some con-
9511: stants whose names start with "REG_"; these are used for setting
9512: options and identifying error codes.
9513:
9514:
9515: COMPILING A PATTERN
9516:
9517: The function regcomp() is called to compile a pattern into an internal
9518: form. The pattern is a C string terminated by a binary zero, and is
9519: passed in the argument pattern. The preg argument is a pointer to a
9520: regex_t structure that is used as a base for storing information about
9521: the compiled regular expression.
9522:
9523: The argument cflags is either zero, or contains one or more of the bits
9524: defined by the following macros:
9525:
9526: REG_DOTALL
9527:
9528: The PCRE_DOTALL option is set when the regular expression is passed for
9529: compilation to the native function. Note that REG_DOTALL is not part of
9530: the POSIX standard.
9531:
9532: REG_ICASE
9533:
9534: The PCRE_CASELESS option is set when the regular expression is passed
9535: for compilation to the native function.
9536:
9537: REG_NEWLINE
9538:
9539: The PCRE_MULTILINE option is set when the regular expression is passed
9540: for compilation to the native function. Note that this does not mimic
9541: the defined POSIX behaviour for REG_NEWLINE (see the following sec-
9542: tion).
9543:
9544: REG_NOSUB
9545:
9546: The PCRE_NO_AUTO_CAPTURE option is set when the regular expression is
9547: passed for compilation to the native function. In addition, when a pat-
9548: tern that is compiled with this flag is passed to regexec() for match-
9549: ing, the nmatch and pmatch arguments are ignored, and no captured
9550: strings are returned.
9551:
9552: REG_UCP
9553:
9554: The PCRE_UCP option is set when the regular expression is passed for
9555: compilation to the native function. This causes PCRE to use Unicode
9556: properties when matchine \d, \w, etc., instead of just recognizing
9557: ASCII values. Note that REG_UTF8 is not part of the POSIX standard.
9558:
9559: REG_UNGREEDY
9560:
9561: The PCRE_UNGREEDY option is set when the regular expression is passed
9562: for compilation to the native function. Note that REG_UNGREEDY is not
9563: part of the POSIX standard.
9564:
9565: REG_UTF8
9566:
9567: The PCRE_UTF8 option is set when the regular expression is passed for
9568: compilation to the native function. This causes the pattern itself and
9569: all data strings used for matching it to be treated as UTF-8 strings.
9570: Note that REG_UTF8 is not part of the POSIX standard.
9571:
9572: In the absence of these flags, no options are passed to the native
9573: function. This means the the regex is compiled with PCRE default
9574: semantics. In particular, the way it handles newline characters in the
9575: subject string is the Perl way, not the POSIX way. Note that setting
9576: PCRE_MULTILINE has only some of the effects specified for REG_NEWLINE.
9577: It does not affect the way newlines are matched by . (they are not) or
9578: by a negative class such as [^a] (they are).
9579:
9580: The yield of regcomp() is zero on success, and non-zero otherwise. The
9581: preg structure is filled in on success, and one member of the structure
9582: is public: re_nsub contains the number of capturing subpatterns in the
9583: regular expression. Various error codes are defined in the header file.
9584:
9585: NOTE: If the yield of regcomp() is non-zero, you must not attempt to
9586: use the contents of the preg structure. If, for example, you pass it to
9587: regexec(), the result is undefined and your program is likely to crash.
9588:
9589:
9590: MATCHING NEWLINE CHARACTERS
9591:
9592: This area is not simple, because POSIX and Perl take different views of
9593: things. It is not possible to get PCRE to obey POSIX semantics, but
9594: then PCRE was never intended to be a POSIX engine. The following table
9595: lists the different possibilities for matching newline characters in
9596: PCRE:
9597:
9598: Default Change with
9599:
9600: . matches newline no PCRE_DOTALL
9601: newline matches [^a] yes not changeable
9602: $ matches \n at end yes PCRE_DOLLARENDONLY
9603: $ matches \n in middle no PCRE_MULTILINE
9604: ^ matches \n in middle no PCRE_MULTILINE
9605:
9606: This is the equivalent table for POSIX:
9607:
9608: Default Change with
9609:
9610: . matches newline yes REG_NEWLINE
9611: newline matches [^a] yes REG_NEWLINE
9612: $ matches \n at end no REG_NEWLINE
9613: $ matches \n in middle no REG_NEWLINE
9614: ^ matches \n in middle no REG_NEWLINE
9615:
9616: PCRE's behaviour is the same as Perl's, except that there is no equiva-
9617: lent for PCRE_DOLLAR_ENDONLY in Perl. In both PCRE and Perl, there is
9618: no way to stop newline from matching [^a].
9619:
9620: The default POSIX newline handling can be obtained by setting
9621: PCRE_DOTALL and PCRE_DOLLAR_ENDONLY, but there is no way to make PCRE
9622: behave exactly as for the REG_NEWLINE action.
