ViewVC logotype

Contents of /code/trunk/doc/pcrepattern.3

Parent Directory Parent Directory | Revision Log Revision Log

Revision 572 - (show annotations)
Wed Nov 17 17:55:57 2010 UTC (9 years, 3 months ago) by ph10
File size: 112119 byte(s)
Error occurred while calculating annotation data.
Documentation updates and tidies.
3 PCRE - Perl-compatible regular expressions
5 .rs
6 .sp
7 The syntax and semantics of the regular expressions that are supported by PCRE
8 are described in detail below. There is a quick-reference syntax summary in the
9 .\" HREF
10 \fBpcresyntax\fP
11 .\"
12 page. PCRE tries to match Perl syntax and semantics as closely as it can. PCRE
13 also supports some alternative regular expression syntax (which does not
14 conflict with the Perl syntax) in order to provide some compatibility with
15 regular expressions in Python, .NET, and Oniguruma.
16 .P
17 Perl's regular expressions are described in its own documentation, and
18 regular expressions in general are covered in a number of books, some of which
19 have copious examples. Jeffrey Friedl's "Mastering Regular Expressions",
20 published by O'Reilly, covers regular expressions in great detail. This
21 description of PCRE's regular expressions is intended as reference material.
22 .P
23 The original operation of PCRE was on strings of one-byte characters. However,
24 there is now also support for UTF-8 character strings. To use this,
25 PCRE must be built to include UTF-8 support, and you must call
26 \fBpcre_compile()\fP or \fBpcre_compile2()\fP with the PCRE_UTF8 option. There
27 is also a special sequence that can be given at the start of a pattern:
28 .sp
29 (*UTF8)
30 .sp
31 Starting a pattern with this sequence is equivalent to setting the PCRE_UTF8
32 option. This feature is not Perl-compatible. How setting UTF-8 mode affects
33 pattern matching is mentioned in several places below. There is also a summary
34 of UTF-8 features in the
35 .\" HTML <a href="pcre.html#utf8support">
36 .\" </a>
37 section on UTF-8 support
38 .\"
39 in the main
40 .\" HREF
41 \fBpcre\fP
42 .\"
43 page.
44 .P
45 Another special sequence that may appear at the start of a pattern or in
46 combination with (*UTF8) is:
47 .sp
48 (*UCP)
49 .sp
50 This has the same effect as setting the PCRE_UCP option: it causes sequences
51 such as \ed and \ew to use Unicode properties to determine character types,
52 instead of recognizing only characters with codes less than 128 via a lookup
53 table.
54 .P
55 The remainder of this document discusses the patterns that are supported by
56 PCRE when its main matching function, \fBpcre_exec()\fP, is used.
57 From release 6.0, PCRE offers a second matching function,
58 \fBpcre_dfa_exec()\fP, which matches using a different algorithm that is not
59 Perl-compatible. Some of the features discussed below are not available when
60 \fBpcre_dfa_exec()\fP is used. The advantages and disadvantages of the
61 alternative function, and how it differs from the normal function, are
62 discussed in the
63 .\" HREF
64 \fBpcrematching\fP
65 .\"
66 page.
67 .
68 .
69 .\" HTML <a name="newlines"></a>
71 .rs
72 .sp
73 PCRE supports five different conventions for indicating line breaks in
74 strings: a single CR (carriage return) character, a single LF (linefeed)
75 character, the two-character sequence CRLF, any of the three preceding, or any
76 Unicode newline sequence. The
77 .\" HREF
78 \fBpcreapi\fP
79 .\"
80 page has
81 .\" HTML <a href="pcreapi.html#newlines">
82 .\" </a>
83 further discussion
84 .\"
85 about newlines, and shows how to set the newline convention in the
86 \fIoptions\fP arguments for the compiling and matching functions.
87 .P
88 It is also possible to specify a newline convention by starting a pattern
89 string with one of the following five sequences:
90 .sp
91 (*CR) carriage return
92 (*LF) linefeed
93 (*CRLF) carriage return, followed by linefeed
94 (*ANYCRLF) any of the three above
95 (*ANY) all Unicode newline sequences
96 .sp
97 These override the default and the options given to \fBpcre_compile()\fP or
98 \fBpcre_compile2()\fP. For example, on a Unix system where LF is the default
99 newline sequence, the pattern
100 .sp
101 (*CR)a.b
102 .sp
103 changes the convention to CR. That pattern matches "a\enb" because LF is no
104 longer a newline. Note that these special settings, which are not
105 Perl-compatible, are recognized only at the very start of a pattern, and that
106 they must be in upper case. If more than one of them is present, the last one
107 is used.
108 .P
109 The newline convention affects the interpretation of the dot metacharacter when
110 PCRE_DOTALL is not set, and also the behaviour of \eN. However, it does not
111 affect what the \eR escape sequence matches. By default, this is any Unicode
112 newline sequence, for Perl compatibility. However, this can be changed; see the
113 description of \eR in the section entitled
114 .\" HTML <a href="#newlineseq">
115 .\" </a>
116 "Newline sequences"
117 .\"
118 below. A change of \eR setting can be combined with a change of newline
119 convention.
120 .
121 .
123 .rs
124 .sp
125 A regular expression is a pattern that is matched against a subject string from
126 left to right. Most characters stand for themselves in a pattern, and match the
127 corresponding characters in the subject. As a trivial example, the pattern
128 .sp
129 The quick brown fox
130 .sp
131 matches a portion of a subject string that is identical to itself. When
132 caseless matching is specified (the PCRE_CASELESS option), letters are matched
133 independently of case. In UTF-8 mode, PCRE always understands the concept of
134 case for characters whose values are less than 128, so caseless matching is
135 always possible. For characters with higher values, the concept of case is
136 supported if PCRE is compiled with Unicode property support, but not otherwise.
137 If you want to use caseless matching for characters 128 and above, you must
138 ensure that PCRE is compiled with Unicode property support as well as with
139 UTF-8 support.
140 .P
141 The power of regular expressions comes from the ability to include alternatives
142 and repetitions in the pattern. These are encoded in the pattern by the use of
143 \fImetacharacters\fP, which do not stand for themselves but instead are
144 interpreted in some special way.
145 .P
146 There are two different sets of metacharacters: those that are recognized
147 anywhere in the pattern except within square brackets, and those that are
148 recognized within square brackets. Outside square brackets, the metacharacters
149 are as follows:
150 .sp
151 \e general escape character with several uses
152 ^ assert start of string (or line, in multiline mode)
153 $ assert end of string (or line, in multiline mode)
154 . match any character except newline (by default)
155 [ start character class definition
156 | start of alternative branch
157 ( start subpattern
158 ) end subpattern
159 ? extends the meaning of (
160 also 0 or 1 quantifier
161 also quantifier minimizer
162 * 0 or more quantifier
163 + 1 or more quantifier
164 also "possessive quantifier"
165 { start min/max quantifier
166 .sp
167 Part of a pattern that is in square brackets is called a "character class". In
168 a character class the only metacharacters are:
169 .sp
170 \e general escape character
171 ^ negate the class, but only if the first character
172 - indicates character range
173 .\" JOIN
174 [ POSIX character class (only if followed by POSIX
175 syntax)
176 ] terminates the character class
177 .sp
178 The following sections describe the use of each of the metacharacters.
179 .
180 .
182 .rs
183 .sp
184 The backslash character has several uses. Firstly, if it is followed by a
185 non-alphanumeric character, it takes away any special meaning that character
186 may have. This use of backslash as an escape character applies both inside and
187 outside character classes.
188 .P
189 For example, if you want to match a * character, you write \e* in the pattern.
190 This escaping action applies whether or not the following character would
191 otherwise be interpreted as a metacharacter, so it is always safe to precede a
192 non-alphanumeric with backslash to specify that it stands for itself. In
193 particular, if you want to match a backslash, you write \e\e.
194 .P
195 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
196 pattern (other than in a character class) and characters between a # outside
197 a character class and the next newline are ignored. An escaping backslash can
198 be used to include a whitespace or # character as part of the pattern.
199 .P
200 If you want to remove the special meaning from a sequence of characters, you
201 can do so by putting them between \eQ and \eE. This is different from Perl in
202 that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
203 Perl, $ and @ cause variable interpolation. Note the following examples:
204 .sp
205 Pattern PCRE matches Perl matches
206 .sp
207 .\" JOIN
208 \eQabc$xyz\eE abc$xyz abc followed by the
209 contents of $xyz
210 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
211 \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
212 .sp
213 The \eQ...\eE sequence is recognized both inside and outside character classes.
214 An isolated \eE that is not preceded by \eQ is ignored.
215 .
216 .
217 .\" HTML <a name="digitsafterbackslash"></a>
218 .SS "Non-printing characters"
219 .rs
220 .sp
221 A second use of backslash provides a way of encoding non-printing characters
222 in patterns in a visible manner. There is no restriction on the appearance of
223 non-printing characters, apart from the binary zero that terminates a pattern,
224 but when a pattern is being prepared by text editing, it is often easier to use
225 one of the following escape sequences than the binary character it represents:
226 .sp
227 \ea alarm, that is, the BEL character (hex 07)
228 \ecx "control-x", where x is any character
229 \ee escape (hex 1B)
230 \ef formfeed (hex 0C)
231 \en linefeed (hex 0A)
232 \er carriage return (hex 0D)
233 \et tab (hex 09)
234 \eddd character with octal code ddd, or back reference
235 \exhh character with hex code hh
236 \ex{hhh..} character with hex code hhh..
237 .sp
238 The precise effect of \ecx is as follows: if x is a lower case letter, it
239 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
240 Thus \ecz becomes hex 1A, but \ec{ becomes hex 3B, while \ec; becomes hex
241 7B.
242 .P
243 After \ex, from zero to two hexadecimal digits are read (letters can be in
244 upper or lower case). Any number of hexadecimal digits may appear between \ex{
245 and }, but the value of the character code must be less than 256 in non-UTF-8
246 mode, and less than 2**31 in UTF-8 mode. That is, the maximum value in
247 hexadecimal is 7FFFFFFF. Note that this is bigger than the largest Unicode code
248 point, which is 10FFFF.
249 .P
250 If characters other than hexadecimal digits appear between \ex{ and }, or if
251 there is no terminating }, this form of escape is not recognized. Instead, the
252 initial \ex will be interpreted as a basic hexadecimal escape, with no
253 following digits, giving a character whose value is zero.
254 .P
255 Characters whose value is less than 256 can be defined by either of the two
256 syntaxes for \ex. There is no difference in the way they are handled. For
257 example, \exdc is exactly the same as \ex{dc}.
258 .P
259 After \e0 up to two further octal digits are read. If there are fewer than two
260 digits, just those that are present are used. Thus the sequence \e0\ex\e07
261 specifies two binary zeros followed by a BEL character (code value 7). Make
262 sure you supply two digits after the initial zero if the pattern character that
263 follows is itself an octal digit.
264 .P
265 The handling of a backslash followed by a digit other than 0 is complicated.
266 Outside a character class, PCRE reads it and any following digits as a decimal
267 number. If the number is less than 10, or if there have been at least that many
268 previous capturing left parentheses in the expression, the entire sequence is
269 taken as a \fIback reference\fP. A description of how this works is given
270 .\" HTML <a href="#backreferences">
271 .\" </a>
272 later,
273 .\"
274 following the discussion of
275 .\" HTML <a href="#subpattern">
276 .\" </a>
277 parenthesized subpatterns.
278 .\"
279 .P
280 Inside a character class, or if the decimal number is greater than 9 and there
281 have not been that many capturing subpatterns, PCRE re-reads up to three octal
282 digits following the backslash, and uses them to generate a data character. Any
283 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
284 character specified in octal must be less than \e400. In UTF-8 mode, values up
285 to \e777 are permitted. For example:
286 .sp
287 \e040 is another way of writing a space
288 .\" JOIN
289 \e40 is the same, provided there are fewer than 40
290 previous capturing subpatterns
291 \e7 is always a back reference
292 .\" JOIN
293 \e11 might be a back reference, or another way of
294 writing a tab
295 \e011 is always a tab
296 \e0113 is a tab followed by the character "3"
297 .\" JOIN
298 \e113 might be a back reference, otherwise the
299 character with octal code 113
300 .\" JOIN
301 \e377 might be a back reference, otherwise
302 the byte consisting entirely of 1 bits
303 .\" JOIN
304 \e81 is either a back reference, or a binary zero
305 followed by the two characters "8" and "1"
306 .sp
307 Note that octal values of 100 or greater must not be introduced by a leading
308 zero, because no more than three octal digits are ever read.
309 .P
310 All the sequences that define a single character value can be used both inside
311 and outside character classes. In addition, inside a character class, the
312 sequence \eb is interpreted as the backspace character (hex 08). The sequences
313 \eB, \eN, \eR, and \eX are not special inside a character class. Like any other
314 unrecognized escape sequences, they are treated as the literal characters "B",
315 "N", "R", and "X" by default, but cause an error if the PCRE_EXTRA option is
316 set. Outside a character class, these sequences have different meanings.
317 .
318 .
319 .SS "Absolute and relative back references"
320 .rs
321 .sp
322 The sequence \eg followed by an unsigned or a negative number, optionally
323 enclosed in braces, is an absolute or relative back reference. A named back
324 reference can be coded as \eg{name}. Back references are discussed
325 .\" HTML <a href="#backreferences">
326 .\" </a>
327 later,
328 .\"
329 following the discussion of
330 .\" HTML <a href="#subpattern">
331 .\" </a>
332 parenthesized subpatterns.
