ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 185 - (show annotations)
Tue Jun 19 13:39:46 2007 UTC (14 years, 3 months ago) by ph10
File size: 80656 byte(s)
Documentation final tidies for 7.2 release.
3 PCRE - Perl-compatible regular expressions
5 .rs
6 .sp
7 The syntax and semantics of the regular expressions supported by PCRE are
8 described below. Regular expressions are also described in the Perl
9 documentation and in a number of books, some of which have copious examples.
10 Jeffrey Friedl's "Mastering Regular Expressions", published by O'Reilly, covers
11 regular expressions in great detail. This description of PCRE's regular
12 expressions is intended as reference material.
13 .P
14 The original operation of PCRE was on strings of one-byte characters. However,
15 there is now also support for UTF-8 character strings. To use this, you must
16 build PCRE to include UTF-8 support, and then call \fBpcre_compile()\fP with
17 the PCRE_UTF8 option. How this affects pattern matching is mentioned in several
18 places below. There is also a summary of UTF-8 features in the
19 .\" HTML <a href="pcre.html#utf8support">
20 .\" </a>
21 section on UTF-8 support
22 .\"
23 in the main
24 .\" HREF
25 \fBpcre\fP
26 .\"
27 page.
28 .P
29 The remainder of this document discusses the patterns that are supported by
30 PCRE when its main matching function, \fBpcre_exec()\fP, is used.
31 From release 6.0, PCRE offers a second matching function,
32 \fBpcre_dfa_exec()\fP, which matches using a different algorithm that is not
33 Perl-compatible. Some of the features discussed below are not available when
34 \fBpcre_dfa_exec()\fP is used. The advantages and disadvantages of the
35 alternative function, and how it differs from the normal function, are
36 discussed in the
37 .\" HREF
38 \fBpcrematching\fP
39 .\"
40 page.
41 .
42 .
44 .rs
45 .sp
46 A regular expression is a pattern that is matched against a subject string from
47 left to right. Most characters stand for themselves in a pattern, and match the
48 corresponding characters in the subject. As a trivial example, the pattern
49 .sp
50 The quick brown fox
51 .sp
52 matches a portion of a subject string that is identical to itself. When
53 caseless matching is specified (the PCRE_CASELESS option), letters are matched
54 independently of case. In UTF-8 mode, PCRE always understands the concept of
55 case for characters whose values are less than 128, so caseless matching is
56 always possible. For characters with higher values, the concept of case is
57 supported if PCRE is compiled with Unicode property support, but not otherwise.
58 If you want to use caseless matching for characters 128 and above, you must
59 ensure that PCRE is compiled with Unicode property support as well as with
60 UTF-8 support.
61 .P
62 The power of regular expressions comes from the ability to include alternatives
63 and repetitions in the pattern. These are encoded in the pattern by the use of
64 \fImetacharacters\fP, which do not stand for themselves but instead are
65 interpreted in some special way.
66 .P
67 There are two different sets of metacharacters: those that are recognized
68 anywhere in the pattern except within square brackets, and those that are
69 recognized within square brackets. Outside square brackets, the metacharacters
70 are as follows:
71 .sp
72 \e general escape character with several uses
73 ^ assert start of string (or line, in multiline mode)
74 $ assert end of string (or line, in multiline mode)
75 . match any character except newline (by default)
76 [ start character class definition
77 | start of alternative branch
78 ( start subpattern
79 ) end subpattern
80 ? extends the meaning of (
81 also 0 or 1 quantifier
82 also quantifier minimizer
83 * 0 or more quantifier
84 + 1 or more quantifier
85 also "possessive quantifier"
86 { start min/max quantifier
87 .sp
88 Part of a pattern that is in square brackets is called a "character class". In
89 a character class the only metacharacters are:
90 .sp
91 \e general escape character
92 ^ negate the class, but only if the first character
93 - indicates character range
94 .\" JOIN
95 [ POSIX character class (only if followed by POSIX
96 syntax)
97 ] terminates the character class
98 .sp
99 The following sections describe the use of each of the metacharacters.
100 .
101 .
103 .rs
104 .sp
105 The backslash character has several uses. Firstly, if it is followed by a
106 non-alphanumeric character, it takes away any special meaning that character
107 may have. This use of backslash as an escape character applies both inside and
108 outside character classes.
109 .P
110 For example, if you want to match a * character, you write \e* in the pattern.
111 This escaping action applies whether or not the following character would
112 otherwise be interpreted as a metacharacter, so it is always safe to precede a
113 non-alphanumeric with backslash to specify that it stands for itself. In
114 particular, if you want to match a backslash, you write \e\e.
115 .P
116 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
117 pattern (other than in a character class) and characters between a # outside
118 a character class and the next newline are ignored. An escaping backslash can
119 be used to include a whitespace or # character as part of the pattern.
120 .P
121 If you want to remove the special meaning from a sequence of characters, you
122 can do so by putting them between \eQ and \eE. This is different from Perl in
123 that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
124 Perl, $ and @ cause variable interpolation. Note the following examples:
125 .sp
126 Pattern PCRE matches Perl matches
127 .sp
128 .\" JOIN
129 \eQabc$xyz\eE abc$xyz abc followed by the
130 contents of $xyz
131 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
132 \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
133 .sp
134 The \eQ...\eE sequence is recognized both inside and outside character classes.
135 .
136 .
137 .\" HTML <a name="digitsafterbackslash"></a>
138 .SS "Non-printing characters"
139 .rs
140 .sp
141 A second use of backslash provides a way of encoding non-printing characters
142 in patterns in a visible manner. There is no restriction on the appearance of
143 non-printing characters, apart from the binary zero that terminates a pattern,
144 but when a pattern is being prepared by text editing, it is usually easier to
145 use one of the following escape sequences than the binary character it
146 represents:
147 .sp
148 \ea alarm, that is, the BEL character (hex 07)
149 \ecx "control-x", where x is any character
150 \ee escape (hex 1B)
151 \ef formfeed (hex 0C)
152 \en newline (hex 0A)
153 \er carriage return (hex 0D)
154 \et tab (hex 09)
155 \eddd character with octal code ddd, or backreference
156 \exhh character with hex code hh
157 \ex{hhh..} character with hex code hhh..