9623:
9624:
9625: MATCHING A PATTERN
9626:
9627: The function regexec() is called to match a compiled pattern preg
9628: against a given string, which is by default terminated by a zero byte
9629: (but see REG_STARTEND below), subject to the options in eflags. These
9630: can be:
9631:
9632: REG_NOTBOL
9633:
9634: The PCRE_NOTBOL option is set when calling the underlying PCRE matching
9635: function.
9636:
9637: REG_NOTEMPTY
9638:
9639: The PCRE_NOTEMPTY option is set when calling the underlying PCRE match-
9640: ing function. Note that REG_NOTEMPTY is not part of the POSIX standard.
9641: However, setting this option can give more POSIX-like behaviour in some
9642: situations.
9643:
9644: REG_NOTEOL
9645:
9646: The PCRE_NOTEOL option is set when calling the underlying PCRE matching
9647: function.
9648:
9649: REG_STARTEND
9650:
9651: The string is considered to start at string + pmatch[0].rm_so and to
9652: have a terminating NUL located at string + pmatch[0].rm_eo (there need
9653: not actually be a NUL at that location), regardless of the value of
9654: nmatch. This is a BSD extension, compatible with but not specified by
9655: IEEE Standard 1003.2 (POSIX.2), and should be used with caution in
9656: software intended to be portable to other systems. Note that a non-zero
9657: rm_so does not imply REG_NOTBOL; REG_STARTEND affects only the location
9658: of the string, not how it is matched.
9659:
9660: If the pattern was compiled with the REG_NOSUB flag, no data about any
9661: matched strings is returned. The nmatch and pmatch arguments of
9662: regexec() are ignored.
9663:
9664: If the value of nmatch is zero, or if the value pmatch is NULL, no data
9665: about any matched strings is returned.
9666:
9667: Otherwise,the portion of the string that was matched, and also any cap-
9668: tured substrings, are returned via the pmatch argument, which points to
9669: an array of nmatch structures of type regmatch_t, containing the mem-
9670: bers rm_so and rm_eo. These contain the offset to the first character
9671: of each substring and the offset to the first character after the end
9672: of each substring, respectively. The 0th element of the vector relates
9673: to the entire portion of string that was matched; subsequent elements
9674: relate to the capturing subpatterns of the regular expression. Unused
9675: entries in the array have both structure members set to -1.
9676:
9677: A successful match yields a zero return; various error codes are
9678: defined in the header file, of which REG_NOMATCH is the "expected"
9679: failure code.
9680:
9681:
9682: ERROR MESSAGES
9683:
9684: The regerror() function maps a non-zero errorcode from either regcomp()
9685: or regexec() to a printable message. If preg is not NULL, the error
9686: should have arisen from the use of that structure. A message terminated
9687: by a binary zero is placed in errbuf. The length of the message,
9688: including the zero, is limited to errbuf_size. The yield of the func-
9689: tion is the size of buffer needed to hold the whole message.
9690:
9691:
9692: MEMORY USAGE
9693:
9694: Compiling a regular expression causes memory to be allocated and asso-
9695: ciated with the preg structure. The function regfree() frees all such
9696: memory, after which preg may no longer be used as a compiled expres-
9697: sion.
9698:
9699:
9700: AUTHOR
9701:
9702: Philip Hazel
9703: University Computing Service
9704: Cambridge CB2 3QH, England.
9705:
9706:
9707: REVISION
9708:
9709: Last updated: 09 January 2012
9710: Copyright (c) 1997-2012 University of Cambridge.
9711: ------------------------------------------------------------------------------
9712:
9713:
9714: PCRECPP(3) Library Functions Manual PCRECPP(3)
9715:
9716:
9717:
9718: NAME
9719: PCRE - Perl-compatible regular expressions.
9720:
9721: SYNOPSIS OF C++ WRAPPER
9722:
9723: #include <pcrecpp.h>
9724:
9725:
9726: DESCRIPTION
9727:
9728: The C++ wrapper for PCRE was provided by Google Inc. Some additional
9729: functionality was added by Giuseppe Maxia. This brief man page was con-
9730: structed from the notes in the pcrecpp.h file, which should be con-
9731: sulted for further details. Note that the C++ wrapper supports only the
9732: original 8-bit PCRE library. There is no 16-bit or 32-bit support at
9733: present.
9734:
9735:
9736: MATCHING INTERFACE
9737:
9738: The "FullMatch" operation checks that supplied text matches a supplied
9739: pattern exactly. If pointer arguments are supplied, it copies matched
9740: sub-strings that match sub-patterns into them.