333 .\"
334 .
335 .
336 .SS "Absolute and relative subroutine calls"
337 .rs
338 .sp
339 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
340 a number enclosed either in angle brackets or single quotes, is an alternative
341 syntax for referencing a subpattern as a "subroutine". Details are discussed
342 .\" HTML <a href="#onigurumasubroutines">
343 .\" </a>
344 later.
345 .\"
346 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
347 synonymous. The former is a back reference; the latter is a
348 .\" HTML <a href="#subpatternsassubroutines">
349 .\" </a>
350 subroutine
351 .\"
352 call.
353 .
354 .
355 .\" HTML <a name="genericchartypes"></a>
356 .SS "Generic character types"
357 .rs
358 .sp
359 Another use of backslash is for specifying generic character types:
360 .sp
361 \ed any decimal digit
362 \eD any character that is not a decimal digit
363 \eh any horizontal whitespace character
364 \eH any character that is not a horizontal whitespace character
365 \es any whitespace character
366 \eS any character that is not a whitespace character
367 \ev any vertical whitespace character
368 \eV any character that is not a vertical whitespace character
369 \ew any "word" character
370 \eW any "non-word" character
371 .sp
372 There is also the single sequence \eN, which matches a non-newline character.
373 This is the same as
374 .\" HTML <a href="#fullstopdot">
375 .\" </a>
376 the "." metacharacter
377 .\"
378 when PCRE_DOTALL is not set.
379 .P
380 Each pair of lower and upper case escape sequences partitions the complete set
381 of characters into two disjoint sets. Any given character matches one, and only
382 one, of each pair. The sequences can appear both inside and outside character
383 classes. They each match one character of the appropriate type. If the current
384 matching point is at the end of the subject string, all of them fail, because
385 there is no character to match.
386 .P
387 For compatibility with Perl, \es does not match the VT character (code 11).
388 This makes it different from the the POSIX "space" class. The \es characters
389 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
390 included in a Perl script, \es may match the VT character. In PCRE, it never
391 does.
392 .P
393 A "word" character is an underscore or any character that is a letter or digit.
394 By default, the definition of letters and digits is controlled by PCRE's
395 low-valued character tables, and may vary if locale-specific matching is taking
396 place (see
397 .\" HTML <a href="pcreapi.html#localesupport">
398 .\" </a>
399 "Locale support"
400 .\"
401 in the
402 .\" HREF
403 \fBpcreapi\fP
404 .\"
405 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
406 or "french" in Windows, some character codes greater than 128 are used for
407 accented letters, and these are then matched by \ew. The use of locales with
408 Unicode is discouraged.
409 .P
410 By default, in UTF-8 mode, characters with values greater than 128 never match
411 \ed, \es, or \ew, and always match \eD, \eS, and \eW. These sequences retain
412 their original meanings from before UTF-8 support was available, mainly for
413 efficiency reasons. However, if PCRE is compiled with Unicode property support,
414 and the PCRE_UCP option is set, the behaviour is changed so that Unicode
415 properties are used to determine character types, as follows:
416 .sp
417 \ed any character that \ep{Nd} matches (decimal digit)
418 \es any character that \ep{Z} matches, plus HT, LF, FF, CR
419 \ew any character that \ep{L} or \ep{N} matches, plus underscore
420 .sp
421 The upper case escapes match the inverse sets of characters. Note that \ed
422 matches only decimal digits, whereas \ew matches any Unicode digit, as well as
423 any Unicode letter, and underscore. Note also that PCRE_UCP affects \eb, and
424 \eB because they are defined in terms of \ew and \eW. Matching these sequences
425 is noticeably slower when PCRE_UCP is set.
426 .P
427 The sequences \eh, \eH, \ev, and \eV are features that were added to Perl at
428 release 5.10. In contrast to the other sequences, which match only ASCII
429 characters by default, these always match certain high-valued codepoints in
430 UTF-8 mode, whether or not PCRE_UCP is set. The horizontal space characters
431 are:
432 .sp
433 U+0009 Horizontal tab
434 U+0020 Space
435 U+00A0 Non-break space
436 U+1680 Ogham space mark
437 U+180E Mongolian vowel separator
438 U+2000 En quad
439 U+2001 Em quad
440 U+2002 En space
441 U+2003 Em space
442 U+2004 Three-per-em space
443 U+2005 Four-per-em space
444 U+2006 Six-per-em space
445 U+2007 Figure space
446 U+2008 Punctuation space
447 U+2009 Thin space
448 U+200A Hair space
449 U+202F Narrow no-break space
450 U+205F Medium mathematical space
451 U+3000 Ideographic space
452 .sp
453 The vertical space characters are:
454 .sp
455 U+000A Linefeed
456 U+000B Vertical tab
457 U+000C Formfeed
458 U+000D Carriage return
459 U+0085 Next line
460 U+2028 Line separator
461 U+2029 Paragraph separator
462 .
463 .
464 .\" HTML <a name="newlineseq"></a>
465 .SS "Newline sequences"
466 .rs
467 .sp
468 Outside a character class, by default, the escape sequence \eR matches any
469 Unicode newline sequence. In non-UTF-8 mode \eR is equivalent to the following:
470 .sp
471 (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
472 .sp
473 This is an example of an "atomic group", details of which are given
474 .\" HTML <a href="#atomicgroup">
475 .\" </a>
476 below.
477 .\"
478 This particular group matches either the two-character sequence CR followed by
479 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
480 U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
481 line, U+0085). The two-character sequence is treated as a single unit that
482 cannot be split.
483 .P
484 In UTF-8 mode, two additional characters whose codepoints are greater than 255
485 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
486 Unicode character property support is not needed for these characters to be
487 recognized.
488 .P
489 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
490 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
491 either at compile time or when the pattern is matched. (BSR is an abbrevation
492 for "backslash R".) This can be made the default when PCRE is built; if this is
493 the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
494 It is also possible to specify these settings by starting a pattern string with
495 one of the following sequences:
496 .sp
497 (*BSR_ANYCRLF) CR, LF, or CRLF only
498 (*BSR_UNICODE) any Unicode newline sequence
499 .sp
500 These override the default and the options given to \fBpcre_compile()\fP or
501 \fBpcre_compile2()\fP, but they can be overridden by options given to
502 \fBpcre_exec()\fP or \fBpcre_dfa_exec()\fP. Note that these special settings,
503 which are not Perl-compatible, are recognized only at the very start of a
504 pattern, and that they must be in upper case. If more than one of them is
505 present, the last one is used. They can be combined with a change of newline
506 convention; for example, a pattern can start with:
507 .sp
509 .sp
510 They can also be combined with the (*UTF8) or (*UCP) special sequences. Inside
511 a character class, \eR is treated as an unrecognized escape sequence, and so
512 matches the letter "R" by default, but causes an error if PCRE_EXTRA is set.
513 .
514 .
515 .\" HTML <a name="uniextseq"></a>
516 .SS Unicode character properties
517 .rs
518 .sp
519 When PCRE is built with Unicode character property support, three additional
520 escape sequences that match characters with specific properties are available.
521 When not in UTF-8 mode, these sequences are of course limited to testing
522 characters whose codepoints are less than 256, but they do work in this mode.
523 The extra escape sequences are:
524 .sp
525 \ep{\fIxx\fP} a character with the \fIxx\fP property
526 \eP{\fIxx\fP} a character without the \fIxx\fP property
527 \eX an extended Unicode sequence
528 .sp
529 The property names represented by \fIxx\fP above are limited to the Unicode
530 script names, the general category properties, "Any", which matches any
531 character (including newline), and some special PCRE properties (described
532 in the
533 .\" HTML <a href="#extraprops">
534 .\" </a>
535 next section).
536 .\"
537 Other Perl properties such as "InMusicalSymbols" are not currently supported by
538 PCRE. Note that \eP{Any} does not match any characters, so always causes a
539 match failure.
540 .P
541 Sets of Unicode characters are defined as belonging to certain scripts. A
542 character from one of these sets can be matched using a script name. For
543 example:
544 .sp
545 \ep{Greek}
546 \eP{Han}
547 .sp
548 Those that are not part of an identified script are lumped together as
549 "Common". The current list of scripts is:
550 .P
551 Arabic,
552 Armenian,
553 Avestan,
554 Balinese,
555 Bamum,
556 Bengali,
557 Bopomofo,
558 Braille,
559 Buginese,
560 Buhid,
561 Canadian_Aboriginal,
562 Carian,
563 Cham,
564 Cherokee,
565 Common,
566 Coptic,
567 Cuneiform,
568 Cypriot,
569 Cyrillic,
570 Deseret,
571 Devanagari,
572 Egyptian_Hieroglyphs,
573 Ethiopic,
574 Georgian,
575 Glagolitic,
576 Gothic,
577 Greek,
578 Gujarati,
579 Gurmukhi,
580 Han,
581 Hangul,
582 Hanunoo,
583 Hebrew,
584 Hiragana,
585 Imperial_Aramaic,
586 Inherited,
587 Inscriptional_Pahlavi,
588 Inscriptional_Parthian,
589 Javanese,
590 Kaithi,
591 Kannada,
592 Katakana,
593 Kayah_Li,
594 Kharoshthi,
595 Khmer,
596 Lao,
597 Latin,
598 Lepcha,
599 Limbu,
600 Linear_B,
601 Lisu,
602 Lycian,
603 Lydian,
604 Malayalam,
605 Meetei_Mayek,
606 Mongolian,
607 Myanmar,
608 New_Tai_Lue,
609 Nko,
610 Ogham,
611 Old_Italic,
612 Old_Persian,
613 Old_South_Arabian,
614 Old_Turkic,
615 Ol_Chiki,
616 Oriya,
617 Osmanya,
618 Phags_Pa,
619 Phoenician,
620 Rejang,
621 Runic,
622 Samaritan,
623 Saurashtra,
624 Shavian,
625 Sinhala,
626 Sundanese,
627 Syloti_Nagri,
628 Syriac,
629 Tagalog,
630 Tagbanwa,
631 Tai_Le,
632 Tai_Tham,
633 Tai_Viet,
634 Tamil,
635 Telugu,
636 Thaana,
637 Thai,
638 Tibetan,
639 Tifinagh,
640 Ugaritic,
641 Vai,
642 Yi.
643 .P
644 Each character has exactly one Unicode general category property, specified by
645 a two-letter abbreviation. For compatibility with Perl, negation can be
646 specified by including a circumflex between the opening brace and the property
647 name. For example, \ep{^Lu} is the same as \eP{Lu}.
648 .P
649 If only one letter is specified with \ep or \eP, it includes all the general
650 category properties that start with that letter. In this case, in the absence
651 of negation, the curly brackets in the escape sequence are optional; these two
652 examples have the same effect:
653 .sp
654 \ep{L}
655 \epL
656 .sp
657 The following general category property codes are supported:
658 .sp
659 C Other
660 Cc Control
661 Cf Format
662 Cn Unassigned
663 Co Private use
664 Cs Surrogate
665 .sp
666 L Letter
667 Ll Lower case letter
668 Lm Modifier letter
669 Lo Other letter
670 Lt Title case letter
671 Lu Upper case letter
672 .sp
673 M Mark
674 Mc Spacing mark
675 Me Enclosing mark
676 Mn Non-spacing mark
677 .sp
678 N Number
679 Nd Decimal number
680 Nl Letter number
681 No Other number
682 .sp
683 P Punctuation
684 Pc Connector punctuation
685 Pd Dash punctuation
686 Pe Close punctuation
687 Pf Final punctuation
688 Pi Initial punctuation
689 Po Other punctuation
690 Ps Open punctuation
691 .sp
692 S Symbol
693 Sc Currency symbol
694 Sk Modifier symbol
695 Sm Mathematical symbol
696 So Other symbol
697 .sp
698 Z Separator
699 Zl Line separator
700 Zp Paragraph separator
701 Zs Space separator
702 .sp
703 The special property L& is also supported: it matches a character that has
704 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
705 a modifier or "other".
706 .P
707 The Cs (Surrogate) property applies only to characters in the range U+D800 to
708 U+DFFF. Such characters are not valid in UTF-8 strings (see RFC 3629) and so
709 cannot be tested by PCRE, unless UTF-8 validity checking has been turned off
710 (see the discussion of PCRE_NO_UTF8_CHECK in the
711 .\" HREF
712 \fBpcreapi\fP
713 .\"
714 page). Perl does not support the Cs property.
715 .P
716 The long synonyms for property names that Perl supports (such as \ep{Letter})
717 are not supported by PCRE, nor is it permitted to prefix any of these
718 properties with "Is".
719 .P
720 No character that is in the Unicode table has the Cn (unassigned) property.
721 Instead, this property is assumed for any code point that is not in the
722 Unicode table.
723 .P
724 Specifying caseless matching does not affect these escape sequences. For
725 example, \ep{Lu} always matches only upper case letters.
726 .P
727 The \eX escape matches any number of Unicode characters that form an extended
728 Unicode sequence. \eX is equivalent to
729 .sp
730 (?>\ePM\epM*)
731 .sp
732 That is, it matches a character without the "mark" property, followed by zero
733 or more characters with the "mark" property, and treats the sequence as an
734 atomic group
735 .\" HTML <a href="#atomicgroup">
736 .\" </a>
737 (see below).