158 .sp
159 The precise effect of \ecx is as follows: if x is a lower case letter, it
160 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
161 Thus \ecz becomes hex 1A, but \ec{ becomes hex 3B, while \ec; becomes hex
162 7B.
163 .P
164 After \ex, from zero to two hexadecimal digits are read (letters can be in
165 upper or lower case). Any number of hexadecimal digits may appear between \ex{
166 and }, but the value of the character code must be less than 256 in non-UTF-8
167 mode, and less than 2**31 in UTF-8 mode (that is, the maximum hexadecimal value
168 is 7FFFFFFF). If characters other than hexadecimal digits appear between \ex{
169 and }, or if there is no terminating }, this form of escape is not recognized.
170 Instead, the initial \ex will be interpreted as a basic hexadecimal escape,
171 with no following digits, giving a character whose value is zero.
172 .P
173 Characters whose value is less than 256 can be defined by either of the two
174 syntaxes for \ex. There is no difference in the way they are handled. For
175 example, \exdc is exactly the same as \ex{dc}.
176 .P
177 After \e0 up to two further octal digits are read. If there are fewer than two
178 digits, just those that are present are used. Thus the sequence \e0\ex\e07
179 specifies two binary zeros followed by a BEL character (code value 7). Make
180 sure you supply two digits after the initial zero if the pattern character that
181 follows is itself an octal digit.
182 .P
183 The handling of a backslash followed by a digit other than 0 is complicated.
184 Outside a character class, PCRE reads it and any following digits as a decimal
185 number. If the number is less than 10, or if there have been at least that many
186 previous capturing left parentheses in the expression, the entire sequence is
187 taken as a \fIback reference\fP. A description of how this works is given
188 .\" HTML <a href="#backreferences">
189 .\" </a>
190 later,
191 .\"
192 following the discussion of
193 .\" HTML <a href="#subpattern">
194 .\" </a>
195 parenthesized subpatterns.
196 .\"
197 .P
198 Inside a character class, or if the decimal number is greater than 9 and there
199 have not been that many capturing subpatterns, PCRE re-reads up to three octal
200 digits following the backslash, and uses them to generate a data character. Any
201 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
202 character specified in octal must be less than \e400. In UTF-8 mode, values up
203 to \e777 are permitted. For example:
204 .sp
205 \e040 is another way of writing a space
206 .\" JOIN
207 \e40 is the same, provided there are fewer than 40
208 previous capturing subpatterns
209 \e7 is always a back reference
210 .\" JOIN
211 \e11 might be a back reference, or another way of
212 writing a tab
213 \e011 is always a tab
214 \e0113 is a tab followed by the character "3"
215 .\" JOIN
216 \e113 might be a back reference, otherwise the
217 character with octal code 113
218 .\" JOIN
219 \e377 might be a back reference, otherwise
220 the byte consisting entirely of 1 bits
221 .\" JOIN
222 \e81 is either a back reference, or a binary zero
223 followed by the two characters "8" and "1"
224 .sp
225 Note that octal values of 100 or greater must not be introduced by a leading
226 zero, because no more than three octal digits are ever read.
227 .P
228 All the sequences that define a single character value can be used both inside
229 and outside character classes. In addition, inside a character class, the
230 sequence \eb is interpreted as the backspace character (hex 08), and the
231 sequences \eR and \eX are interpreted as the characters "R" and "X",
232 respectively. Outside a character class, these sequences have different
233 meanings
234 .\" HTML <a href="#uniextseq">
235 .\" </a>
236 (see below).
237 .\"
238 .
239 .
240 .SS "Absolute and relative back references"
241 .rs
242 .sp
243 The sequence \eg followed by a positive or negative number, optionally enclosed
244 in braces, is an absolute or relative back reference. A named back reference
245 can be coded as \eg{name}. Back references are discussed
246 .\" HTML <a href="#backreferences">
247 .\" </a>
248 later,
249 .\"
250 following the discussion of
251 .\" HTML <a href="#subpattern">
252 .\" </a>
253 parenthesized subpatterns.
254 .\"
255 .
256 .
257 .SS "Generic character types"
258 .rs
259 .sp
260 Another use of backslash is for specifying generic character types. The
261 following are always recognized:
262 .sp
263 \ed any decimal digit
264 \eD any character that is not a decimal digit
265 \eh any horizontal whitespace character
266 \eH any character that is not a horizontal whitespace character
267 \es any whitespace character
268 \eS any character that is not a whitespace character
269 \ev any vertical whitespace character
270 \eV any character that is not a vertical whitespace character
271 \ew any "word" character
272 \eW any "non-word" character
273 .sp
274 Each pair of escape sequences partitions the complete set of characters into
275 two disjoint sets. Any given character matches one, and only one, of each pair.
276 .P
277 These character type sequences can appear both inside and outside character
278 classes. They each match one character of the appropriate type. If the current
279 matching point is at the end of the subject string, all of them fail, since
280 there is no character to match.
281 .P
282 For compatibility with Perl, \es does not match the VT character (code 11).
283 This makes it different from the the POSIX "space" class. The \es characters
284 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
285 included in a Perl script, \es may match the VT character. In PCRE, it never
286 does.
287 .P
288 In UTF-8 mode, characters with values greater than 128 never match \ed, \es, or
289 \ew, and always match \eD, \eS, and \eW. This is true even when Unicode
290 character property support is available. These sequences retain their original
291 meanings from before UTF-8 support was available, mainly for efficiency
292 reasons.