9741:
9742: Example: successful match
9743: pcrecpp::RE re("h.*o");
9744: re.FullMatch("hello");
9745:
9746: Example: unsuccessful match (requires full match):
9747: pcrecpp::RE re("e");
9748: !re.FullMatch("hello");
9749:
9750: Example: creating a temporary RE object:
9751: pcrecpp::RE("h.*o").FullMatch("hello");
9752:
9753: You can pass in a "const char*" or a "string" for "text". The examples
9754: below tend to use a const char*. You can, as in the different examples
9755: above, store the RE object explicitly in a variable or use a temporary
9756: RE object. The examples below use one mode or the other arbitrarily.
9757: Either could correctly be used for any of these examples.
9758:
9759: You must supply extra pointer arguments to extract matched subpieces.
9760:
9761: Example: extracts "ruby" into "s" and 1234 into "i"
9762: int i;
9763: string s;
9764: pcrecpp::RE re("(\\w+):(\\d+)");
9765: re.FullMatch("ruby:1234", &s, &i);
9766:
9767: Example: does not try to extract any extra sub-patterns
9768: re.FullMatch("ruby:1234", &s);
9769:
9770: Example: does not try to extract into NULL
9771: re.FullMatch("ruby:1234", NULL, &i);
9772:
9773: Example: integer overflow causes failure
9774: !re.FullMatch("ruby:1234567891234", NULL, &i);
9775:
9776: Example: fails because there aren't enough sub-patterns:
9777: !pcrecpp::RE("\\w+:\\d+").FullMatch("ruby:1234", &s);
9778:
9779: Example: fails because string cannot be stored in integer
9780: !pcrecpp::RE("(.*)").FullMatch("ruby", &i);
9781:
9782: The provided pointer arguments can be pointers to any scalar numeric
9783: type, or one of:
9784:
9785: string (matched piece is copied to string)
9786: StringPiece (StringPiece is mutated to point to matched piece)
9787: T (where "bool T::ParseFrom(const char*, int)" exists)
9788: NULL (the corresponding matched sub-pattern is not copied)
9789:
9790: The function returns true iff all of the following conditions are sat-
9791: isfied:
9792:
9793: a. "text" matches "pattern" exactly;
9794:
9795: b. The number of matched sub-patterns is >= number of supplied
9796: pointers;
9797:
9798: c. The "i"th argument has a suitable type for holding the
9799: string captured as the "i"th sub-pattern. If you pass in
9800: void * NULL for the "i"th argument, or a non-void * NULL
9801: of the correct type, or pass fewer arguments than the
9802: number of sub-patterns, "i"th captured sub-pattern is
9803: ignored.
9804:
9805: CAVEAT: An optional sub-pattern that does not exist in the matched
9806: string is assigned the empty string. Therefore, the following will
9807: return false (because the empty string is not a valid number):
9808:
9809: int number;
9810: pcrecpp::RE::FullMatch("abc", "[a-z]+(\\d+)?", &number);
9811:
9812: The matching interface supports at most 16 arguments per call. If you
9813: need more, consider using the more general interface
9814: pcrecpp::RE::DoMatch. See pcrecpp.h for the signature for DoMatch.
9815:
9816: NOTE: Do not use no_arg, which is used internally to mark the end of a
9817: list of optional arguments, as a placeholder for missing arguments, as
9818: this can lead to segfaults.
9819:
9820:
9821: QUOTING METACHARACTERS
9822:
9823: You can use the "QuoteMeta" operation to insert backslashes before all
9824: potentially meaningful characters in a string. The returned string,
9825: used as a regular expression, will exactly match the original string.
9826:
9827: Example:
9828: string quoted = RE::QuoteMeta(unquoted);
9829:
9830: Note that it's legal to escape a character even if it has no special
9831: meaning in a regular expression -- so this function does that. (This
9832: also makes it identical to the perl function of the same name; see
9833: "perldoc -f quotemeta".) For example, "1.5-2.0?" becomes
9834: "1\.5\-2\.0\?".
9835:
9836:
9837: PARTIAL MATCHES
9838:
9839: You can use the "PartialMatch" operation when you want the pattern to
9840: match any substring of the text.
9841:
9842: Example: simple search for a string:
9843: pcrecpp::RE("ell").PartialMatch("hello");
9844:
9845: Example: find first number in a string:
9846: int number;
9847: pcrecpp::RE re("(\\d+)");
9848: re.PartialMatch("x*100 + 20", &number);
9849: assert(number == 100);
9850:
9851:
9852: UTF-8 AND THE MATCHING INTERFACE
9853:
9854: By default, pattern and text are plain text, one byte per character.
9855: The UTF8 flag, passed to the constructor, causes both pattern and
9856: string to be treated as UTF-8 text, still a byte stream but potentially
9857: multiple bytes per character. In practice, the text is likelier to be
9858: UTF-8 than the pattern, but the match returned may depend on the UTF8
9859: flag, so always use it when matching UTF8 text. For example, "." will
9860: match one byte normally but with UTF8 set may match up to three bytes
9861: of a multi-byte character.