738 .\"
739 Characters with the "mark" property are typically accents that affect the
740 preceding character. None of them have codepoints less than 256, so in
741 non-UTF-8 mode \eX matches any one character.
742 .P
743 Matching characters by Unicode property is not fast, because PCRE has to search
744 a structure that contains data for over fifteen thousand characters. That is
745 why the traditional escape sequences such as \ed and \ew do not use Unicode
746 properties in PCRE by default, though you can make them do so by setting the
747 PCRE_UCP option for \fBpcre_compile()\fP or by starting the pattern with
748 (*UCP).
749 .
750 .
751 .\" HTML <a name="extraprops"></a>
752 .SS PCRE's additional properties
753 .rs
754 .sp
755 As well as the standard Unicode properties described in the previous
756 section, PCRE supports four more that make it possible to convert traditional
757 escape sequences such as \ew and \es and POSIX character classes to use Unicode
758 properties. PCRE uses these non-standard, non-Perl properties internally when
759 PCRE_UCP is set. They are:
760 .sp
761 Xan Any alphanumeric character
762 Xps Any POSIX space character
763 Xsp Any Perl space character
764 Xwd Any Perl "word" character
765 .sp
766 Xan matches characters that have either the L (letter) or the N (number)
767 property. Xps matches the characters tab, linefeed, vertical tab, formfeed, or
768 carriage return, and any other character that has the Z (separator) property.
769 Xsp is the same as Xps, except that vertical tab is excluded. Xwd matches the
770 same characters as Xan, plus underscore.
771 .
772 .
773 .\" HTML <a name="resetmatchstart"></a>
774 .SS "Resetting the match start"
775 .rs
776 .sp
777 The escape sequence \eK causes any previously matched characters not to be
778 included in the final matched sequence. For example, the pattern:
779 .sp
780 foo\eKbar
781 .sp
782 matches "foobar", but reports that it has matched "bar". This feature is
783 similar to a lookbehind assertion
784 .\" HTML <a href="#lookbehind">
785 .\" </a>
786 (described below).
787 .\"
788 However, in this case, the part of the subject before the real match does not
789 have to be of fixed length, as lookbehind assertions do. The use of \eK does
790 not interfere with the setting of
791 .\" HTML <a href="#subpattern">
792 .\" </a>
793 captured substrings.
794 .\"
795 For example, when the pattern
796 .sp
797 (foo)\eKbar
798 .sp
799 matches "foobar", the first substring is still set to "foo".
800 .P
801 Perl documents that the use of \eK within assertions is "not well defined". In
802 PCRE, \eK is acted upon when it occurs inside positive assertions, but is
803 ignored in negative assertions.
804 .
805 .
806 .\" HTML <a name="smallassertions"></a>
807 .SS "Simple assertions"
808 .rs
809 .sp
810 The final use of backslash is for certain simple assertions. An assertion
811 specifies a condition that has to be met at a particular point in a match,
812 without consuming any characters from the subject string. The use of
813 subpatterns for more complicated assertions is described
814 .\" HTML <a href="#bigassertions">
815 .\" </a>
816 below.
817 .\"
818 The backslashed assertions are:
819 .sp
820 \eb matches at a word boundary
821 \eB matches when not at a word boundary
822 \eA matches at the start of the subject
823 \eZ matches at the end of the subject
824 also matches before a newline at the end of the subject
825 \ez matches only at the end of the subject
826 \eG matches at the first matching position in the subject
827 .sp
828 Inside a character class, \eb has a different meaning; it matches the backspace
829 character. If any other of these assertions appears in a character class, by
830 default it matches the corresponding literal character (for example, \eB
831 matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
832 escape sequence" error is generated instead.
833 .P
834 A word boundary is a position in the subject string where the current character
835 and the previous character do not both match \ew or \eW (i.e. one matches
836 \ew and the other matches \eW), or the start or end of the string if the
837 first or last character matches \ew, respectively. In UTF-8 mode, the meanings
838 of \ew and \eW can be changed by setting the PCRE_UCP option. When this is
839 done, it also affects \eb and \eB. Neither PCRE nor Perl has a separate "start
840 of word" or "end of word" metasequence. However, whatever follows \eb normally
841 determines which it is. For example, the fragment \eba matches "a" at the start
842 of a word.
843 .P
844 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
845 dollar (described in the next section) in that they only ever match at the very
846 start and end of the subject string, whatever options are set. Thus, they are
847 independent of multiline mode. These three assertions are not affected by the
848 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
849 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
850 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
851 at a point other than the beginning of the subject, \eA can never match. The
852 difference between \eZ and \ez is that \eZ matches before a newline at the end
853 of the string as well as at the very end, whereas \ez matches only at the end.
854 .P
855 The \eG assertion is true only when the current matching position is at the
856 start point of the match, as specified by the \fIstartoffset\fP argument of
857 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
858 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
859 arguments, you can mimic Perl's /g option, and it is in this kind of
860 implementation where \eG can be useful.
861 .P
862 Note, however, that PCRE's interpretation of \eG, as the start of the current
863 match, is subtly different from Perl's, which defines it as the end of the
864 previous match. In Perl, these can be different when the previously matched
865 string was empty. Because PCRE does just one match at a time, it cannot
866 reproduce this behaviour.
867 .P
868 If all the alternatives of a pattern begin with \eG, the expression is anchored
869 to the starting match position, and the "anchored" flag is set in the compiled
870 regular expression.
871 .
872 .
874 .rs
875 .sp
876 Outside a character class, in the default matching mode, the circumflex
877 character is an assertion that is true only if the current matching point is
878 at the start of the subject string. If the \fIstartoffset\fP argument of
879 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
880 option is unset. Inside a character class, circumflex has an entirely different
881 meaning
882 .\" HTML <a href="#characterclass">
883 .\" </a>
884 (see below).
885 .\"
886 .P
887 Circumflex need not be the first character of the pattern if a number of
888 alternatives are involved, but it should be the first thing in each alternative
889 in which it appears if the pattern is ever to match that branch. If all
890 possible alternatives start with a circumflex, that is, if the pattern is
891 constrained to match only at the start of the subject, it is said to be an
892 "anchored" pattern. (There are also other constructs that can cause a pattern
893 to be anchored.)
894 .P
895 A dollar character is an assertion that is true only if the current matching
896 point is at the end of the subject string, or immediately before a newline
897 at the end of the string (by default). Dollar need not be the last character of
898 the pattern if a number of alternatives are involved, but it should be the last
899 item in any branch in which it appears. Dollar has no special meaning in a
900 character class.
901 .P
902 The meaning of dollar can be changed so that it matches only at the very end of
903 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
904 does not affect the \eZ assertion.
905 .P
906 The meanings of the circumflex and dollar characters are changed if the
907 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
908 immediately after internal newlines as well as at the start of the subject
909 string. It does not match after a newline that ends the string. A dollar
910 matches before any newlines in the string, as well as at the very end, when
911 PCRE_MULTILINE is set. When newline is specified as the two-character
912 sequence CRLF, isolated CR and LF characters do not indicate newlines.
913 .P
914 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
915 \en represents a newline) in multiline mode, but not otherwise. Consequently,
916 patterns that are anchored in single line mode because all branches start with
917 ^ are not anchored in multiline mode, and a match for circumflex is possible
918 when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
919 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
920 .P
921 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
922 end of the subject in both modes, and if all branches of a pattern start with
923 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
924 .
925 .
926 .\" HTML <a name="fullstopdot"></a>
928 .rs
929 .sp
930 Outside a character class, a dot in the pattern matches any one character in
931 the subject string except (by default) a character that signifies the end of a
932 line. In UTF-8 mode, the matched character may be more than one byte long.
933 .P
934 When a line ending is defined as a single character, dot never matches that
935 character; when the two-character sequence CRLF is used, dot does not match CR
936 if it is immediately followed by LF, but otherwise it matches all characters
937 (including isolated CRs and LFs). When any Unicode line endings are being
938 recognized, dot does not match CR or LF or any of the other line ending
939 characters.
940 .P
941 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
942 option is set, a dot matches any one character, without exception. If the
943 two-character sequence CRLF is present in the subject string, it takes two dots
944 to match it.
945 .P
946 The handling of dot is entirely independent of the handling of circumflex and
947 dollar, the only relationship being that they both involve newlines. Dot has no
948 special meaning in a character class.
949 .P
950 The escape sequence \eN behaves like a dot, except that it is not affected by
951 the PCRE_DOTALL option. In other words, it matches any character except one
952 that signifies the end of a line.
953 .
954 .
956 .rs
957 .sp
958 Outside a character class, the escape sequence \eC matches any one byte, both
959 in and out of UTF-8 mode. Unlike a dot, it always matches any line-ending
960 characters. The feature is provided in Perl in order to match individual bytes
961 in UTF-8 mode. Because it breaks up UTF-8 characters into individual bytes, the
962 rest of the string may start with a malformed UTF-8 character. For this reason,
963 the \eC escape sequence is best avoided.
964 .P
965 PCRE does not allow \eC to appear in lookbehind assertions
966 .\" HTML <a href="#lookbehind">
967 .\" </a>
968 (described below),
969 .\"
970 because in UTF-8 mode this would make it impossible to calculate the length of
971 the lookbehind.
972 .
973 .
974 .\" HTML <a name="characterclass"></a>
976 .rs
977 .sp
978 An opening square bracket introduces a character class, terminated by a closing
979 square bracket. A closing square bracket on its own is not special by default.
980 However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
981 bracket causes a compile-time error. If a closing square bracket is required as
982 a member of the class, it should be the first data character in the class
983 (after an initial circumflex, if present) or escaped with a backslash.
984 .P
985 A character class matches a single character in the subject. In UTF-8 mode, the
986 character may be more than one byte long. A matched character must be in the
987 set of characters defined by the class, unless the first character in the class
988 definition is a circumflex, in which case the subject character must not be in
989 the set defined by the class. If a circumflex is actually required as a member
990 of the class, ensure it is not the first character, or escape it with a
991 backslash.
992 .P
993 For example, the character class [aeiou] matches any lower case vowel, while
994 [^aeiou] matches any character that is not a lower case vowel. Note that a
995 circumflex is just a convenient notation for specifying the characters that
996 are in the class by enumerating those that are not. A class that starts with a
997 circumflex is not an assertion; it still consumes a character from the subject
998 string, and therefore it fails if the current pointer is at the end of the
999 string.
1000 .P
1001 In UTF-8 mode, characters with values greater than 255 can be included in a
1002 class as a literal string of bytes, or by using the \ex{ escaping mechanism.
1003 .P
1004 When caseless matching is set, any letters in a class represent both their
1005 upper case and lower case versions, so for example, a caseless [aeiou] matches
1006 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1007 caseful version would. In UTF-8 mode, PCRE always understands the concept of
1008 case for characters whose values are less than 128, so caseless matching is
1009 always possible. For characters with higher values, the concept of case is
1010 supported if PCRE is compiled with Unicode property support, but not otherwise.
1011 If you want to use caseless matching in UTF8-mode for characters 128 and above,
1012 you must ensure that PCRE is compiled with Unicode property support as well as
1013 with UTF-8 support.
1014 .P
1015 Characters that might indicate line breaks are never treated in any special way
1016 when matching character classes, whatever line-ending sequence is in use, and
1017 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
1018 such as [^a] always matches one of these characters.
1019 .P
1020 The minus (hyphen) character can be used to specify a range of characters in a
1021 character class. For example, [d-m] matches any letter between d and m,
1022 inclusive. If a minus character is required in a class, it must be escaped with
1023 a backslash or appear in a position where it cannot be interpreted as
1024 indicating a range, typically as the first or last character in the class.
1025 .P
1026 It is not possible to have the literal character "]" as the end character of a
1027 range. A pattern such as [W-]46] is interpreted as a class of two characters
1028 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1029 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1030 the end of range, so [W-\e]46] is interpreted as a class containing a range
1031 followed by two other characters. The octal or hexadecimal representation of
1032 "]" can also be used to end a range.
1033 .P
1034 Ranges operate in the collating sequence of character values. They can also be
1035 used for characters specified numerically, for example [\e000-\e037]. In UTF-8
1036 mode, ranges can include characters whose values are greater than 255, for
1037 example [\ex{100}-\ex{2ff}].
1038 .P
1039 If a range that includes letters is used when caseless matching is set, it
1040 matches the letters in either case. For example, [W-c] is equivalent to
1041 [][\e\e^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
1042 tables for a French locale are in use, [\exc8-\excb] matches accented E
1043 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
1044 characters with values greater than 128 only when it is compiled with Unicode
1045 property support.
1046 .P
1047 The character types \ed, \eD, \eh, \eH, \ep, \eP, \es, \eS, \ev, \eV, \ew, and
1048 \eW may also appear in a character class, and add the characters that they
1049 match to the class. For example, [\edABCDEF] matches any hexadecimal digit. A
1050 circumflex can conveniently be used with the upper case character types to
1051 specify a more restricted set of characters than the matching lower case type.
1052 For example, the class [^\eW_] matches any letter or digit, but not underscore.
1053 .P
1054 The only metacharacters that are recognized in character classes are backslash,
1055 hyphen (only where it can be interpreted as specifying a range), circumflex
1056 (only at the start), opening square bracket (only when it can be interpreted as
1057 introducing a POSIX class name - see the next section), and the terminating
1058 closing square bracket. However, escaping other non-alphanumeric characters
1059 does no harm.
1060 .
1061 .