293 .P
294 The sequences \eh, \eH, \ev, and \eV are Perl 5.10 features. In contrast to the
295 other sequences, these do match certain high-valued codepoints in UTF-8 mode.
296 The horizontal space characters are:
297 .sp
298 U+0009 Horizontal tab
299 U+0020 Space
300 U+00A0 Non-break space
301 U+1680 Ogham space mark
302 U+180E Mongolian vowel separator
303 U+2000 En quad
304 U+2001 Em quad
305 U+2002 En space
306 U+2003 Em space
307 U+2004 Three-per-em space
308 U+2005 Four-per-em space
309 U+2006 Six-per-em space
310 U+2007 Figure space
311 U+2008 Punctuation space
312 U+2009 Thin space
313 U+200A Hair space
314 U+202F Narrow no-break space
315 U+205F Medium mathematical space
316 U+3000 Ideographic space
317 .sp
318 The vertical space characters are:
319 .sp
320 U+000A Linefeed
321 U+000B Vertical tab
322 U+000C Formfeed
323 U+000D Carriage return
324 U+0085 Next line
325 U+2028 Line separator
326 U+2029 Paragraph separator
327 .P
328 A "word" character is an underscore or any character less than 256 that is a
329 letter or digit. The definition of letters and digits is controlled by PCRE's
330 low-valued character tables, and may vary if locale-specific matching is taking
331 place (see
332 .\" HTML <a href="pcreapi.html#localesupport">
333 .\" </a>
334 "Locale support"
335 .\"
336 in the
337 .\" HREF
338 \fBpcreapi\fP
339 .\"
340 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
341 or "french" in Windows, some character codes greater than 128 are used for
342 accented letters, and these are matched by \ew. The use of locales with Unicode
343 is discouraged.
344 .
345 .
346 .SS "Newline sequences"
347 .rs
348 .sp
349 Outside a character class, the escape sequence \eR matches any Unicode newline
350 sequence. This is a Perl 5.10 feature. In non-UTF-8 mode \eR is equivalent to
351 the following:
352 .sp
353 (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
354 .sp
355 This is an example of an "atomic group", details of which are given
356 .\" HTML <a href="#atomicgroup">
357 .\" </a>
358 below.
359 .\"
360 This particular group matches either the two-character sequence CR followed by
361 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
362 U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
363 line, U+0085). The two-character sequence is treated as a single unit that
364 cannot be split.
365 .P
366 In UTF-8 mode, two additional characters whose codepoints are greater than 255
367 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
368 Unicode character property support is not needed for these characters to be
369 recognized.
370 .P
371 Inside a character class, \eR matches the letter "R".
372 .
373 .
374 .\" HTML <a name="uniextseq"></a>
375 .SS Unicode character properties
376 .rs
377 .sp
378 When PCRE is built with Unicode character property support, three additional
379 escape sequences that match characters with specific properties are available.
380 When not in UTF-8 mode, these sequences are of course limited to testing
381 characters whose codepoints are less than 256, but they do work in this mode.
382 The extra escape sequences are:
383 .sp
384 \ep{\fIxx\fP} a character with the \fIxx\fP property
385 \eP{\fIxx\fP} a character without the \fIxx\fP property
386 \eX an extended Unicode sequence
387 .sp
388 The property names represented by \fIxx\fP above are limited to the Unicode
389 script names, the general category properties, and "Any", which matches any
390 character (including newline). Other properties such as "InMusicalSymbols" are
391 not currently supported by PCRE. Note that \eP{Any} does not match any
392 characters, so always causes a match failure.
393 .P
394 Sets of Unicode characters are defined as belonging to certain scripts. A
395 character from one of these sets can be matched using a script name. For
396 example:
397 .sp
398 \ep{Greek}
399 \eP{Han}
400 .sp
401 Those that are not part of an identified script are lumped together as
402 "Common". The current list of scripts is:
403 .P
404 Arabic,
405 Armenian,
406 Balinese,
407 Bengali,
408 Bopomofo,
409 Braille,
410 Buginese,
411 Buhid,
412 Canadian_Aboriginal,
413 Cherokee,
414 Common,
415 Coptic,
416 Cuneiform,
417 Cypriot,
418 Cyrillic,
419 Deseret,
420 Devanagari,
421 Ethiopic,
422 Georgian,
423 Glagolitic,
424 Gothic,
425 Greek,
426 Gujarati,
427 Gurmukhi,
428 Han,
429 Hangul,
430 Hanunoo,
431 Hebrew,
432 Hiragana,
433 Inherited,
434 Kannada,
435 Katakana,
436 Kharoshthi,
437 Khmer,
438 Lao,
439 Latin,
440 Limbu,
441 Linear_B,
442 Malayalam,
443 Mongolian,
444 Myanmar,
445 New_Tai_Lue,
446 Nko,
447 Ogham,
448 Old_Italic,
449 Old_Persian,
450 Oriya,
451 Osmanya,
452 Phags_Pa,
453 Phoenician,
454 Runic,
455 Shavian,
456 Sinhala,
457 Syloti_Nagri,
458 Syriac,
459 Tagalog,
460 Tagbanwa,
461 Tai_Le,
462 Tamil,
463 Telugu,
464 Thaana,
465 Thai,
466 Tibetan,
467 Tifinagh,
468 Ugaritic,
469 Yi.
470 .P
471 Each character has exactly one general category property, specified by a
472 two-letter abbreviation. For compatibility with Perl, negation can be specified
473 by including a circumflex between the opening brace and the property name. For
474 example, \ep{^Lu} is the same as \eP{Lu}.