9862:
9863: Example:
9864: pcrecpp::RE_Options options;
9865: options.set_utf8();
9866: pcrecpp::RE re(utf8_pattern, options);
9867: re.FullMatch(utf8_string);
9868:
9869: Example: using the convenience function UTF8():
9870: pcrecpp::RE re(utf8_pattern, pcrecpp::UTF8());
9871: re.FullMatch(utf8_string);
9872:
9873: NOTE: The UTF8 flag is ignored if pcre was not configured with the
9874: --enable-utf8 flag.
9875:
9876:
9877: PASSING MODIFIERS TO THE REGULAR EXPRESSION ENGINE
9878:
9879: PCRE defines some modifiers to change the behavior of the regular
9880: expression engine. The C++ wrapper defines an auxiliary class,
9881: RE_Options, as a vehicle to pass such modifiers to a RE class. Cur-
9882: rently, the following modifiers are supported:
9883:
9884: modifier description Perl corresponding
9885:
9886: PCRE_CASELESS case insensitive match /i
9887: PCRE_MULTILINE multiple lines match /m
9888: PCRE_DOTALL dot matches newlines /s
9889: PCRE_DOLLAR_ENDONLY $ matches only at end N/A
9890: PCRE_EXTRA strict escape parsing N/A
9891: PCRE_EXTENDED ignore white spaces /x
9892: PCRE_UTF8 handles UTF8 chars built-in
9893: PCRE_UNGREEDY reverses * and *? N/A
9894: PCRE_NO_AUTO_CAPTURE disables capturing parens N/A (*)
9895:
9896: (*) Both Perl and PCRE allow non capturing parentheses by means of the
9897: "?:" modifier within the pattern itself. e.g. (?:ab|cd) does not cap-
9898: ture, while (ab|cd) does.
9899:
9900: For a full account on how each modifier works, please check the PCRE
9901: API reference page.
9902:
9903: For each modifier, there are two member functions whose name is made
9904: out of the modifier in lowercase, without the "PCRE_" prefix. For
9905: instance, PCRE_CASELESS is handled by
9906:
9907: bool caseless()
9908:
9909: which returns true if the modifier is set, and
9910:
9911: RE_Options & set_caseless(bool)
9912:
9913: which sets or unsets the modifier. Moreover, PCRE_EXTRA_MATCH_LIMIT can
9914: be accessed through the set_match_limit() and match_limit() member
9915: functions. Setting match_limit to a non-zero value will limit the exe-
9916: cution of pcre to keep it from doing bad things like blowing the stack
9917: or taking an eternity to return a result. A value of 5000 is good
9918: enough to stop stack blowup in a 2MB thread stack. Setting match_limit
9919: to zero disables match limiting. Alternatively, you can call
9920: match_limit_recursion() which uses PCRE_EXTRA_MATCH_LIMIT_RECURSION to
9921: limit how much PCRE recurses. match_limit() limits the number of
9922: matches PCRE does; match_limit_recursion() limits the depth of internal
9923: recursion, and therefore the amount of stack that is used.
9924:
9925: Normally, to pass one or more modifiers to a RE class, you declare a
9926: RE_Options object, set the appropriate options, and pass this object to
9927: a RE constructor. Example:
9928:
9929: RE_Options opt;
9930: opt.set_caseless(true);
9931: if (RE("HELLO", opt).PartialMatch("hello world")) ...
9932:
9933: RE_options has two constructors. The default constructor takes no argu-
9934: ments and creates a set of flags that are off by default. The optional
9935: parameter option_flags is to facilitate transfer of legacy code from C
9936: programs. This lets you do
9937:
9938: RE(pattern,
9939: RE_Options(PCRE_CASELESS|PCRE_MULTILINE)).PartialMatch(str);
9940:
9941: However, new code is better off doing
9942:
9943: RE(pattern,
9944: RE_Options().set_caseless(true).set_multiline(true))
9945: .PartialMatch(str);
9946:
9947: If you are going to pass one of the most used modifiers, there are some
9948: convenience functions that return a RE_Options class with the appropri-
9949: ate modifier already set: CASELESS(), UTF8(), MULTILINE(), DOTALL(),
9950: and EXTENDED().
9951:
9952: If you need to set several options at once, and you don't want to go
9953: through the pains of declaring a RE_Options object and setting several
9954: options, there is a parallel method that give you such ability on the
9955: fly. You can concatenate several set_xxxxx() member functions, since
9956: each of them returns a reference to its class object. For example, to
9957: pass PCRE_CASELESS, PCRE_EXTENDED, and PCRE_MULTILINE to a RE with one
9958: statement, you may write:
9959:
9960: RE(" ^ xyz \\s+ .* blah$",
9961: RE_Options()
9962: .set_caseless(true)
9963: .set_extended(true)
9964: .set_multiline(true)).PartialMatch(sometext);
9965:
9966:
9967: SCANNING TEXT INCREMENTALLY
9968:
9969: The "Consume" operation may be useful if you want to repeatedly match
9970: regular expressions at the front of a string and skip over them as they
9971: match. This requires use of the "StringPiece" type, which represents a
9972: sub-range of a real string. Like RE, StringPiece is defined in the
9973: pcrecpp namespace.