1063 .rs
1064 .sp
1065 Perl supports the POSIX notation for character classes. This uses names
1066 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1067 this notation. For example,
1068 .sp
1069 [01[:alpha:]%]
1070 .sp
1071 matches "0", "1", any alphabetic character, or "%". The supported class names
1072 are:
1073 .sp
1074 alnum letters and digits
1075 alpha letters
1076 ascii character codes 0 - 127
1077 blank space or tab only
1078 cntrl control characters
1079 digit decimal digits (same as \ed)
1080 graph printing characters, excluding space
1081 lower lower case letters
1082 print printing characters, including space
1083 punct printing characters, excluding letters and digits and space
1084 space white space (not quite the same as \es)
1085 upper upper case letters
1086 word "word" characters (same as \ew)
1087 xdigit hexadecimal digits
1088 .sp
1089 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1090 space (32). Notice that this list includes the VT character (code 11). This
1091 makes "space" different to \es, which does not include VT (for Perl
1092 compatibility).
1093 .P
1094 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1095 5.8. Another Perl extension is negation, which is indicated by a ^ character
1096 after the colon. For example,
1097 .sp
1098 [12[:^digit:]]
1099 .sp
1100 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1101 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1102 supported, and an error is given if they are encountered.
1103 .P
1104 By default, in UTF-8 mode, characters with values greater than 128 do not match
1105 any of the POSIX character classes. However, if the PCRE_UCP option is passed
1106 to \fBpcre_compile()\fP, some of the classes are changed so that Unicode
1107 character properties are used. This is achieved by replacing the POSIX classes
1108 by other sequences, as follows:
1109 .sp
1110 [:alnum:] becomes \ep{Xan}
1111 [:alpha:] becomes \ep{L}
1112 [:blank:] becomes \eh
1113 [:digit:] becomes \ep{Nd}
1114 [:lower:] becomes \ep{Ll}
1115 [:space:] becomes \ep{Xps}
1116 [:upper:] becomes \ep{Lu}
1117 [:word:] becomes \ep{Xwd}
1118 .sp
1119 Negated versions, such as [:^alpha:] use \eP instead of \ep. The other POSIX
1120 classes are unchanged, and match only characters with code points less than
1121 128.
1122 .
1123 .
1125 .rs
1126 .sp
1127 Vertical bar characters are used to separate alternative patterns. For example,
1128 the pattern
1129 .sp
1130 gilbert|sullivan
1131 .sp
1132 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1133 and an empty alternative is permitted (matching the empty string). The matching
1134 process tries each alternative in turn, from left to right, and the first one
1135 that succeeds is used. If the alternatives are within a subpattern
1136 .\" HTML <a href="#subpattern">
1137 .\" </a>
1138 (defined below),
1139 .\"
1140 "succeeds" means matching the rest of the main pattern as well as the
1141 alternative in the subpattern.
1142 .
1143 .
1145 .rs
1146 .sp
1147 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1148 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1149 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1150 The option letters are
1151 .sp
1152 i for PCRE_CASELESS
1154 s for PCRE_DOTALL
1155 x for PCRE_EXTENDED
1156 .sp
1157 For example, (?im) sets caseless, multiline matching. It is also possible to
1158 unset these options by preceding the letter with a hyphen, and a combined
1159 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1160 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1161 permitted. If a letter appears both before and after the hyphen, the option is
1162 unset.
1163 .P
1164 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1165 changed in the same way as the Perl-compatible options by using the characters
1166 J, U and X respectively.
1167 .P
1168 When one of these option changes occurs at top level (that is, not inside
1169 subpattern parentheses), the change applies to the remainder of the pattern
1170 that follows. If the change is placed right at the start of a pattern, PCRE
1171 extracts it into the global options (and it will therefore show up in data
1172 extracted by the \fBpcre_fullinfo()\fP function).
1173 .P
1174 An option change within a subpattern (see below for a description of
1175 subpatterns) affects only that part of the subpattern that follows it, so
1176 .sp
1177 (a(?i)b)c
1178 .sp
1179 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1180 By this means, options can be made to have different settings in different
1181 parts of the pattern. Any changes made in one alternative do carry on
1182 into subsequent branches within the same subpattern. For example,
1183 .sp
1184 (a(?i)b|c)
1185 .sp
1186 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1187 branch is abandoned before the option setting. This is because the effects of
1188 option settings happen at compile time. There would be some very weird
1189 behaviour otherwise.
1190 .P
1191 \fBNote:\fP There are other PCRE-specific options that can be set by the
1192 application when the compile or match functions are called. In some cases the
1193 pattern can contain special leading sequences such as (*CRLF) to override what
1194 the application has set or what has been defaulted. Details are given in the
1195 section entitled
1196 .\" HTML <a href="#newlineseq">
1197 .\" </a>
1198 "Newline sequences"
1199 .\"
1200 above. There are also the (*UTF8) and (*UCP) leading sequences that can be used
1201 to set UTF-8 and Unicode property modes; they are equivalent to setting the
1202 PCRE_UTF8 and the PCRE_UCP options, respectively.
1203 .
1204 .
1205 .\" HTML <a name="subpattern"></a>
1207 .rs
1208 .sp
1209 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1210 Turning part of a pattern into a subpattern does two things:
1211 .sp
1212 1. It localizes a set of alternatives. For example, the pattern
1213 .sp
1214 cat(aract|erpillar|)
1215 .sp
1216 matches "cataract", "caterpillar", or "cat". Without the parentheses, it would
1217 match "cataract", "erpillar" or an empty string.
1218 .sp
1219 2. It sets up the subpattern as a capturing subpattern. This means that, when
1220 the whole pattern matches, that portion of the subject string that matched the
1221 subpattern is passed back to the caller via the \fIovector\fP argument of
1222 \fBpcre_exec()\fP. Opening parentheses are counted from left to right (starting
1223 from 1) to obtain numbers for the capturing subpatterns. For example, if the
1224 string "the red king" is matched against the pattern
1225 .sp
1226 the ((red|white) (king|queen))
1227 .sp
1228 the captured substrings are "red king", "red", and "king", and are numbered 1,
1229 2, and 3, respectively.
1230 .P
1231 The fact that plain parentheses fulfil two functions is not always helpful.
1232 There are often times when a grouping subpattern is required without a
1233 capturing requirement. If an opening parenthesis is followed by a question mark
1234 and a colon, the subpattern does not do any capturing, and is not counted when
1235 computing the number of any subsequent capturing subpatterns. For example, if
1236 the string "the white queen" is matched against the pattern
1237 .sp
1238 the ((?:red|white) (king|queen))
1239 .sp
1240 the captured substrings are "white queen" and "queen", and are numbered 1 and
1241 2. The maximum number of capturing subpatterns is 65535.
1242 .P
1243 As a convenient shorthand, if any option settings are required at the start of
1244 a non-capturing subpattern, the option letters may appear between the "?" and
1245 the ":". Thus the two patterns
1246 .sp
1247 (?i:saturday|sunday)
1248 (?:(?i)saturday|sunday)
1249 .sp
1250 match exactly the same set of strings. Because alternative branches are tried
1251 from left to right, and options are not reset until the end of the subpattern
1252 is reached, an option setting in one branch does affect subsequent branches, so
1253 the above patterns match "SUNDAY" as well as "Saturday".
1254 .
1255 .
1256 .\" HTML <a name="dupsubpatternnumber"></a>
1258 .rs
1259 .sp
1260 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1261 the same numbers for its capturing parentheses. Such a subpattern starts with
1262 (?| and is itself a non-capturing subpattern. For example, consider this
1263 pattern:
1264 .sp
1265 (?|(Sat)ur|(Sun))day
1266 .sp
1267 Because the two alternatives are inside a (?| group, both sets of capturing
1268 parentheses are numbered one. Thus, when the pattern matches, you can look
1269 at captured substring number one, whichever alternative matched. This construct
1270 is useful when you want to capture part, but not all, of one of a number of
1271 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1272 number is reset at the start of each branch. The numbers of any capturing
1273 parentheses that follow the subpattern start after the highest number used in
1274 any branch. The following example is taken from the Perl documentation. The
1275 numbers underneath show in which buffer the captured content will be stored.
1276 .sp
1277 # before ---------------branch-reset----------- after
1278 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1279 # 1 2 2 3 2 3 4
1280 .sp
1281 A back reference to a numbered subpattern uses the most recent value that is
1282 set for that number by any subpattern. The following pattern matches "abcabc"
1283 or "defdef":
1284 .sp
1285 /(?|(abc)|(def))\e1/
1286 .sp
1287 In contrast, a recursive or "subroutine" call to a numbered subpattern always
1288 refers to the first one in the pattern with the given number. The following
1289 pattern matches "abcabc" or "defabc":
1290 .sp
1291 /(?|(abc)|(def))(?1)/
1292 .sp
1293 If a
1294 .\" HTML <a href="#conditions">
1295 .\" </a>
1296 condition test
1297 .\"
1298 for a subpattern's having matched refers to a non-unique number, the test is
1299 true if any of the subpatterns of that number have matched.
1300 .P
1301 An alternative approach to using this "branch reset" feature is to use
1302 duplicate named subpatterns, as described in the next section.
1303 .
1304 .
1306 .rs
1307 .sp
1308 Identifying capturing parentheses by number is simple, but it can be very hard
1309 to keep track of the numbers in complicated regular expressions. Furthermore,
1310 if an expression is modified, the numbers may change. To help with this
1311 difficulty, PCRE supports the naming of subpatterns. This feature was not
1312 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1313 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1314 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1315 have different names, but PCRE does not.
1316 .P
1317 In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1318 (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1319 parentheses from other parts of the pattern, such as
1320 .\" HTML <a href="#backreferences">
1321 .\" </a>
1322 back references,
1323 .\"
1324 .\" HTML <a href="#recursion">
1325 .\" </a>
1326 recursion,
1327 .\"
1328 and
1329 .\" HTML <a href="#conditions">
1330 .\" </a>
1331 conditions,
1332 .\"
1333 can be made by name as well as by number.
1334 .P
1335 Names consist of up to 32 alphanumeric characters and underscores. Named
1336 capturing parentheses are still allocated numbers as well as names, exactly as
1337 if the names were not present. The PCRE API provides function calls for
1338 extracting the name-to-number translation table from a compiled pattern. There
1339 is also a convenience function for extracting a captured substring by name.
1340 .P
1341 By default, a name must be unique within a pattern, but it is possible to relax
1342 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1343 names are also always permitted for subpatterns with the same number, set up as
1344 described in the previous section.) Duplicate names can be useful for patterns
1345 where only one instance of the named parentheses can match. Suppose you want to
1346 match the name of a weekday, either as a 3-letter abbreviation or as the full
1347 name, and in both cases you want to extract the abbreviation. This pattern
1348 (ignoring the line breaks) does the job:
1349 .sp
1350 (?<DN>Mon|Fri|Sun)(?:day)?|
1351 (?<DN>Tue)(?:sday)?|
1352 (?<DN>Wed)(?:nesday)?|
1353 (?<DN>Thu)(?:rsday)?|
1354 (?<DN>Sat)(?:urday)?
1355 .sp
1356 There are five capturing substrings, but only one is ever set after a match.
1357 (An alternative way of solving this problem is to use a "branch reset"
1358 subpattern, as described in the previous section.)
1359 .P
1360 The convenience function for extracting the data by name returns the substring
1361 for the first (and in this example, the only) subpattern of that name that
1362 matched. This saves searching to find which numbered subpattern it was.
1363 .P
1364 If you make a back reference to a non-unique named subpattern from elsewhere in
1365 the pattern, the one that corresponds to the first occurrence of the name is
1366 used. In the absence of duplicate numbers (see the previous section) this is
1367 the one with the lowest number. If you use a named reference in a condition
1368 test (see the
1369 .\"
1370 .\" HTML <a href="#conditions">
1371 .\" </a>
1372 section about conditions
1373 .\"
1374 below), either to check whether a subpattern has matched, or to check for
1375 recursion, all subpatterns with the same name are tested. If the condition is
1376 true for any one of them, the overall condition is true. This is the same
1377 behaviour as testing by number. For further details of the interfaces for
1378 handling named subpatterns, see the
1379 .\" HREF
1380 \fBpcreapi\fP
1381 .\"
1382 documentation.
1383 .P
1384 \fBWarning:\fP You cannot use different names to distinguish between two
1385 subpatterns with the same number because PCRE uses only the numbers when
1386 matching. For this reason, an error is given at compile time if different names
1387 are given to subpatterns with the same number. However, you can give the same
1388 name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
1389 .
1390 .
1392 .rs
1393 .sp
1394 Repetition is specified by quantifiers, which can follow any of the following
1395 items:
1396 .sp
1397 a literal data character
1398 the dot metacharacter
1399 the \eC escape sequence
1400 the \eX escape sequence (in UTF-8 mode with Unicode properties)
1401 the \eR escape sequence
1402 an escape such as \ed or \epL that matches a single character
1403 a character class
1404 a back reference (see next section)
1405 a parenthesized subpattern (unless it is an assertion)
1406 a recursive or "subroutine" call to a subpattern
1407 .sp
1408 The general repetition quantifier specifies a minimum and maximum number of
1409 permitted matches, by giving the two numbers in curly brackets (braces),
1410 separated by a comma. The numbers must be less than 65536, and the first must
1411 be less than or equal to the second. For example:
1412 .sp
1413 z{2,4}
1414 .sp
1415 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1416 character. If the second number is omitted, but the comma is present, there is
1417 no upper limit; if the second number and the comma are both omitted, the
1418 quantifier specifies an exact number of required matches. Thus
1419 .sp
1420 [aeiou]{3,}
1421 .sp
1422 matches at least 3 successive vowels, but may match many more, while
1423 .sp
1424 \ed{8}
1425 .sp
1426 matches exactly 8 digits. An opening curly bracket that appears in a position
1427 where a quantifier is not allowed, or one that does not match the syntax of a
1428 quantifier, is taken as a literal character. For example, {,6} is not a
1429 quantifier, but a literal string of four characters.