475 .P
476 If only one letter is specified with \ep or \eP, it includes all the general
477 category properties that start with that letter. In this case, in the absence
478 of negation, the curly brackets in the escape sequence are optional; these two
479 examples have the same effect:
480 .sp
481 \ep{L}
482 \epL
483 .sp
484 The following general category property codes are supported:
485 .sp
486 C Other
487 Cc Control
488 Cf Format
489 Cn Unassigned
490 Co Private use
491 Cs Surrogate
492 .sp
493 L Letter
494 Ll Lower case letter
495 Lm Modifier letter
496 Lo Other letter
497 Lt Title case letter
498 Lu Upper case letter
499 .sp
500 M Mark
501 Mc Spacing mark
502 Me Enclosing mark
503 Mn Non-spacing mark
504 .sp
505 N Number
506 Nd Decimal number
507 Nl Letter number
508 No Other number
509 .sp
510 P Punctuation
511 Pc Connector punctuation
512 Pd Dash punctuation
513 Pe Close punctuation
514 Pf Final punctuation
515 Pi Initial punctuation
516 Po Other punctuation
517 Ps Open punctuation
518 .sp
519 S Symbol
520 Sc Currency symbol
521 Sk Modifier symbol
522 Sm Mathematical symbol
523 So Other symbol
524 .sp
525 Z Separator
526 Zl Line separator
527 Zp Paragraph separator
528 Zs Space separator
529 .sp
530 The special property L& is also supported: it matches a character that has
531 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
532 a modifier or "other".
533 .P
534 The long synonyms for these properties that Perl supports (such as \ep{Letter})
535 are not supported by PCRE, nor is it permitted to prefix any of these
536 properties with "Is".
537 .P
538 No character that is in the Unicode table has the Cn (unassigned) property.
539 Instead, this property is assumed for any code point that is not in the
540 Unicode table.
541 .P
542 Specifying caseless matching does not affect these escape sequences. For
543 example, \ep{Lu} always matches only upper case letters.
544 .P
545 The \eX escape matches any number of Unicode characters that form an extended
546 Unicode sequence. \eX is equivalent to
547 .sp
548 (?>\ePM\epM*)
549 .sp
550 That is, it matches a character without the "mark" property, followed by zero
551 or more characters with the "mark" property, and treats the sequence as an
552 atomic group
553 .\" HTML <a href="#atomicgroup">
554 .\" </a>
555 (see below).
556 .\"
557 Characters with the "mark" property are typically accents that affect the
558 preceding character. None of them have codepoints less than 256, so in
559 non-UTF-8 mode \eX matches any one character.
560 .P
561 Matching characters by Unicode property is not fast, because PCRE has to search
562 a structure that contains data for over fifteen thousand characters. That is
563 why the traditional escape sequences such as \ed and \ew do not use Unicode
564 properties in PCRE.
565 .
566 .
567 .\" HTML <a name="resetmatchstart"></a>
568 .SS "Resetting the match start"
569 .rs
570 .sp
571 The escape sequence \eK, which is a Perl 5.10 feature, causes any previously
572 matched characters not to be included in the final matched sequence. For
573 example, the pattern:
574 .sp
575 foo\eKbar
576 .sp
577 matches "foobar", but reports that it has matched "bar". This feature is
578 similar to a lookbehind assertion
579 .\" HTML <a href="#lookbehind">
580 .\" </a>
581 (described below).
582 .\"
583 However, in this case, the part of the subject before the real match does not
584 have to be of fixed length, as lookbehind assertions do. The use of \eK does
585 not interfere with the setting of
586 .\" HTML <a href="#subpattern">
587 .\" </a>
588 captured substrings.
589 .\"
590 For example, when the pattern
591 .sp
592 (foo)\eKbar
593 .sp
594 matches "foobar", the first substring is still set to "foo".
595 .
596 .
597 .\" HTML <a name="smallassertions"></a>
598 .SS "Simple assertions"
599 .rs
600 .sp
601 The final use of backslash is for certain simple assertions. An assertion
602 specifies a condition that has to be met at a particular point in a match,
603 without consuming any characters from the subject string. The use of
604 subpatterns for more complicated assertions is described
605 .\" HTML <a href="#bigassertions">
606 .\" </a>
607 below.
608 .\"
609 The backslashed assertions are:
610 .sp
611 \eb matches at a word boundary
612 \eB matches when not at a word boundary
613 \eA matches at the start of the subject
614 \eZ matches at the end of the subject
615 also matches before a newline at the end of the subject
616 \ez matches only at the end of the subject
617 \eG matches at the first matching position in the subject
618 .sp
619 These assertions may not appear in character classes (but note that \eb has a
620 different meaning, namely the backspace character, inside a character class).
621 .P
622 A word boundary is a position in the subject string where the current character
623 and the previous character do not both match \ew or \eW (i.e. one matches
624 \ew and the other matches \eW), or the start or end of the string if the
625 first or last character matches \ew, respectively.
626 .P
627 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
628 dollar (described in the next section) in that they only ever match at the very
629 start and end of the subject string, whatever options are set. Thus, they are
630 independent of multiline mode. These three assertions are not affected by the
631 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
632 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
633 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
634 at a point other than the beginning of the subject, \eA can never match. The
635 difference between \eZ and \ez is that \eZ matches before a newline at the end
636 of the string as well as at the very end, whereas \ez matches only at the end.
637 .P
638 The \eG assertion is true only when the current matching position is at the
639 start point of the match, as specified by the \fIstartoffset\fP argument of
640 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
641 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
642 arguments, you can mimic Perl's /g option, and it is in this kind of
643 implementation where \eG can be useful.
644 .P
645 Note, however, that PCRE's interpretation of \eG, as the start of the current
646 match, is subtly different from Perl's, which defines it as the end of the
647 previous match. In Perl, these can be different when the previously matched
648 string was empty. Because PCRE does just one match at a time, it cannot
649 reproduce this behaviour.
650 .P
651 If all the alternatives of a pattern begin with \eG, the expression is anchored
652 to the starting match position, and the "anchored" flag is set in the compiled
653 regular expression.
654 .
655 .