9974:
9975: Example: read lines of the form "var = value" from a string.
9976: string contents = ...; // Fill string somehow
9977: pcrecpp::StringPiece input(contents); // Wrap in a StringPiece
9978:
9979: string var;
9980: int value;
9981: pcrecpp::RE re("(\\w+) = (\\d+)\n");
9982: while (re.Consume(&input, &var, &value)) {
9983: ...;
9984: }
9985:
9986: Each successful call to "Consume" will set "var/value", and also
9987: advance "input" so it points past the matched text.
9988:
9989: The "FindAndConsume" operation is similar to "Consume" but does not
9990: anchor your match at the beginning of the string. For example, you
9991: could extract all words from a string by repeatedly calling
9992:
9993: pcrecpp::RE("(\\w+)").FindAndConsume(&input, &word)
9994:
9995:
9996: PARSING HEX/OCTAL/C-RADIX NUMBERS
9997:
9998: By default, if you pass a pointer to a numeric value, the corresponding
9999: text is interpreted as a base-10 number. You can instead wrap the
10000: pointer with a call to one of the operators Hex(), Octal(), or CRadix()
10001: to interpret the text in another base. The CRadix operator interprets
10002: C-style "0" (base-8) and "0x" (base-16) prefixes, but defaults to
10003: base-10.
10004:
10005: Example:
10006: int a, b, c, d;
10007: pcrecpp::RE re("(.*) (.*) (.*) (.*)");
10008: re.FullMatch("100 40 0100 0x40",
10009: pcrecpp::Octal(&a), pcrecpp::Hex(&b),
10010: pcrecpp::CRadix(&c), pcrecpp::CRadix(&d));
10011:
10012: will leave 64 in a, b, c, and d.
10013:
10014:
10015: REPLACING PARTS OF STRINGS
10016:
10017: You can replace the first match of "pattern" in "str" with "rewrite".
10018: Within "rewrite", backslash-escaped digits (\1 to \9) can be used to
10019: insert text matching corresponding parenthesized group from the pat-
10020: tern. \0 in "rewrite" refers to the entire matching text. For example:
10021:
10022: string s = "yabba dabba doo";
10023: pcrecpp::RE("b+").Replace("d", &s);
10024:
10025: will leave "s" containing "yada dabba doo". The result is true if the
10026: pattern matches and a replacement occurs, false otherwise.
10027:
10028: GlobalReplace is like Replace except that it replaces all occurrences
10029: of the pattern in the string with the rewrite. Replacements are not
10030: subject to re-matching. For example:
10031:
10032: string s = "yabba dabba doo";
10033: pcrecpp::RE("b+").GlobalReplace("d", &s);
10034:
10035: will leave "s" containing "yada dada doo". It returns the number of
10036: replacements made.
10037:
10038: Extract is like Replace, except that if the pattern matches, "rewrite"
10039: is copied into "out" (an additional argument) with substitutions. The
10040: non-matching portions of "text" are ignored. Returns true iff a match
10041: occurred and the extraction happened successfully; if no match occurs,
10042: the string is left unaffected.
10043:
10044:
10045: AUTHOR
10046:
10047: The C++ wrapper was contributed by Google Inc.
10048: Copyright (c) 2007 Google Inc.
10049:
10050:
10051: REVISION
10052:
10053: Last updated: 08 January 2012
10054: ------------------------------------------------------------------------------
10055:
10056:
10057: PCRESAMPLE(3) Library Functions Manual PCRESAMPLE(3)
10058:
10059:
10060:
10061: NAME
10062: PCRE - Perl-compatible regular expressions
10063:
10064: PCRE SAMPLE PROGRAM
10065:
10066: A simple, complete demonstration program, to get you started with using
10067: PCRE, is supplied in the file pcredemo.c in the PCRE distribution. A
10068: listing of this program is given in the pcredemo documentation. If you
10069: do not have a copy of the PCRE distribution, you can save this listing
10070: to re-create pcredemo.c.
10071:
10072: The demonstration program, which uses the original PCRE 8-bit library,
10073: compiles the regular expression that is its first argument, and matches
10074: it against the subject string in its second argument. No PCRE options
10075: are set, and default character tables are used. If matching succeeds,
10076: the program outputs the portion of the subject that matched, together
10077: with the contents of any captured substrings.