1430 .P
1431 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1432 bytes. Thus, for example, \ex{100}{2} matches two UTF-8 characters, each of
1433 which is represented by a two-byte sequence. Similarly, when Unicode property
1434 support is available, \eX{3} matches three Unicode extended sequences, each of
1435 which may be several bytes long (and they may be of different lengths).
1436 .P
1437 The quantifier {0} is permitted, causing the expression to behave as if the
1438 previous item and the quantifier were not present. This may be useful for
1439 subpatterns that are referenced as
1440 .\" HTML <a href="#subpatternsassubroutines">
1441 .\" </a>
1442 subroutines
1443 .\"
1444 from elsewhere in the pattern (but see also the section entitled
1445 .\" HTML <a href="#subdefine">
1446 .\" </a>
1447 "Defining subpatterns for use by reference only"
1448 .\"
1449 below). Items other than subpatterns that have a {0} quantifier are omitted
1450 from the compiled pattern.
1451 .P
1452 For convenience, the three most common quantifiers have single-character
1453 abbreviations:
1454 .sp
1455 * is equivalent to {0,}
1456 + is equivalent to {1,}
1457 ? is equivalent to {0,1}
1458 .sp
1459 It is possible to construct infinite loops by following a subpattern that can
1460 match no characters with a quantifier that has no upper limit, for example:
1461 .sp
1462 (a?)*
1463 .sp
1464 Earlier versions of Perl and PCRE used to give an error at compile time for
1465 such patterns. However, because there are cases where this can be useful, such
1466 patterns are now accepted, but if any repetition of the subpattern does in fact
1467 match no characters, the loop is forcibly broken.
1468 .P
1469 By default, the quantifiers are "greedy", that is, they match as much as
1470 possible (up to the maximum number of permitted times), without causing the
1471 rest of the pattern to fail. The classic example of where this gives problems
1472 is in trying to match comments in C programs. These appear between /* and */
1473 and within the comment, individual * and / characters may appear. An attempt to
1474 match C comments by applying the pattern
1475 .sp
1476 /\e*.*\e*/
1477 .sp
1478 to the string
1479 .sp
1480 /* first comment */ not comment /* second comment */
1481 .sp
1482 fails, because it matches the entire string owing to the greediness of the .*
1483 item.
1484 .P
1485 However, if a quantifier is followed by a question mark, it ceases to be
1486 greedy, and instead matches the minimum number of times possible, so the
1487 pattern
1488 .sp
1489 /\e*.*?\e*/
1490 .sp
1491 does the right thing with the C comments. The meaning of the various
1492 quantifiers is not otherwise changed, just the preferred number of matches.
1493 Do not confuse this use of question mark with its use as a quantifier in its
1494 own right. Because it has two uses, it can sometimes appear doubled, as in
1495 .sp
1496 \ed??\ed
1497 .sp
1498 which matches one digit by preference, but can match two if that is the only
1499 way the rest of the pattern matches.
1500 .P
1501 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1502 the quantifiers are not greedy by default, but individual ones can be made
1503 greedy by following them with a question mark. In other words, it inverts the
1504 default behaviour.
1505 .P
1506 When a parenthesized subpattern is quantified with a minimum repeat count that
1507 is greater than 1 or with a limited maximum, more memory is required for the
1508 compiled pattern, in proportion to the size of the minimum or maximum.
1509 .P
1510 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1511 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1512 implicitly anchored, because whatever follows will be tried against every
1513 character position in the subject string, so there is no point in retrying the
1514 overall match at any position after the first. PCRE normally treats such a
1515 pattern as though it were preceded by \eA.
1516 .P
1517 In cases where it is known that the subject string contains no newlines, it is
1518 worth setting PCRE_DOTALL in order to obtain this optimization, or
1519 alternatively using ^ to indicate anchoring explicitly.
1520 .P
1521 However, there is one situation where the optimization cannot be used. When .*
1522 is inside capturing parentheses that are the subject of a back reference
1523 elsewhere in the pattern, a match at the start may fail where a later one
1524 succeeds. Consider, for example:
1525 .sp
1526 (.*)abc\e1
1527 .sp
1528 If the subject is "xyz123abc123" the match point is the fourth character. For
1529 this reason, such a pattern is not implicitly anchored.
1530 .P
1531 When a capturing subpattern is repeated, the value captured is the substring
1532 that matched the final iteration. For example, after
1533 .sp
1534 (tweedle[dume]{3}\es*)+
1535 .sp
1536 has matched "tweedledum tweedledee" the value of the captured substring is
1537 "tweedledee". However, if there are nested capturing subpatterns, the
1538 corresponding captured values may have been set in previous iterations. For
1539 example, after
1540 .sp
1541 /(a|(b))+/
1542 .sp
1543 matches "aba" the value of the second captured substring is "b".
1544 .
1545 .
1546 .\" HTML <a name="atomicgroup"></a>
1548 .rs
1549 .sp
1550 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1551 repetition, failure of what follows normally causes the repeated item to be
1552 re-evaluated to see if a different number of repeats allows the rest of the
1553 pattern to match. Sometimes it is useful to prevent this, either to change the
1554 nature of the match, or to cause it fail earlier than it otherwise might, when
1555 the author of the pattern knows there is no point in carrying on.
1556 .P
1557 Consider, for example, the pattern \ed+foo when applied to the subject line
1558 .sp
1559 123456bar
1560 .sp
1561 After matching all 6 digits and then failing to match "foo", the normal
1562 action of the matcher is to try again with only 5 digits matching the \ed+
1563 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1564 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1565 that once a subpattern has matched, it is not to be re-evaluated in this way.
1566 .P
1567 If we use atomic grouping for the previous example, the matcher gives up
1568 immediately on failing to match "foo" the first time. The notation is a kind of
1569 special parenthesis, starting with (?> as in this example:
1570 .sp
1571 (?>\ed+)foo
1572 .sp
1573 This kind of parenthesis "locks up" the part of the pattern it contains once
1574 it has matched, and a failure further into the pattern is prevented from
1575 backtracking into it. Backtracking past it to previous items, however, works as
1576 normal.
1577 .P
1578 An alternative description is that a subpattern of this type matches the string
1579 of characters that an identical standalone pattern would match, if anchored at
1580 the current point in the subject string.
1581 .P
1582 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1583 the above example can be thought of as a maximizing repeat that must swallow
1584 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1585 number of digits they match in order to make the rest of the pattern match,
1586 (?>\ed+) can only match an entire sequence of digits.
1587 .P
1588 Atomic groups in general can of course contain arbitrarily complicated
1589 subpatterns, and can be nested. However, when the subpattern for an atomic
1590 group is just a single repeated item, as in the example above, a simpler
1591 notation, called a "possessive quantifier" can be used. This consists of an
1592 additional + character following a quantifier. Using this notation, the
1593 previous example can be rewritten as
1594 .sp
1595 \ed++foo
1596 .sp
1597 Note that a possessive quantifier can be used with an entire group, for
1598 example:
1599 .sp
1600 (abc|xyz){2,3}+
1601 .sp
1602 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1603 option is ignored. They are a convenient notation for the simpler forms of
1604 atomic group. However, there is no difference in the meaning of a possessive
1605 quantifier and the equivalent atomic group, though there may be a performance
1606 difference; possessive quantifiers should be slightly faster.
1607 .P
1608 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1609 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1610 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1611 package, and PCRE copied it from there. It ultimately found its way into Perl
1612 at release 5.10.
1613 .P
1614 PCRE has an optimization that automatically "possessifies" certain simple
1615 pattern constructs. For example, the sequence A+B is treated as A++B because
1616 there is no point in backtracking into a sequence of A's when B must follow.
1617 .P
1618 When a pattern contains an unlimited repeat inside a subpattern that can itself
1619 be repeated an unlimited number of times, the use of an atomic group is the
1620 only way to avoid some failing matches taking a very long time indeed. The
1621 pattern
1622 .sp
1623 (\eD+|<\ed+>)*[!?]
1624 .sp
1625 matches an unlimited number of substrings that either consist of non-digits, or
1626 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1627 quickly. However, if it is applied to
1628 .sp
1629 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1630 .sp
1631 it takes a long time before reporting failure. This is because the string can
1632 be divided between the internal \eD+ repeat and the external * repeat in a
1633 large number of ways, and all have to be tried. (The example uses [!?] rather
1634 than a single character at the end, because both PCRE and Perl have an
1635 optimization that allows for fast failure when a single character is used. They
1636 remember the last single character that is required for a match, and fail early
1637 if it is not present in the string.) If the pattern is changed so that it uses
1638 an atomic group, like this:
1639 .sp
1640 ((?>\eD+)|<\ed+>)*[!?]
1641 .sp
1642 sequences of non-digits cannot be broken, and failure happens quickly.
1643 .
1644 .
1645 .\" HTML <a name="backreferences"></a>
1647 .rs
1648 .sp
1649 Outside a character class, a backslash followed by a digit greater than 0 (and
1650 possibly further digits) is a back reference to a capturing subpattern earlier
1651 (that is, to its left) in the pattern, provided there have been that many
1652 previous capturing left parentheses.
1653 .P
1654 However, if the decimal number following the backslash is less than 10, it is
1655 always taken as a back reference, and causes an error only if there are not
1656 that many capturing left parentheses in the entire pattern. In other words, the
1657 parentheses that are referenced need not be to the left of the reference for
1658 numbers less than 10. A "forward back reference" of this type can make sense
1659 when a repetition is involved and the subpattern to the right has participated
1660 in an earlier iteration.
1661 .P
1662 It is not possible to have a numerical "forward back reference" to a subpattern
1663 whose number is 10 or more using this syntax because a sequence such as \e50 is
1664 interpreted as a character defined in octal. See the subsection entitled
1665 "Non-printing characters"
1666 .\" HTML <a href="#digitsafterbackslash">
1667 .\" </a>
1668 above
1669 .\"
1670 for further details of the handling of digits following a backslash. There is
1671 no such problem when named parentheses are used. A back reference to any
1672 subpattern is possible using named parentheses (see below).
1673 .P
1674 Another way of avoiding the ambiguity inherent in the use of digits following a
1675 backslash is to use the \eg escape sequence. This escape must be followed by an
1676 unsigned number or a negative number, optionally enclosed in braces. These
1677 examples are all identical:
1678 .sp
1679 (ring), \e1
1680 (ring), \eg1
1681 (ring), \eg{1}
1682 .sp
1683 An unsigned number specifies an absolute reference without the ambiguity that
1684 is present in the older syntax. It is also useful when literal digits follow
1685 the reference. A negative number is a relative reference. Consider this
1686 example:
1687 .sp
1688 (abc(def)ghi)\eg{-1}
1689 .sp
1690 The sequence \eg{-1} is a reference to the most recently started capturing
1691 subpattern before \eg, that is, is it equivalent to \e2 in this example.
1692 Similarly, \eg{-2} would be equivalent to \e1. The use of relative references
1693 can be helpful in long patterns, and also in patterns that are created by
1694 joining together fragments that contain references within themselves.
1695 .P
1696 A back reference matches whatever actually matched the capturing subpattern in
1697 the current subject string, rather than anything matching the subpattern
1698 itself (see
1699 .\" HTML <a href="#subpatternsassubroutines">
1700 .\" </a>
1701 "Subpatterns as subroutines"
1702 .\"
1703 below for a way of doing that). So the pattern
1704 .sp
1705 (sens|respons)e and \e1ibility
1706 .sp
1707 matches "sense and sensibility" and "response and responsibility", but not
1708 "sense and responsibility". If caseful matching is in force at the time of the
1709 back reference, the case of letters is relevant. For example,
1710 .sp
1711 ((?i)rah)\es+\e1
1712 .sp
1713 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1714 capturing subpattern is matched caselessly.
1715 .P
1716 There are several different ways of writing back references to named
1717 subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1718 \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1719 back reference syntax, in which \eg can be used for both numeric and named
1720 references, is also supported. We could rewrite the above example in any of
1721 the following ways:
1722 .sp
1723 (?<p1>(?i)rah)\es+\ek<p1>
1724 (?'p1'(?i)rah)\es+\ek{p1}
1725 (?P<p1>(?i)rah)\es+(?P=p1)
1726 (?<p1>(?i)rah)\es+\eg{p1}
1727 .sp
1728 A subpattern that is referenced by name may appear in the pattern before or
1729 after the reference.
1730 .P
1731 There may be more than one back reference to the same subpattern. If a
1732 subpattern has not actually been used in a particular match, any back
1733 references to it always fail by default. For example, the pattern
1734 .sp
1735 (a|(bc))\e2
1736 .sp
1737 always fails if it starts to match "a" rather than "bc". However, if the
1738 PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1739 unset value matches an empty string.
1740 .P
1741 Because there may be many capturing parentheses in a pattern, all digits
1742 following a backslash are taken as part of a potential back reference number.