657 .rs
658 .sp
659 Outside a character class, in the default matching mode, the circumflex
660 character is an assertion that is true only if the current matching point is
661 at the start of the subject string. If the \fIstartoffset\fP argument of
662 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
663 option is unset. Inside a character class, circumflex has an entirely different
664 meaning
665 .\" HTML <a href="#characterclass">
666 .\" </a>
667 (see below).
668 .\"
669 .P
670 Circumflex need not be the first character of the pattern if a number of
671 alternatives are involved, but it should be the first thing in each alternative
672 in which it appears if the pattern is ever to match that branch. If all
673 possible alternatives start with a circumflex, that is, if the pattern is
674 constrained to match only at the start of the subject, it is said to be an
675 "anchored" pattern. (There are also other constructs that can cause a pattern
676 to be anchored.)
677 .P
678 A dollar character is an assertion that is true only if the current matching
679 point is at the end of the subject string, or immediately before a newline
680 at the end of the string (by default). Dollar need not be the last character of
681 the pattern if a number of alternatives are involved, but it should be the last
682 item in any branch in which it appears. Dollar has no special meaning in a
683 character class.
684 .P
685 The meaning of dollar can be changed so that it matches only at the very end of
686 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
687 does not affect the \eZ assertion.
688 .P
689 The meanings of the circumflex and dollar characters are changed if the
690 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
691 immediately after internal newlines as well as at the start of the subject
692 string. It does not match after a newline that ends the string. A dollar
693 matches before any newlines in the string, as well as at the very end, when
694 PCRE_MULTILINE is set. When newline is specified as the two-character
695 sequence CRLF, isolated CR and LF characters do not indicate newlines.
696 .P
697 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
698 \en represents a newline) in multiline mode, but not otherwise. Consequently,
699 patterns that are anchored in single line mode because all branches start with
700 ^ are not anchored in multiline mode, and a match for circumflex is possible
701 when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
702 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
703 .P
704 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
705 end of the subject in both modes, and if all branches of a pattern start with
706 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
707 .
708 .
710 .rs
711 .sp
712 Outside a character class, a dot in the pattern matches any one character in
713 the subject string except (by default) a character that signifies the end of a
714 line. In UTF-8 mode, the matched character may be more than one byte long.
715 .P
716 When a line ending is defined as a single character, dot never matches that
717 character; when the two-character sequence CRLF is used, dot does not match CR
718 if it is immediately followed by LF, but otherwise it matches all characters
719 (including isolated CRs and LFs). When any Unicode line endings are being
720 recognized, dot does not match CR or LF or any of the other line ending
721 characters.
722 .P
723 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
724 option is set, a dot matches any one character, without exception. If the
725 two-character sequence CRLF is present in the subject string, it takes two dots
726 to match it.
727 .P
728 The handling of dot is entirely independent of the handling of circumflex and
729 dollar, the only relationship being that they both involve newlines. Dot has no
730 special meaning in a character class.
731 .
732 .
734 .rs
735 .sp
736 Outside a character class, the escape sequence \eC matches any one byte, both
737 in and out of UTF-8 mode. Unlike a dot, it always matches any line-ending
738 characters. The feature is provided in Perl in order to match individual bytes
739 in UTF-8 mode. Because it breaks up UTF-8 characters into individual bytes,
740 what remains in the string may be a malformed UTF-8 string. For this reason,
741 the \eC escape sequence is best avoided.
742 .P
743 PCRE does not allow \eC to appear in lookbehind assertions
744 .\" HTML <a href="#lookbehind">
745 .\" </a>
746 (described below),
747 .\"
748 because in UTF-8 mode this would make it impossible to calculate the length of
749 the lookbehind.
750 .
751 .
752 .\" HTML <a name="characterclass"></a>
754 .rs
755 .sp
756 An opening square bracket introduces a character class, terminated by a closing
757 square bracket. A closing square bracket on its own is not special. If a
758 closing square bracket is required as a member of the class, it should be the
759 first data character in the class (after an initial circumflex, if present) or
760 escaped with a backslash.
761 .P
762 A character class matches a single character in the subject. In UTF-8 mode, the
763 character may occupy more than one byte. A matched character must be in the set
764 of characters defined by the class, unless the first character in the class
765 definition is a circumflex, in which case the subject character must not be in
766 the set defined by the class. If a circumflex is actually required as a member
767 of the class, ensure it is not the first character, or escape it with a
768 backslash.
769 .P
770 For example, the character class [aeiou] matches any lower case vowel, while
771 [^aeiou] matches any character that is not a lower case vowel. Note that a
772 circumflex is just a convenient notation for specifying the characters that
773 are in the class by enumerating those that are not. A class that starts with a
774 circumflex is not an assertion: it still consumes a character from the subject
775 string, and therefore it fails if the current pointer is at the end of the
776 string.
777 .P
778 In UTF-8 mode, characters with values greater than 255 can be included in a
779 class as a literal string of bytes, or by using the \ex{ escaping mechanism.
780 .P
781 When caseless matching is set, any letters in a class represent both their
782 upper case and lower case versions, so for example, a caseless [aeiou] matches
783 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
784 caseful version would. In UTF-8 mode, PCRE always understands the concept of
785 case for characters whose values are less than 128, so caseless matching is
786 always possible. For characters with higher values, the concept of case is
787 supported if PCRE is compiled with Unicode property support, but not otherwise.
788 If you want to use caseless matching for characters 128 and above, you must
789 ensure that PCRE is compiled with Unicode property support as well as with
790 UTF-8 support.
791 .P
792 Characters that might indicate line breaks are never treated in any special way
793 when matching character classes, whatever line-ending sequence is in use, and
794 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
795 such as [^a] always matches one of these characters.
796 .P
797 The minus (hyphen) character can be used to specify a range of characters in a
798 character class. For example, [d-m] matches any letter between d and m,
799 inclusive. If a minus character is required in a class, it must be escaped with
800 a backslash or appear in a position where it cannot be interpreted as
801 indicating a range, typically as the first or last character in the class.