10078:
10079: If the -g option is given on the command line, the program then goes on
10080: to check for further matches of the same regular expression in the same
10081: subject string. The logic is a little bit tricky because of the possi-
10082: bility of matching an empty string. Comments in the code explain what
10083: is going on.
10084:
10085: If PCRE is installed in the standard include and library directories
10086: for your operating system, you should be able to compile the demonstra-
10087: tion program using this command:
10088:
10089: gcc -o pcredemo pcredemo.c -lpcre
10090:
10091: If PCRE is installed elsewhere, you may need to add additional options
10092: to the command line. For example, on a Unix-like system that has PCRE
10093: installed in /usr/local, you can compile the demonstration program
10094: using a command like this:
10095:
10096: gcc -o pcredemo -I/usr/local/include pcredemo.c \
10097: -L/usr/local/lib -lpcre
10098:
10099: In a Windows environment, if you want to statically link the program
10100: against a non-dll pcre.a file, you must uncomment the line that defines
10101: PCRE_STATIC before including pcre.h, because otherwise the pcre_mal-
10102: loc() and pcre_free() exported functions will be declared
10103: __declspec(dllimport), with unwanted results.
10104:
10105: Once you have compiled and linked the demonstration program, you can
10106: run simple tests like this:
10107:
10108: ./pcredemo 'cat|dog' 'the cat sat on the mat'
10109: ./pcredemo -g 'cat|dog' 'the dog sat on the cat'
10110:
10111: Note that there is a much more comprehensive test program, called
10112: pcretest, which supports many more facilities for testing regular
10113: expressions and both PCRE libraries. The pcredemo program is provided
10114: as a simple coding example.
10115:
10116: If you try to run pcredemo when PCRE is not installed in the standard
10117: library directory, you may get an error like this on some operating
10118: systems (e.g. Solaris):
10119:
10120: ld.so.1: a.out: fatal: libpcre.so.0: open failed: No such file or
10121: directory
10122:
10123: This is caused by the way shared library support works on those sys-
10124: tems. You need to add
10125:
10126: -R/usr/local/lib
10127:
10128: (for example) to the compile command to get round this problem.
10129:
10130:
10131: AUTHOR
10132:
10133: Philip Hazel
10134: University Computing Service
10135: Cambridge CB2 3QH, England.
10136:
10137:
10138: REVISION
10139:
10140: Last updated: 10 January 2012
10141: Copyright (c) 1997-2012 University of Cambridge.
10142: ------------------------------------------------------------------------------
10143: PCRELIMITS(3) Library Functions Manual PCRELIMITS(3)
10144:
10145:
10146:
10147: NAME
10148: PCRE - Perl-compatible regular expressions
10149:
10150: SIZE AND OTHER LIMITATIONS
10151:
10152: There are some size limitations in PCRE but it is hoped that they will
10153: never in practice be relevant.
10154:
10155: The maximum length of a compiled pattern is approximately 64K data
10156: units (bytes for the 8-bit library, 16-bit units for the 16-bit
10157: library, and 32-bit units for the 32-bit library) if PCRE is compiled
10158: with the default internal linkage size, which is 2 bytes for the 8-bit
10159: and 16-bit libraries, and 4 bytes for the 32-bit library. If you want
10160: to process regular expressions that are truly enormous, you can compile
10161: PCRE with an internal linkage size of 3 or 4 (when building the 16-bit
10162: or 32-bit library, 3 is rounded up to 4). See the README file in the
10163: source distribution and the pcrebuild documentation for details. In
10164: these cases the limit is substantially larger. However, the speed of
10165: execution is slower.
10166:
10167: All values in repeating quantifiers must be less than 65536.
10168:
10169: There is no limit to the number of parenthesized subpatterns, but there
10170: can be no more than 65535 capturing subpatterns. There is, however, a
10171: limit to the depth of nesting of parenthesized subpatterns of all
10172: kinds. This is imposed in order to limit the amount of system stack
10173: used at compile time. The limit can be specified when PCRE is built;
10174: the default is 250.
10175:
10176: There is a limit to the number of forward references to subsequent sub-
10177: patterns of around 200,000. Repeated forward references with fixed
10178: upper limits, for example, (?2){0,100} when subpattern number 2 is to
10179: the right, are included in the count. There is no limit to the number
10180: of backward references.
10181:
10182: The maximum length of name for a named subpattern is 32 characters, and
10183: the maximum number of named subpatterns is 10000.
10184:
10185: The maximum length of a name in a (*MARK), (*PRUNE), (*SKIP), or
10186: (*THEN) verb is 255 for the 8-bit library and 65535 for the 16-bit and
10187: 32-bit libraries.
10188:
10189: The maximum length of a subject string is the largest positive number
10190: that an integer variable can hold. However, when using the traditional
10191: matching function, PCRE uses recursion to handle subpatterns and indef-
10192: inite repetition. This means that the available stack space may limit
10193: the size of a subject string that can be processed by certain patterns.