1743 If the pattern continues with a digit character, some delimiter must be used to
1744 terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1745 whitespace. Otherwise, the \eg{ syntax or an empty comment (see
1746 .\" HTML <a href="#comments">
1747 .\" </a>
1748 "Comments"
1749 .\"
1750 below) can be used.
1751 .
1752 .SS "Recursive back references"
1753 .rs
1754 .sp
1755 A back reference that occurs inside the parentheses to which it refers fails
1756 when the subpattern is first used, so, for example, (a\e1) never matches.
1757 However, such references can be useful inside repeated subpatterns. For
1758 example, the pattern
1759 .sp
1760 (a|b\e1)+
1761 .sp
1762 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1763 the subpattern, the back reference matches the character string corresponding
1764 to the previous iteration. In order for this to work, the pattern must be such
1765 that the first iteration does not need to match the back reference. This can be
1766 done using alternation, as in the example above, or by a quantifier with a
1767 minimum of zero.
1768 .P
1769 Back references of this type cause the group that they reference to be treated
1770 as an
1771 .\" HTML <a href="#atomicgroup">
1772 .\" </a>
1773 atomic group.
1774 .\"
1775 Once the whole group has been matched, a subsequent matching failure cannot
1776 cause backtracking into the middle of the group.
1777 .
1778 .
1779 .\" HTML <a name="bigassertions"></a>
1781 .rs
1782 .sp
1783 An assertion is a test on the characters following or preceding the current
1784 matching point that does not actually consume any characters. The simple
1785 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1786 .\" HTML <a href="#smallassertions">
1787 .\" </a>
1788 above.
1789 .\"
1790 .P
1791 More complicated assertions are coded as subpatterns. There are two kinds:
1792 those that look ahead of the current position in the subject string, and those
1793 that look behind it. An assertion subpattern is matched in the normal way,
1794 except that it does not cause the current matching position to be changed.
1795 .P
1796 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1797 because it makes no sense to assert the same thing several times. If any kind
1798 of assertion contains capturing subpatterns within it, these are counted for
1799 the purposes of numbering the capturing subpatterns in the whole pattern.
1800 However, substring capturing is carried out only for positive assertions,
1801 because it does not make sense for negative assertions.
1802 .
1803 .
1804 .SS "Lookahead assertions"
1805 .rs
1806 .sp
1807 Lookahead assertions start with (?= for positive assertions and (?! for
1808 negative assertions. For example,
1809 .sp
1810 \ew+(?=;)
1811 .sp
1812 matches a word followed by a semicolon, but does not include the semicolon in
1813 the match, and
1814 .sp
1815 foo(?!bar)
1816 .sp
1817 matches any occurrence of "foo" that is not followed by "bar". Note that the
1818 apparently similar pattern
1819 .sp
1820 (?!foo)bar
1821 .sp
1822 does not find an occurrence of "bar" that is preceded by something other than
1823 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1824 (?!foo) is always true when the next three characters are "bar". A
1825 lookbehind assertion is needed to achieve the other effect.
1826 .P
1827 If you want to force a matching failure at some point in a pattern, the most
1828 convenient way to do it is with (?!) because an empty string always matches, so
1829 an assertion that requires there not to be an empty string must always fail.
1830 The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).
1831 .
1832 .
1833 .\" HTML <a name="lookbehind"></a>
1834 .SS "Lookbehind assertions"
1835 .rs
1836 .sp
1837 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1838 negative assertions. For example,
1839 .sp
1840 (?<!foo)bar
1841 .sp
1842 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1843 a lookbehind assertion are restricted such that all the strings it matches must
1844 have a fixed length. However, if there are several top-level alternatives, they
1845 do not all have to have the same fixed length. Thus
1846 .sp
1847 (?<=bullock|donkey)
1848 .sp
1849 is permitted, but
1850 .sp
1851 (?<!dogs?|cats?)
1852 .sp
1853 causes an error at compile time. Branches that match different length strings
1854 are permitted only at the top level of a lookbehind assertion. This is an
1855 extension compared with Perl, which requires all branches to match the same
1856 length of string. An assertion such as
1857 .sp
1858 (?<=ab(c|de))
1859 .sp
1860 is not permitted, because its single top-level branch can match two different
1861 lengths, but it is acceptable to PCRE if rewritten to use two top-level
1862 branches:
1863 .sp
1864 (?<=abc|abde)
1865 .sp
1866 In some cases, the escape sequence \eK
1867 .\" HTML <a href="#resetmatchstart">
1868 .\" </a>
1869 (see above)
1870 .\"
1871 can be used instead of a lookbehind assertion to get round the fixed-length
1872 restriction.
1873 .P
1874 The implementation of lookbehind assertions is, for each alternative, to
1875 temporarily move the current position back by the fixed length and then try to
1876 match. If there are insufficient characters before the current position, the
1877 assertion fails.
1878 .P
1879 PCRE does not allow the \eC escape (which matches a single byte in UTF-8 mode)
1880 to appear in lookbehind assertions, because it makes it impossible to calculate
1881 the length of the lookbehind. The \eX and \eR escapes, which can match
1882 different numbers of bytes, are also not permitted.
1883 .P
1884 .\" HTML <a href="#subpatternsassubroutines">
1885 .\" </a>
1886 "Subroutine"
1887 .\"
1888 calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
1889 as the subpattern matches a fixed-length string.
1890 .\" HTML <a href="#recursion">
1891 .\" </a>
1892 Recursion,
1893 .\"
1894 however, is not supported.
1895 .P
1896 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1897 specify efficient matching of fixed-length strings at the end of subject
1898 strings. Consider a simple pattern such as
1899 .sp
1900 abcd$
1901 .sp
1902 when applied to a long string that does not match. Because matching proceeds
1903 from left to right, PCRE will look for each "a" in the subject and then see if
1904 what follows matches the rest of the pattern. If the pattern is specified as
1905 .sp
1906 ^.*abcd$
1907 .sp
1908 the initial .* matches the entire string at first, but when this fails (because
1909 there is no following "a"), it backtracks to match all but the last character,
1910 then all but the last two characters, and so on. Once again the search for "a"
1911 covers the entire string, from right to left, so we are no better off. However,
1912 if the pattern is written as
1913 .sp
1914 ^.*+(?<=abcd)
1915 .sp
1916 there can be no backtracking for the .*+ item; it can match only the entire
1917 string. The subsequent lookbehind assertion does a single test on the last four
1918 characters. If it fails, the match fails immediately. For long strings, this
1919 approach makes a significant difference to the processing time.
1920 .
1921 .
1922 .SS "Using multiple assertions"
1923 .rs
1924 .sp
1925 Several assertions (of any sort) may occur in succession. For example,
1926 .sp
1927 (?<=\ed{3})(?<!999)foo
1928 .sp
1929 matches "foo" preceded by three digits that are not "999". Notice that each of
1930 the assertions is applied independently at the same point in the subject
1931 string. First there is a check that the previous three characters are all
1932 digits, and then there is a check that the same three characters are not "999".
1933 This pattern does \fInot\fP match "foo" preceded by six characters, the first
1934 of which are digits and the last three of which are not "999". For example, it
1935 doesn't match "123abcfoo". A pattern to do that is
1936 .sp
1937 (?<=\ed{3}...)(?<!999)foo
1938 .sp
1939 This time the first assertion looks at the preceding six characters, checking
1940 that the first three are digits, and then the second assertion checks that the
1941 preceding three characters are not "999".
1942 .P
1943 Assertions can be nested in any combination. For example,
1944 .sp
1945 (?<=(?<!foo)bar)baz
1946 .sp
1947 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1948 preceded by "foo", while
1949 .sp
1950 (?<=\ed{3}(?!999)...)foo
1951 .sp
1952 is another pattern that matches "foo" preceded by three digits and any three
1953 characters that are not "999".
1954 .
1955 .
1956 .\" HTML <a name="conditions"></a>
1958 .rs
1959 .sp
1960 It is possible to cause the matching process to obey a subpattern
1961 conditionally or to choose between two alternative subpatterns, depending on
1962 the result of an assertion, or whether a specific capturing subpattern has
1963 already been matched. The two possible forms of conditional subpattern are:
1964 .sp
1965 (?(condition)yes-pattern)
1966 (?(condition)yes-pattern|no-pattern)
1967 .sp
1968 If the condition is satisfied, the yes-pattern is used; otherwise the
1969 no-pattern (if present) is used. If there are more than two alternatives in the
1970 subpattern, a compile-time error occurs. Each of the two alternatives may
1971 itself contain nested subpatterns of any form, including conditional
1972 subpatterns; the restriction to two alternatives applies only at the level of
1973 the condition. This pattern fragment is an example where the alternatives are
1974 complex:
1975 .sp
1976 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
1977 .sp
1978 .P
1979 There are four kinds of condition: references to subpatterns, references to
1980 recursion, a pseudo-condition called DEFINE, and assertions.
1981 .
1982 .SS "Checking for a used subpattern by number"
1983 .rs
1984 .sp
1985 If the text between the parentheses consists of a sequence of digits, the
1986 condition is true if a capturing subpattern of that number has previously
1987 matched. If there is more than one capturing subpattern with the same number
1988 (see the earlier
1989 .\"
1990 .\" HTML <a href="#recursion">
1991 .\" </a>
1992 section about duplicate subpattern numbers),
1993 .\"
1994 the condition is true if any of them have matched. An alternative notation is
1995 to precede the digits with a plus or minus sign. In this case, the subpattern
1996 number is relative rather than absolute. The most recently opened parentheses
1997 can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside
1998 loops it can also make sense to refer to subsequent groups. The next
1999 parentheses to be opened can be referenced as (?(+1), and so on. (The value
2000 zero in any of these forms is not used; it provokes a compile-time error.)
2001 .P
2002 Consider the following pattern, which contains non-significant white space to
2003 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
2004 three parts for ease of discussion:
2005 .sp
2006 ( \e( )? [^()]+ (?(1) \e) )
2007 .sp
2008 The first part matches an optional opening parenthesis, and if that
2009 character is present, sets it as the first captured substring. The second part
2010 matches one or more characters that are not parentheses. The third part is a
2011 conditional subpattern that tests whether or not the first set of parentheses
2012 matched. If they did, that is, if subject started with an opening parenthesis,
2013 the condition is true, and so the yes-pattern is executed and a closing
2014 parenthesis is required. Otherwise, since no-pattern is not present, the
2015 subpattern matches nothing. In other words, this pattern matches a sequence of
2016 non-parentheses, optionally enclosed in parentheses.
2017 .P
2018 If you were embedding this pattern in a larger one, you could use a relative
2019 reference:
2020 .sp
2021 ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
2022 .sp
2023 This makes the fragment independent of the parentheses in the larger pattern.
2024 .
2025 .SS "Checking for a used subpattern by name"
2026 .rs
2027 .sp
2028 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
2029 subpattern by name. For compatibility with earlier versions of PCRE, which had
2030 this facility before Perl, the syntax (?(name)...) is also recognized. However,
2031 there is a possible ambiguity with this syntax, because subpattern names may
2032 consist entirely of digits. PCRE looks first for a named subpattern; if it
2033 cannot find one and the name consists entirely of digits, PCRE looks for a
2034 subpattern of that number, which must be greater than zero. Using subpattern
2035 names that consist entirely of digits is not recommended.
2036 .P
2037 Rewriting the above example to use a named subpattern gives this:
2038 .sp
2039 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
2040 .sp
2041 If the name used in a condition of this kind is a duplicate, the test is
2042 applied to all subpatterns of the same name, and is true if any one of them has
2043 matched.
2044 .
2045 .SS "Checking for pattern recursion"
2046 .rs
2047 .sp
2048 If the condition is the string (R), and there is no subpattern with the name R,
2049 the condition is true if a recursive call to the whole pattern or any
2050 subpattern has been made. If digits or a name preceded by ampersand follow the
2051 letter R, for example:
2052 .sp
2053 (?(R3)...) or (?(R&name)...)
2054 .sp
2055 the condition is true if the most recent recursion is into a subpattern whose
2056 number or name is given. This condition does not check the entire recursion
2057 stack. If the name used in a condition of this kind is a duplicate, the test is
2058 applied to all subpatterns of the same name, and is true if any one of them is
2059 the most recent recursion.
2060 .P
2061 At "top level", all these recursion test conditions are false.
2062 .\" HTML <a href="#recursion">
2063 .\" </a>
2064 The syntax for recursive patterns
2065 .\"
2066 is described below.
2067 .
2068 .\" HTML <a name="subdefine"></a>
2069 .SS "Defining subpatterns for use by reference only"
2070 .rs
2071 .sp
2072 If the condition is the string (DEFINE), and there is no subpattern with the
2073 name DEFINE, the condition is always false. In this case, there may be only one
2074 alternative in the subpattern. It is always skipped if control reaches this
2075 point in the pattern; the idea of DEFINE is that it can be used to define
2076 "subroutines" that can be referenced from elsewhere. (The use of
2077 .\" HTML <a href="#subpatternsassubroutines">
2078 .\" </a>
2079 "subroutines"
2080 .\"
2081 is described below.) For example, a pattern to match an IPv4 address such as
2082 "" could be written like this (ignore whitespace and line
2083 breaks):
2084 .sp
2085 (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
2086 \eb (?&byte) (\e.(?&byte)){3} \eb
2087 .sp
2088 The first part of the pattern is a DEFINE group inside which a another group
2089 named "byte" is defined. This matches an individual component of an IPv4
2090 address (a number less than 256). When matching takes place, this part of the
2091 pattern is skipped because DEFINE acts like a false condition. The rest of the
2092 pattern uses references to the named group to match the four dot-separated
2093 components of an IPv4 address, insisting on a word boundary at each end.