802 .P
803 It is not possible to have the literal character "]" as the end character of a
804 range. A pattern such as [W-]46] is interpreted as a class of two characters
805 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
806 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
807 the end of range, so [W-\e]46] is interpreted as a class containing a range
808 followed by two other characters. The octal or hexadecimal representation of
809 "]" can also be used to end a range.
810 .P
811 Ranges operate in the collating sequence of character values. They can also be
812 used for characters specified numerically, for example [\e000-\e037]. In UTF-8
813 mode, ranges can include characters whose values are greater than 255, for
814 example [\ex{100}-\ex{2ff}].
815 .P
816 If a range that includes letters is used when caseless matching is set, it
817 matches the letters in either case. For example, [W-c] is equivalent to
818 [][\e\e^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
819 tables for a French locale are in use, [\exc8-\excb] matches accented E
820 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
821 characters with values greater than 128 only when it is compiled with Unicode
822 property support.
823 .P
824 The character types \ed, \eD, \ep, \eP, \es, \eS, \ew, and \eW may also appear
825 in a character class, and add the characters that they match to the class. For
826 example, [\edABCDEF] matches any hexadecimal digit. A circumflex can
827 conveniently be used with the upper case character types to specify a more
828 restricted set of characters than the matching lower case type. For example,
829 the class [^\eW_] matches any letter or digit, but not underscore.
830 .P
831 The only metacharacters that are recognized in character classes are backslash,
832 hyphen (only where it can be interpreted as specifying a range), circumflex
833 (only at the start), opening square bracket (only when it can be interpreted as
834 introducing a POSIX class name - see the next section), and the terminating
835 closing square bracket. However, escaping other non-alphanumeric characters
836 does no harm.
837 .
838 .
840 .rs
841 .sp
842 Perl supports the POSIX notation for character classes. This uses names
843 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
844 this notation. For example,
845 .sp
846 [01[:alpha:]%]
847 .sp
848 matches "0", "1", any alphabetic character, or "%". The supported class names
849 are
850 .sp
851 alnum letters and digits
852 alpha letters
853 ascii character codes 0 - 127
854 blank space or tab only
855 cntrl control characters
856 digit decimal digits (same as \ed)
857 graph printing characters, excluding space
858 lower lower case letters
859 print printing characters, including space
860 punct printing characters, excluding letters and digits
861 space white space (not quite the same as \es)
862 upper upper case letters
863 word "word" characters (same as \ew)
864 xdigit hexadecimal digits
865 .sp
866 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
867 space (32). Notice that this list includes the VT character (code 11). This
868 makes "space" different to \es, which does not include VT (for Perl
869 compatibility).
870 .P
871 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
872 5.8. Another Perl extension is negation, which is indicated by a ^ character
873 after the colon. For example,
874 .sp
875 [12[:^digit:]]
876 .sp
877 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
878 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
879 supported, and an error is given if they are encountered.
880 .P
881 In UTF-8 mode, characters with values greater than 128 do not match any of
882 the POSIX character classes.
883 .
884 .
886 .rs
887 .sp
888 Vertical bar characters are used to separate alternative patterns. For example,
889 the pattern
890 .sp
891 gilbert|sullivan
892 .sp
893 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
894 and an empty alternative is permitted (matching the empty string). The matching
895 process tries each alternative in turn, from left to right, and the first one
896 that succeeds is used. If the alternatives are within a subpattern
897 .\" HTML <a href="#subpattern">
898 .\" </a>
899 (defined below),
900 .\"
901 "succeeds" means matching the rest of the main pattern as well as the
902 alternative in the subpattern.
903 .
904 .
906 .rs
907 .sp
909 PCRE_EXTENDED options can be changed from within the pattern by a sequence of
910 Perl option letters enclosed between "(?" and ")". The option letters are
911 .sp
914 s for PCRE_DOTALL
916 .sp
917 For example, (?im) sets caseless, multiline matching. It is also possible to
918 unset these options by preceding the letter with a hyphen, and a combined
919 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
920 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
921 permitted. If a letter appears both before and after the hyphen, the option is
922 unset.
923 .P
924 When an option change occurs at top level (that is, not inside subpattern
925 parentheses), the change applies to the remainder of the pattern that follows.
926 If the change is placed right at the start of a pattern, PCRE extracts it into
927 the global options (and it will therefore show up in data extracted by the
928 \fBpcre_fullinfo()\fP function).
929 .P
930 An option change within a subpattern (see below for a description of
931 subpatterns) affects only that part of the current pattern that follows it, so
932 .sp
933 (a(?i)b)c
934 .sp
935 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
936 By this means, options can be made to have different settings in different
937 parts of the pattern. Any changes made in one alternative do carry on
938 into subsequent branches within the same subpattern. For example,
939 .sp
940 (a(?i)b|c)
941 .sp
942 matches "ab", "aB", "c", and "C", even though when matching "C" the first
943 branch is abandoned before the option setting. This is because the effects of
944 option settings happen at compile time. There would be some very weird
945 behaviour otherwise.
946 .P
947 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
948 changed in the same way as the Perl-compatible options by using the characters
949 J, U and X respectively.
950 .
951 .
952 .\" HTML <a name="subpattern"></a>
954 .rs
955 .sp
956 Subpatterns are delimited by parentheses (round brackets), which can be nested.
957 Turning part of a pattern into a subpattern does two things:
958 .sp
959 1. It localizes a set of alternatives. For example, the pattern
960 .sp
961 cat(aract|erpillar|)
962 .sp
963 matches one of the words "cat", "cataract", or "caterpillar". Without the
964 parentheses, it would match "cataract", "erpillar" or an empty string.
965 .sp
966 2. It sets up the subpattern as a capturing subpattern. This means that, when
967 the whole pattern matches, that portion of the subject string that matched the
968 subpattern is passed back to the caller via the \fIovector\fP argument of
969 \fBpcre_exec()\fP. Opening parentheses are counted from left to right (starting
970 from 1) to obtain numbers for the capturing subpatterns.