10194: For a discussion of stack issues, see the pcrestack documentation.
10195:
10196:
10197: AUTHOR
10198:
10199: Philip Hazel
10200: University Computing Service
10201: Cambridge CB2 3QH, England.
10202:
10203:
10204: REVISION
10205:
10206: Last updated: 05 November 2013
10207: Copyright (c) 1997-2013 University of Cambridge.
10208: ------------------------------------------------------------------------------
10209:
10210:
10211: PCRESTACK(3) Library Functions Manual PCRESTACK(3)
10212:
10213:
10214:
10215: NAME
10216: PCRE - Perl-compatible regular expressions
10217:
10218: PCRE DISCUSSION OF STACK USAGE
10219:
10220: When you call pcre[16|32]_exec(), it makes use of an internal function
10221: called match(). This calls itself recursively at branch points in the
10222: pattern, in order to remember the state of the match so that it can
10223: back up and try a different alternative if the first one fails. As
10224: matching proceeds deeper and deeper into the tree of possibilities, the
10225: recursion depth increases. The match() function is also called in other
10226: circumstances, for example, whenever a parenthesized sub-pattern is
10227: entered, and in certain cases of repetition.
10228:
10229: Not all calls of match() increase the recursion depth; for an item such
10230: as a* it may be called several times at the same level, after matching
10231: different numbers of a's. Furthermore, in a number of cases where the
10232: result of the recursive call would immediately be passed back as the
10233: result of the current call (a "tail recursion"), the function is just
10234: restarted instead.
10235:
10236: The above comments apply when pcre[16|32]_exec() is run in its normal
10237: interpretive manner. If the pattern was studied with the
10238: PCRE_STUDY_JIT_COMPILE option, and just-in-time compiling was success-
10239: ful, and the options passed to pcre[16|32]_exec() were not incompati-
10240: ble, the matching process uses the JIT-compiled code instead of the
10241: match() function. In this case, the memory requirements are handled
10242: entirely differently. See the pcrejit documentation for details.
10243:
10244: The pcre[16|32]_dfa_exec() function operates in an entirely different
10245: way, and uses recursion only when there is a regular expression recur-
10246: sion or subroutine call in the pattern. This includes the processing of
10247: assertion and "once-only" subpatterns, which are handled like subrou-
10248: tine calls. Normally, these are never very deep, and the limit on the
10249: complexity of pcre[16|32]_dfa_exec() is controlled by the amount of
10250: workspace it is given. However, it is possible to write patterns with
10251: runaway infinite recursions; such patterns will cause
10252: pcre[16|32]_dfa_exec() to run out of stack. At present, there is no
10253: protection against this.
10254:
10255: The comments that follow do NOT apply to pcre[16|32]_dfa_exec(); they
10256: are relevant only for pcre[16|32]_exec() without the JIT optimization.
10257:
10258: Reducing pcre[16|32]_exec()'s stack usage
10259:
10260: Each time that match() is actually called recursively, it uses memory
10261: from the process stack. For certain kinds of pattern and data, very
10262: large amounts of stack may be needed, despite the recognition of "tail
10263: recursion". You can often reduce the amount of recursion, and there-
10264: fore the amount of stack used, by modifying the pattern that is being
10265: matched. Consider, for example, this pattern:
10266:
10267: ([^<]|<(?!inet))+
10268:
10269: It matches from wherever it starts until it encounters "<inet" or the
10270: end of the data, and is the kind of pattern that might be used when
10271: processing an XML file. Each iteration of the outer parentheses matches
10272: either one character that is not "<" or a "<" that is not followed by
10273: "inet". However, each time a parenthesis is processed, a recursion
10274: occurs, so this formulation uses a stack frame for each matched charac-
10275: ter. For a long string, a lot of stack is required. Consider now this
10276: rewritten pattern, which matches exactly the same strings:
10277:
10278: ([^<]++|<(?!inet))+
10279:
10280: This uses very much less stack, because runs of characters that do not
10281: contain "<" are "swallowed" in one item inside the parentheses. Recur-
10282: sion happens only when a "<" character that is not followed by "inet"
10283: is encountered (and we assume this is relatively rare). A possessive
10284: quantifier is used to stop any backtracking into the runs of non-"<"
10285: characters, but that is not related to stack usage.
10286:
10287: This example shows that one way of avoiding stack problems when match-
10288: ing long subject strings is to write repeated parenthesized subpatterns
10289: to match more than one character whenever possible.