2094 .
2095 .SS "Assertion conditions"
2096 .rs
2097 .sp
2098 If the condition is not in any of the above formats, it must be an assertion.
2099 This may be a positive or negative lookahead or lookbehind assertion. Consider
2100 this pattern, again containing non-significant white space, and with the two
2101 alternatives on the second line:
2102 .sp
2103 (?(?=[^a-z]*[a-z])
2104 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
2105 .sp
2106 The condition is a positive lookahead assertion that matches an optional
2107 sequence of non-letters followed by a letter. In other words, it tests for the
2108 presence of at least one letter in the subject. If a letter is found, the
2109 subject is matched against the first alternative; otherwise it is matched
2110 against the second. This pattern matches strings in one of the two forms
2111 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2112 .
2113 .
2114 .\" HTML <a name="comments"></a>
2116 .rs
2117 .sp
2118 There are two ways of including comments in patterns that are processed by
2119 PCRE. In both cases, the start of the comment must not be in a character class,
2120 nor in the middle of any other sequence of related characters such as (?: or a
2121 subpattern name or number. The characters that make up a comment play no part
2122 in the pattern matching.
2123 .P
2124 The sequence (?# marks the start of a comment that continues up to the next
2125 closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED
2126 option is set, an unescaped # character also introduces a comment, which in
2127 this case continues to immediately after the next newline character or
2128 character sequence in the pattern. Which characters are interpreted as newlines
2129 is controlled by the options passed to \fBpcre_compile()\fP or by a special
2130 sequence at the start of the pattern, as described in the section entitled
2131 .\" HTML <a href="#newlines">
2132 .\" </a>
2133 "Newline conventions"
2134 .\"
2135 above. Note that the end of this type of comment is a literal newline sequence
2136 in the pattern; escape sequences that happen to represent a newline do not
2137 count. For example, consider this pattern when PCRE_EXTENDED is set, and the
2138 default newline convention is in force:
2139 .sp
2140 abc #comment \en still comment
2141 .sp
2142 On encountering the # character, \fBpcre_compile()\fP skips along, looking for
2143 a newline in the pattern. The sequence \en is still literal at this stage, so
2144 it does not terminate the comment. Only an actual character with the code value
2145 0x0a (the default newline) does so.
2146 .
2147 .
2148 .\" HTML <a name="recursion"></a>
2150 .rs
2151 .sp
2152 Consider the problem of matching a string in parentheses, allowing for
2153 unlimited nested parentheses. Without the use of recursion, the best that can
2154 be done is to use a pattern that matches up to some fixed depth of nesting. It
2155 is not possible to handle an arbitrary nesting depth.
2156 .P
2157 For some time, Perl has provided a facility that allows regular expressions to
2158 recurse (amongst other things). It does this by interpolating Perl code in the
2159 expression at run time, and the code can refer to the expression itself. A Perl
2160 pattern using code interpolation to solve the parentheses problem can be
2161 created like this:
2162 .sp
2163 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
2164 .sp
2165 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2166 recursively to the pattern in which it appears.
2167 .P
2168 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2169 supports special syntax for recursion of the entire pattern, and also for
2170 individual subpattern recursion. After its introduction in PCRE and Python,
2171 this kind of recursion was subsequently introduced into Perl at release 5.10.
2172 .P
2173 A special item that consists of (? followed by a number greater than zero and a
2174 closing parenthesis is a recursive call of the subpattern of the given number,
2175 provided that it occurs inside that subpattern. (If not, it is a
2176 .\" HTML <a href="#subpatternsassubroutines">
2177 .\" </a>
2178 "subroutine"
2179 .\"
2180 call, which is described in the next section.) The special item (?R) or (?0) is
2181 a recursive call of the entire regular expression.
2182 .P
2183 This PCRE pattern solves the nested parentheses problem (assume the
2184 PCRE_EXTENDED option is set so that white space is ignored):
2185 .sp
2186 \e( ( [^()]++ | (?R) )* \e)
2187 .sp
2188 First it matches an opening parenthesis. Then it matches any number of
2189 substrings which can either be a sequence of non-parentheses, or a recursive
2190 match of the pattern itself (that is, a correctly parenthesized substring).
2191 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2192 to avoid backtracking into sequences of non-parentheses.
2193 .P
2194 If this were part of a larger pattern, you would not want to recurse the entire
2195 pattern, so instead you could use this:
2196 .sp
2197 ( \e( ( [^()]++ | (?1) )* \e) )
2198 .sp
2199 We have put the pattern into parentheses, and caused the recursion to refer to
2200 them instead of the whole pattern.
2201 .P
2202 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2203 is made easier by the use of relative references. Instead of (?1) in the
2204 pattern above you can write (?-2) to refer to the second most recently opened
2205 parentheses preceding the recursion. In other words, a negative number counts
2206 capturing parentheses leftwards from the point at which it is encountered.
2207 .P
2208 It is also possible to refer to subsequently opened parentheses, by writing
2209 references such as (?+2). However, these cannot be recursive because the
2210 reference is not inside the parentheses that are referenced. They are always
2211 .\" HTML <a href="#subpatternsassubroutines">
2212 .\" </a>
2213 "subroutine"
2214 .\"
2215 calls, as described in the next section.
2216 .P
2217 An alternative approach is to use named parentheses instead. The Perl syntax
2218 for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
2219 could rewrite the above example as follows:
2220 .sp
2221 (?<pn> \e( ( [^()]++ | (?&pn) )* \e) )
2222 .sp
2223 If there is more than one subpattern with the same name, the earliest one is
2224 used.
2225 .P
2226 This particular example pattern that we have been looking at contains nested
2227 unlimited repeats, and so the use of a possessive quantifier for matching
2228 strings of non-parentheses is important when applying the pattern to strings
2229 that do not match. For example, when this pattern is applied to
2230 .sp
2231 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2232 .sp
2233 it yields "no match" quickly. However, if a possessive quantifier is not used,
2234 the match runs for a very long time indeed because there are so many different
2235 ways the + and * repeats can carve up the subject, and all have to be tested
2236 before failure can be reported.
2237 .P
2238 At the end of a match, the values of capturing parentheses are those from
2239 the outermost level. If you want to obtain intermediate values, a callout
2240 function can be used (see below and the
2241 .\" HREF
2242 \fBpcrecallout\fP
2243 .\"
2244 documentation). If the pattern above is matched against
2245 .sp
2246 (ab(cd)ef)
2247 .sp
2248 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2249 the last value taken on at the top level. If a capturing subpattern is not
2250 matched at the top level, its final value is unset, even if it is (temporarily)
2251 set at a deeper level.
2252 .P
2253 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2254 obtain extra memory to store data during a recursion, which it does by using
2255 \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no memory can
2256 be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2257 .P
2258 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2259 Consider this pattern, which matches text in angle brackets, allowing for
2260 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2261 recursing), whereas any characters are permitted at the outer level.
2262 .sp
2263 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
2264 .sp
2265 In this pattern, (?(R) is the start of a conditional subpattern, with two
2266 different alternatives for the recursive and non-recursive cases. The (?R) item
2267 is the actual recursive call.
2268 .
2269 .
2270 .\" HTML <a name="recursiondifference"></a>
2271 .SS "Recursion difference from Perl"
2272 .rs
2273 .sp
2274 In PCRE (like Python, but unlike Perl), a recursive subpattern call is always
2275 treated as an atomic group. That is, once it has matched some of the subject
2276 string, it is never re-entered, even if it contains untried alternatives and
2277 there is a subsequent matching failure. This can be illustrated by the
2278 following pattern, which purports to match a palindromic string that contains
2279 an odd number of characters (for example, "a", "aba", "abcba", "abcdcba"):
2280 .sp
2281 ^(.|(.)(?1)\e2)$
2282 .sp
2283 The idea is that it either matches a single character, or two identical
2284 characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2285 it does not if the pattern is longer than three characters. Consider the
2286 subject string "abcba":
2287 .P
2288 At the top level, the first character is matched, but as it is not at the end
2289 of the string, the first alternative fails; the second alternative is taken
2290 and the recursion kicks in. The recursive call to subpattern 1 successfully
2291 matches the next character ("b"). (Note that the beginning and end of line
2292 tests are not part of the recursion).
2293 .P
2294 Back at the top level, the next character ("c") is compared with what
2295 subpattern 2 matched, which was "a". This fails. Because the recursion is
2296 treated as an atomic group, there are now no backtracking points, and so the
2297 entire match fails. (Perl is able, at this point, to re-enter the recursion and
2298 try the second alternative.) However, if the pattern is written with the
2299 alternatives in the other order, things are different:
2300 .sp
2301 ^((.)(?1)\e2|.)$
2302 .sp
2303 This time, the recursing alternative is tried first, and continues to recurse
2304 until it runs out of characters, at which point the recursion fails. But this
2305 time we do have another alternative to try at the higher level. That is the big
2306 difference: in the previous case the remaining alternative is at a deeper
2307 recursion level, which PCRE cannot use.
2308 .P
2309 To change the pattern so that it matches all palindromic strings, not just
2310 those with an odd number of characters, it is tempting to change the pattern to
2311 this:
2312 .sp
2313 ^((.)(?1)\e2|.?)$
2314 .sp
2315 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2316 deeper recursion has matched a single character, it cannot be entered again in
2317 order to match an empty string. The solution is to separate the two cases, and
2318 write out the odd and even cases as alternatives at the higher level:
2319 .sp
2320 ^(?:((.)(?1)\e2|)|((.)(?3)\e4|.))
2321 .sp
2322 If you want to match typical palindromic phrases, the pattern has to ignore all
2323 non-word characters, which can be done like this:
2324 .sp
2325 ^\eW*+(?:((.)\eW*+(?1)\eW*+\e2|)|((.)\eW*+(?3)\eW*+\e4|\eW*+.\eW*+))\eW*+$
2326 .sp
2327 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2328 man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2329 the use of the possessive quantifier *+ to avoid backtracking into sequences of
2330 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2331 more) to match typical phrases, and Perl takes so long that you think it has
2332 gone into a loop.
2333 .P
2334 \fBWARNING\fP: The palindrome-matching patterns above work only if the subject
2335 string does not start with a palindrome that is shorter than the entire string.
2336 For example, although "abcba" is correctly matched, if the subject is "ababa",
2337 PCRE finds the palindrome "aba" at the start, then fails at top level because
2338 the end of the string does not follow. Once again, it cannot jump back into the
2339 recursion to try other alternatives, so the entire match fails.
2340 .
2341 .
2342 .\" HTML <a name="subpatternsassubroutines"></a>
2344 .rs
2345 .sp
2346 If the syntax for a recursive subpattern reference (either by number or by
2347 name) is used outside the parentheses to which it refers, it operates like a
2348 subroutine in a programming language. The "called" subpattern may be defined
2349 before or after the reference. A numbered reference can be absolute or
2350 relative, as in these examples:
2351 .sp
2352 (...(absolute)...)...(?2)...
2353 (...(relative)...)...(?-1)...
2354 (...(?+1)...(relative)...
2355 .sp
2356 An earlier example pointed out that the pattern
2357 .sp
2358 (sens|respons)e and \e1ibility
2359 .sp
2360 matches "sense and sensibility" and "response and responsibility", but not
2361 "sense and responsibility". If instead the pattern
2362 .sp
2363 (sens|respons)e and (?1)ibility
2364 .sp
2365 is used, it does match "sense and responsibility" as well as the other two
2366 strings. Another example is given in the discussion of DEFINE above.
2367 .P
2368 Like recursive subpatterns, a subroutine call is always treated as an atomic
2369 group. That is, once it has matched some of the subject string, it is never
2370 re-entered, even if it contains untried alternatives and there is a subsequent
2371 matching failure. Any capturing parentheses that are set during the subroutine
2372 call revert to their previous values afterwards.
2373 .P
2374 When a subpattern is used as a subroutine, processing options such as
2375 case-independence are fixed when the subpattern is defined. They cannot be
2376 changed for different calls. For example, consider this pattern:
2377 .sp
2378 (abc)(?i:(?-1))
2379 .sp
2380 It matches "abcabc". It does not match "abcABC" because the change of
2381 processing option does not affect the called subpattern.
2382 .
2383 .
2384 .\" HTML <a name="onigurumasubroutines"></a>
2386 .rs
2387 .sp
2388 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
2389 a number enclosed either in angle brackets or single quotes, is an alternative
2390 syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2391 are two of the examples used above, rewritten using this syntax:
2392 .sp
2393 (?<pn> \e( ( (?>[^()]+) | \eg<pn> )* \e) )
2394 (sens|respons)e and \eg'1'ibility
2395 .sp
2396 PCRE supports an extension to Oniguruma: if a number is preceded by a
2397 plus or a minus sign it is taken as a relative reference. For example:
2398 .sp
2399 (abc)(?i:\eg<-1>)
2400 .sp
2401 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
2402 synonymous. The former is a back reference; the latter is a subroutine call.
2403 .
2404 .
2406 .rs
2407 .sp
2408 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2409 code to be obeyed in the middle of matching a regular expression. This makes it
2410 possible, amongst other things, to extract different substrings that match the
2411 same pair of parentheses when there is a repetition.