971 .P
972 For example, if the string "the red king" is matched against the pattern
973 .sp
974 the ((red|white) (king|queen))
975 .sp
976 the captured substrings are "red king", "red", and "king", and are numbered 1,
977 2, and 3, respectively.
978 .P
979 The fact that plain parentheses fulfil two functions is not always helpful.
980 There are often times when a grouping subpattern is required without a
981 capturing requirement. If an opening parenthesis is followed by a question mark
982 and a colon, the subpattern does not do any capturing, and is not counted when
983 computing the number of any subsequent capturing subpatterns. For example, if
984 the string "the white queen" is matched against the pattern
985 .sp
986 the ((?:red|white) (king|queen))
987 .sp
988 the captured substrings are "white queen" and "queen", and are numbered 1 and
989 2. The maximum number of capturing subpatterns is 65535.
990 .P
991 As a convenient shorthand, if any option settings are required at the start of
992 a non-capturing subpattern, the option letters may appear between the "?" and
993 the ":". Thus the two patterns
994 .sp
995 (?i:saturday|sunday)
996 (?:(?i)saturday|sunday)
997 .sp
998 match exactly the same set of strings. Because alternative branches are tried
999 from left to right, and options are not reset until the end of the subpattern
1000 is reached, an option setting in one branch does affect subsequent branches, so
1001 the above patterns match "SUNDAY" as well as "Saturday".
1002 .
1003 .
1005 .rs
1006 .sp
1007 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1008 the same numbers for its capturing parentheses. Such a subpattern starts with
1009 (?| and is itself a non-capturing subpattern. For example, consider this
1010 pattern:
1011 .sp
1012 (?|(Sat)ur|(Sun))day
1013 .sp
1014 Because the two alternatives are inside a (?| group, both sets of capturing
1015 parentheses are numbered one. Thus, when the pattern matches, you can look
1016 at captured substring number one, whichever alternative matched. This construct
1017 is useful when you want to capture part, but not all, of one of a number of
1018 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1019 number is reset at the start of each branch. The numbers of any capturing
1020 buffers that follow the subpattern start after the highest number used in any
1021 branch. The following example is taken from the Perl documentation.
1022 The numbers underneath show in which buffer the captured content will be
1023 stored.
1024 .sp
1025 # before ---------------branch-reset----------- after
1026 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1027 # 1 2 2 3 2 3 4
1028 .sp
1029 A backreference or a recursive call to a numbered subpattern always refers to
1030 the first one in the pattern with the given number.
1031 .P
1032 An alternative approach to using this "branch reset" feature is to use
1033 duplicate named subpatterns, as described in the next section.
1034 .
1035 .
1037 .rs
1038 .sp
1039 Identifying capturing parentheses by number is simple, but it can be very hard
1040 to keep track of the numbers in complicated regular expressions. Furthermore,
1041 if an expression is modified, the numbers may change. To help with this
1042 difficulty, PCRE supports the naming of subpatterns. This feature was not
1043 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1044 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1045 the Perl and the Python syntax.
1046 .P
1047 In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1048 (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1049 parentheses from other parts of the pattern, such as
1050 .\" HTML <a href="#backreferences">
1051 .\" </a>
1052 backreferences,
1053 .\"
1054 .\" HTML <a href="#recursion">
1055 .\" </a>
1056 recursion,
1057 .\"
1058 and
1059 .\" HTML <a href="#conditions">
1060 .\" </a>
1061 conditions,
1062 .\"
1063 can be made by name as well as by number.
1064 .P
1065 Names consist of up to 32 alphanumeric characters and underscores. Named
1066 capturing parentheses are still allocated numbers as well as names, exactly as
1067 if the names were not present. The PCRE API provides function calls for
1068 extracting the name-to-number translation table from a compiled pattern. There
1069 is also a convenience function for extracting a captured substring by name.
1070 .P
1071 By default, a name must be unique within a pattern, but it is possible to relax
1072 this constraint by setting the PCRE_DUPNAMES option at compile time. This can
1073 be useful for patterns where only one instance of the named parentheses can
1074 match. Suppose you want to match the name of a weekday, either as a 3-letter
1075 abbreviation or as the full name, and in both cases you want to extract the
1076 abbreviation. This pattern (ignoring the line breaks) does the job:
1077 .sp
1078 (?<DN>Mon|Fri|Sun)(?:day)?|
1079 (?<DN>Tue)(?:sday)?|
1080 (?<DN>Wed)(?:nesday)?|
1081 (?<DN>Thu)(?:rsday)?|
1082 (?<DN>Sat)(?:urday)?
1083 .sp
1084 There are five capturing substrings, but only one is ever set after a match.
1085 (An alternative way of solving this problem is to use a "branch reset"
1086 subpattern, as described in the previous section.)
1087 .P
1088 The convenience function for extracting the data by name returns the substring
1089 for the first (and in this example, the only) subpattern of that name that
1090 matched. This saves searching to find which numbered subpattern it was. If you
1091 make a reference to a non-unique named subpattern from elsewhere in the
1092 pattern, the one that corresponds to the lowest number is used. For further
1093 details of the interfaces for handling named subpatterns, see the
1094 .\" HREF
1095 \fBpcreapi\fP
1096 .\"
1097 documentation.
1098 .
1099 .