10290:
10291: Compiling PCRE to use heap instead of stack for pcre[16|32]_exec()
10292:
10293: In environments where stack memory is constrained, you might want to
10294: compile PCRE to use heap memory instead of stack for remembering back-
10295: up points when pcre[16|32]_exec() is running. This makes it run a lot
10296: more slowly, however. Details of how to do this are given in the pcre-
10297: build documentation. When built in this way, instead of using the
10298: stack, PCRE obtains and frees memory by calling the functions that are
10299: pointed to by the pcre[16|32]_stack_malloc and pcre[16|32]_stack_free
10300: variables. By default, these point to malloc() and free(), but you can
10301: replace the pointers to cause PCRE to use your own functions. Since the
10302: block sizes are always the same, and are always freed in reverse order,
10303: it may be possible to implement customized memory handlers that are
10304: more efficient than the standard functions.
10305:
10306: Limiting pcre[16|32]_exec()'s stack usage
10307:
10308: You can set limits on the number of times that match() is called, both
10309: in total and recursively. If a limit is exceeded, pcre[16|32]_exec()
10310: returns an error code. Setting suitable limits should prevent it from
10311: running out of stack. The default values of the limits are very large,
10312: and unlikely ever to operate. They can be changed when PCRE is built,
10313: and they can also be set when pcre[16|32]_exec() is called. For details
10314: of these interfaces, see the pcrebuild documentation and the section on
10315: extra data for pcre[16|32]_exec() in the pcreapi documentation.
10316:
10317: As a very rough rule of thumb, you should reckon on about 500 bytes per
10318: recursion. Thus, if you want to limit your stack usage to 8Mb, you
10319: should set the limit at 16000 recursions. A 64Mb stack, on the other
10320: hand, can support around 128000 recursions.
10321:
10322: In Unix-like environments, the pcretest test program has a command line
10323: option (-S) that can be used to increase the size of its stack. As long
10324: as the stack is large enough, another option (-M) can be used to find
10325: the smallest limits that allow a particular pattern to match a given
10326: subject string. This is done by calling pcre[16|32]_exec() repeatedly
10327: with different limits.
10328:
10329: Obtaining an estimate of stack usage
10330:
10331: The actual amount of stack used per recursion can vary quite a lot,
10332: depending on the compiler that was used to build PCRE and the optimiza-
10333: tion or debugging options that were set for it. The rule of thumb value
10334: of 500 bytes mentioned above may be larger or smaller than what is
10335: actually needed. A better approximation can be obtained by running this
10336: command:
10337:
10338: pcretest -m -C
10339:
10340: The -C option causes pcretest to output information about the options
10341: with which PCRE was compiled. When -m is also given (before -C), infor-
10342: mation about stack use is given in a line like this:
10343:
10344: Match recursion uses stack: approximate frame size = 640 bytes
10345:
10346: The value is approximate because some recursions need a bit more (up to
10347: perhaps 16 more bytes).
10348:
10349: If the above command is given when PCRE is compiled to use the heap
10350: instead of the stack for recursion, the value that is output is the
10351: size of each block that is obtained from the heap.
10352:
10353: Changing stack size in Unix-like systems
10354:
10355: In Unix-like environments, there is not often a problem with the stack
10356: unless very long strings are involved, though the default limit on
10357: stack size varies from system to system. Values from 8Mb to 64Mb are
10358: common. You can find your default limit by running the command:
10359:
10360: ulimit -s
10361:
10362: Unfortunately, the effect of running out of stack is often SIGSEGV,
10363: though sometimes a more explicit error message is given. You can nor-
10364: mally increase the limit on stack size by code such as this:
10365:
10366: struct rlimit rlim;
10367: getrlimit(RLIMIT_STACK, &rlim);
10368: rlim.rlim_cur = 100*1024*1024;
10369: setrlimit(RLIMIT_STACK, &rlim);
10370:
10371: This reads the current limits (soft and hard) using getrlimit(), then
10372: attempts to increase the soft limit to 100Mb using setrlimit(). You
10373: must do this before calling pcre[16|32]_exec().
10374:
10375: Changing stack size in Mac OS X
10376:
10377: Using setrlimit(), as described above, should also work on Mac OS X. It
10378: is also possible to set a stack size when linking a program. There is a
10379: discussion about stack sizes in Mac OS X at this web site:
10380: http://developer.apple.com/qa/qa2005/qa1419.html.
10381:
10382:
10383: AUTHOR
10384:
10385: Philip Hazel
10386: University Computing Service
10387: Cambridge CB2 3QH, England.
10388:
10389:
10390: REVISION
10391:
10392: Last updated: 24 June 2012
10393: Copyright (c) 1997-2012 University of Cambridge.
10394: ------------------------------------------------------------------------------
10395:
10396:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>
![[BACK]](/icons/cvsweb/back.gif){kind=icon}
Return to pcre.txt CVS log ![[TXT]](/icons/cvsweb/text.gif){kind=icon}
![[DIR]](/icons/cvsweb/dir.gif){kind=icon}
Up to [ELWIX - Embedded LightWeight unIX -] / embedaddon / pcre / doc