2412 .P
2413 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2414 code. The feature is called "callout". The caller of PCRE provides an external
2415 function by putting its entry point in the global variable \fIpcre_callout\fP.
2416 By default, this variable contains NULL, which disables all calling out.
2417 .P
2418 Within a regular expression, (?C) indicates the points at which the external
2419 function is to be called. If you want to identify different callout points, you
2420 can put a number less than 256 after the letter C. The default value is zero.
2421 For example, this pattern has two callout points:
2422 .sp
2423 (?C1)abc(?C2)def
2424 .sp
2425 If the PCRE_AUTO_CALLOUT flag is passed to \fBpcre_compile()\fP, callouts are
2426 automatically installed before each item in the pattern. They are all numbered
2427 255.
2428 .P
2429 During matching, when PCRE reaches a callout point (and \fIpcre_callout\fP is
2430 set), the external function is called. It is provided with the number of the
2431 callout, the position in the pattern, and, optionally, one item of data
2432 originally supplied by the caller of \fBpcre_exec()\fP. The callout function
2433 may cause matching to proceed, to backtrack, or to fail altogether. A complete
2434 description of the interface to the callout function is given in the
2435 .\" HREF
2436 \fBpcrecallout\fP
2437 .\"
2438 documentation.
2439 .
2440 .
2441 .\" HTML <a name="backtrackcontrol"></a>
2443 .rs
2444 .sp
2445 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2446 are described in the Perl documentation as "experimental and subject to change
2447 or removal in a future version of Perl". It goes on to say: "Their usage in
2448 production code should be noted to avoid problems during upgrades." The same
2449 remarks apply to the PCRE features described in this section.
2450 .P
2451 Since these verbs are specifically related to backtracking, most of them can be
2452 used only when the pattern is to be matched using \fBpcre_exec()\fP, which uses
2453 a backtracking algorithm. With the exception of (*FAIL), which behaves like a
2454 failing negative assertion, they cause an error if encountered by
2455 \fBpcre_dfa_exec()\fP.
2456 .P
2457 If any of these verbs are used in an assertion or subroutine subpattern
2458 (including recursive subpatterns), their effect is confined to that subpattern;
2459 it does not extend to the surrounding pattern. Note that such subpatterns are
2460 processed as anchored at the point where they are tested.
2461 .P
2462 The new verbs make use of what was previously invalid syntax: an opening
2463 parenthesis followed by an asterisk. They are generally of the form
2464 (*VERB) or (*VERB:NAME). Some may take either form, with differing behaviour,
2465 depending on whether or not an argument is present. An name is a sequence of
2466 letters, digits, and underscores. If the name is empty, that is, if the closing
2467 parenthesis immediately follows the colon, the effect is as if the colon were
2468 not there. Any number of these verbs may occur in a pattern.
2469 .P
2470 PCRE contains some optimizations that are used to speed up matching by running
2471 some checks at the start of each match attempt. For example, it may know the
2472 minimum length of matching subject, or that a particular character must be
2473 present. When one of these optimizations suppresses the running of a match, any
2474 included backtracking verbs will not, of course, be processed. You can suppress
2475 the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2476 when calling \fBpcre_exec()\fP.
2477 .
2478 .
2479 .SS "Verbs that act immediately"
2480 .rs
2481 .sp
2482 The following verbs act as soon as they are encountered. They may not be
2483 followed by a name.
2484 .sp
2485 (*ACCEPT)
2486 .sp
2487 This verb causes the match to end successfully, skipping the remainder of the
2488 pattern. When inside a recursion, only the innermost pattern is ended
2489 immediately. If (*ACCEPT) is inside capturing parentheses, the data so far is
2490 captured. (This feature was added to PCRE at release 8.00.) For example:
2491 .sp
2492 A((?:A|B(*ACCEPT)|C)D)
2493 .sp
2494 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2495 the outer parentheses.
2496 .sp
2497 (*FAIL) or (*F)
2498 .sp
2499 This verb causes the match to fail, forcing backtracking to occur. It is
2500 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2501 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2502 Perl features that are not present in PCRE. The nearest equivalent is the
2503 callout feature, as for example in this pattern:
2504 .sp
2505 a+(?C)(*FAIL)
2506 .sp
2507 A match with the string "aaaa" always fails, but the callout is taken before
2508 each backtrack happens (in this example, 10 times).
2509 .
2510 .
2511 .SS "Recording which path was taken"
2512 .rs
2513 .sp
2514 There is one verb whose main purpose is to track how a match was arrived at,
2515 though it also has a secondary use in conjunction with advancing the match
2516 starting point (see (*SKIP) below).
2517 .sp
2518 (*MARK:NAME) or (*:NAME)
2519 .sp
2520 A name is always required with this verb. There may be as many instances of
2521 (*MARK) as you like in a pattern, and their names do not have to be unique.
2522 .P
2523 When a match succeeds, the name of the last-encountered (*MARK) is passed back
2524 to the caller via the \fIpcre_extra\fP data structure, as described in the
2525 .\" HTML <a href="pcreapi.html#extradata">
2526 .\" </a>
2527 section on \fIpcre_extra\fP
2528 .\"
2529 in the
2530 .\" HREF
2531 \fBpcreapi\fP
2532 .\"
2533 documentation. No data is returned for a partial match. Here is an example of
2534 \fBpcretest\fP output, where the /K modifier requests the retrieval and
2535 outputting of (*MARK) data:
2536 .sp
2537 /X(*MARK:A)Y|X(*MARK:B)Z/K
2538 XY
2539 0: XY
2540 MK: A
2541 XZ
2542 0: XZ
2543 MK: B
2544 .sp
2545 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2546 indicates which of the two alternatives matched. This is a more efficient way
2547 of obtaining this information than putting each alternative in its own
2548 capturing parentheses.
2549 .P
2550 A name may also be returned after a failed match if the final path through the
2551 pattern involves (*MARK). However, unless (*MARK) used in conjunction with
2552 (*COMMIT), this is unlikely to happen for an unanchored pattern because, as the
2553 starting point for matching is advanced, the final check is often with an empty
2554 string, causing a failure before (*MARK) is reached. For example:
2555 .sp
2556 /X(*MARK:A)Y|X(*MARK:B)Z/K
2557 XP
2558 No match
2559 .sp
2560 There are three potential starting points for this match (starting with X,
2561 starting with P, and with an empty string). If the pattern is anchored, the
2562 result is different:
2563 .sp
2564 /^X(*MARK:A)Y|^X(*MARK:B)Z/K
2565 XP
2566 No match, mark = B
2567 .sp
2568 PCRE's start-of-match optimizations can also interfere with this. For example,
2569 if, as a result of a call to \fBpcre_study()\fP, it knows the minimum
2570 subject length for a match, a shorter subject will not be scanned at all.
2571 .P
2572 Note that similar anomalies (though different in detail) exist in Perl, no
2573 doubt for the same reasons. The use of (*MARK) data after a failed match of an
2574 unanchored pattern is not recommended, unless (*COMMIT) is involved.
2575 .
2576 .
2577 .SS "Verbs that act after backtracking"
2578 .rs
2579 .sp
2580 The following verbs do nothing when they are encountered. Matching continues
2581 with what follows, but if there is no subsequent match, causing a backtrack to
2582 the verb, a failure is forced. That is, backtracking cannot pass to the left of
2583 the verb. However, when one of these verbs appears inside an atomic group, its
2584 effect is confined to that group, because once the group has been matched,
2585 there is never any backtracking into it. In this situation, backtracking can
2586 "jump back" to the left of the entire atomic group. (Remember also, as stated
2587 above, that this localization also applies in subroutine calls and assertions.)
2588 .P
2589 These verbs differ in exactly what kind of failure occurs when backtracking
2590 reaches them.
2591 .sp
2592 (*COMMIT)
2593 .sp
2594 This verb, which may not be followed by a name, causes the whole match to fail
2595 outright if the rest of the pattern does not match. Even if the pattern is
2596 unanchored, no further attempts to find a match by advancing the starting point
2597 take place. Once (*COMMIT) has been passed, \fBpcre_exec()\fP is committed to
2598 finding a match at the current starting point, or not at all. For example:
2599 .sp
2600 a+(*COMMIT)b
2601 .sp
2602 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2603 dynamic anchor, or "I've started, so I must finish." The name of the most
2604 recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2605 match failure.
2606 .P
2607 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2608 unless PCRE's start-of-match optimizations are turned off, as shown in this
2609 \fBpcretest\fP example:
2610 .sp
2611 /(*COMMIT)abc/
2612 xyzabc
2613 0: abc
2614 xyzabc\eY
2615 No match
2616 .sp
2617 PCRE knows that any match must start with "a", so the optimization skips along
2618 the subject to "a" before running the first match attempt, which succeeds. When
2619 the optimization is disabled by the \eY escape in the second subject, the match
2620 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2621 starting points.
2622 .sp
2623 (*PRUNE) or (*PRUNE:NAME)
2624 .sp
2625 This verb causes the match to fail at the current starting position in the
2626 subject if the rest of the pattern does not match. If the pattern is
2627 unanchored, the normal "bumpalong" advance to the next starting character then
2628 happens. Backtracking can occur as usual to the left of (*PRUNE), before it is
2629 reached, or when matching to the right of (*PRUNE), but if there is no match to
2630 the right, backtracking cannot cross (*PRUNE). In simple cases, the use of
2631 (*PRUNE) is just an alternative to an atomic group or possessive quantifier,
2632 but there are some uses of (*PRUNE) that cannot be expressed in any other way.
2633 The behaviour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE) when the
2634 match fails completely; the name is passed back if this is the final attempt.
2635 (*PRUNE:NAME) does not pass back a name if the match succeeds. In an anchored
2636 pattern (*PRUNE) has the same effect as (*COMMIT).
2637 .sp
2638 (*SKIP)
2639 .sp
2640 This verb, when given without a name, is like (*PRUNE), except that if the
2641 pattern is unanchored, the "bumpalong" advance is not to the next character,
2642 but to the position in the subject where (*SKIP) was encountered. (*SKIP)
2643 signifies that whatever text was matched leading up to it cannot be part of a
2644 successful match. Consider:
2645 .sp
2646 a+(*SKIP)b
2647 .sp
2648 If the subject is "aaaac...", after the first match attempt fails (starting at
2649 the first character in the string), the starting point skips on to start the
2650 next attempt at "c". Note that a possessive quantifer does not have the same
2651 effect as this example; although it would suppress backtracking during the
2652 first match attempt, the second attempt would start at the second character
2653 instead of skipping on to "c".
2654 .sp
2655 (*SKIP:NAME)
2656 .sp
2657 When (*SKIP) has an associated name, its behaviour is modified. If the
2658 following pattern fails to match, the previous path through the pattern is
2659 searched for the most recent (*MARK) that has the same name. If one is found,
2660 the "bumpalong" advance is to the subject position that corresponds to that
2661 (*MARK) instead of to where (*SKIP) was encountered. If no (*MARK) with a
2662 matching name is found, normal "bumpalong" of one character happens (the
2663 (*SKIP) is ignored).
2664 .sp
2665 (*THEN) or (*THEN:NAME)
2666 .sp
2667 This verb causes a skip to the next alternation in the innermost enclosing
2668 group if the rest of the pattern does not match. That is, it cancels pending
2669 backtracking, but only within the current alternation. Its name comes from the
2670 observation that it can be used for a pattern-based if-then-else block:
2671 .sp
2672 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2673 .sp
2674 If the COND1 pattern matches, FOO is tried (and possibly further items after
2675 the end of the group if FOO succeeds); on failure the matcher skips to the
2676 second alternative and tries COND2, without backtracking into COND1. The
2677 behaviour of (*THEN:NAME) is exactly the same as (*MARK:NAME)(*THEN) if the
2678 overall match fails. If (*THEN) is not directly inside an alternation, it acts
2679 like (*PRUNE).
2680 .
2681 .P
2682 The above verbs provide four different "strengths" of control when subsequent
2683 matching fails. (*THEN) is the weakest, carrying on the match at the next
2684 alternation. (*PRUNE) comes next, failing the match at the current starting
2685 position, but allowing an advance to the next character (for an unanchored
2686 pattern). (*SKIP) is similar, except that the advance may be more than one
2687 character. (*COMMIT) is the strongest, causing the entire match to fail.
2688 .P
2689 If more than one is present in a pattern, the "stongest" one wins. For example,
2690 consider this pattern, where A, B, etc. are complex pattern fragments:
2691 .sp
2692 (A(*COMMIT)B(*THEN)C|D)
2693 .sp
2694 Once A has matched, PCRE is committed to this match, at the current starting
2695 position. If subsequently B matches, but C does not, the normal (*THEN) action
2696 of trying the next alternation (that is, D) does not happen because (*COMMIT)
2697 overrides.
2698 .
2699 .
2700 .SH "SEE ALSO"
2701 .rs
2702 .sp
2703 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3),
2704 \fBpcresyntax\fP(3), \fBpcre\fP(3).
2705 .
2706 .
2708 .rs
2709 .sp
2710 .nf
2711 Philip Hazel
2712 University Computing Service
2713 Cambridge CB2 3QH, England.
2714 .fi
2715 .
2716 .
2718 .rs
2719 .sp
2720 .nf
2721 Last updated: 17 November 2010
2722 Copyright (c) 1997-2010 University of Cambridge.
2723 .fi


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

  ViewVC Help
Powered by ViewVC 1.1.5