1101 .rs
1102 .sp
1103 Repetition is specified by quantifiers, which can follow any of the following
1104 items:
1105 .sp
1106 a literal data character
1107 the dot metacharacter
1108 the \eC escape sequence
1109 the \eX escape sequence (in UTF-8 mode with Unicode properties)
1110 the \eR escape sequence
1111 an escape such as \ed that matches a single character
1112 a character class
1113 a back reference (see next section)
1114 a parenthesized subpattern (unless it is an assertion)
1115 .sp
1116 The general repetition quantifier specifies a minimum and maximum number of
1117 permitted matches, by giving the two numbers in curly brackets (braces),
1118 separated by a comma. The numbers must be less than 65536, and the first must
1119 be less than or equal to the second. For example:
1120 .sp
1121 z{2,4}
1122 .sp
1123 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1124 character. If the second number is omitted, but the comma is present, there is
1125 no upper limit; if the second number and the comma are both omitted, the
1126 quantifier specifies an exact number of required matches. Thus
1127 .sp
1128 [aeiou]{3,}
1129 .sp
1130 matches at least 3 successive vowels, but may match many more, while
1131 .sp
1132 \ed{8}
1133 .sp
1134 matches exactly 8 digits. An opening curly bracket that appears in a position
1135 where a quantifier is not allowed, or one that does not match the syntax of a
1136 quantifier, is taken as a literal character. For example, {,6} is not a
1137 quantifier, but a literal string of four characters.
1138 .P
1139 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1140 bytes. Thus, for example, \ex{100}{2} matches two UTF-8 characters, each of
1141 which is represented by a two-byte sequence. Similarly, when Unicode property
1142 support is available, \eX{3} matches three Unicode extended sequences, each of
1143 which may be several bytes long (and they may be of different lengths).
1144 .P
1145 The quantifier {0} is permitted, causing the expression to behave as if the
1146 previous item and the quantifier were not present.
1147 .P
1148 For convenience, the three most common quantifiers have single-character
1149 abbreviations:
1150 .sp
1151 * is equivalent to {0,}
1152 + is equivalent to {1,}
1153 ? is equivalent to {0,1}
1154 .sp
1155 It is possible to construct infinite loops by following a subpattern that can
1156 match no characters with a quantifier that has no upper limit, for example:
1157 .sp
1158 (a?)*
1159 .sp
1160 Earlier versions of Perl and PCRE used to give an error at compile time for
1161 such patterns. However, because there are cases where this can be useful, such
1162 patterns are now accepted, but if any repetition of the subpattern does in fact
1163 match no characters, the loop is forcibly broken.
1164 .P
1165 By default, the quantifiers are "greedy", that is, they match as much as
1166 possible (up to the maximum number of permitted times), without causing the
1167 rest of the pattern to fail. The classic example of where this gives problems
1168 is in trying to match comments in C programs. These appear between /* and */
1169 and within the comment, individual * and / characters may appear. An attempt to
1170 match C comments by applying the pattern
1171 .sp
1172 /\e*.*\e*/
1173 .sp
1174 to the string
1175 .sp
1176 /* first comment */ not comment /* second comment */
1177 .sp
1178 fails, because it matches the entire string owing to the greediness of the .*
1179 item.
1180 .P
1181 However, if a quantifier is followed by a question mark, it ceases to be
1182 greedy, and instead matches the minimum number of times possible, so the
1183 pattern
1184 .sp
1185 /\e*.*?\e*/
1186 .sp
1187 does the right thing with the C comments. The meaning of the various
1188 quantifiers is not otherwise changed, just the preferred number of matches.
1189 Do not confuse this use of question mark with its use as a quantifier in its
1190 own right. Because it has two uses, it can sometimes appear doubled, as in
1191 .sp
1192 \ed??\ed
1193 .sp
1194 which matches one digit by preference, but can match two if that is the only
1195 way the rest of the pattern matches.
1196 .P
1197 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1198 the quantifiers are not greedy by default, but individual ones can be made
1199 greedy by following them with a question mark. In other words, it inverts the
1200 default behaviour.
1201 .P
1202 When a parenthesized subpattern is quantified with a minimum repeat count that
1203 is greater than 1 or with a limited maximum, more memory is required for the
1204 compiled pattern, in proportion to the size of the minimum or maximum.
1205 .P
1206 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1207 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1208 implicitly anchored, because whatever follows will be tried against every
1209 character position in the subject string, so there is no point in retrying the
1210 overall match at any position after the first. PCRE normally treats such a
1211 pattern as though it were preceded by \eA.
1212 .P
1213 In cases where it is known that the subject string contains no newlines, it is
1214 worth setting PCRE_DOTALL in order to obtain this optimization, or
1215 alternatively using ^ to indicate anchoring explicitly.
1216 .P
1217 However, there is one situation where the optimization cannot be used. When .*
1218 is inside capturing parentheses that are the subject of a backreference
1219 elsewhere in the pattern, a match at the start may fail where a later one
1220 succeeds. Consider, for example:
1221 .sp
1222 (.*)abc\e1
1223 .sp
1224 If the subject is "xyz123abc123" the match point is the fourth character. For
1225 this reason, such a pattern is not implicitly anchored.
1226 .P
1227 When a capturing subpattern is repeated, the value captured is the substring
1228 that matched the final iteration. For example, after
1229 .sp
1230 (tweedle[dume]{3}\es*)+
1231 .sp
1232 has matched "tweedledum tweedledee" the value of the captured substring is
1233 "tweedledee". However, if there are nested capturing subpatterns, the
1234 corresponding captured values may have been set in previous iterations. For
1235 example, after
1236 .sp
1237 /(a|(b))+/
1238 .sp
1239 matches "aba" the value of the second captured substring is "b".
1240 .
1241 .
1242 .\" HTML <a name="atomicgroup"></a>
1244 .rs
1245 .sp
1246 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1247 repetition, failure of what follows normally causes the repeated item to be
1248 re-evaluated to see if a different number of repeats allows the rest of the
1249 pattern to match. Sometimes it is useful to prevent this, either to change the
1250 nature of the match, or to cause it fail earlier than it otherwise might, when
1251 the author of the pattern knows there is no point in carrying on.
1252 .P
1253 Consider, for example, the pattern \ed+foo when applied to the subject line
1254 .sp
1255 123456bar
1256 .sp
1257 After matching all 6 digits and then failing to match "foo", the normal
1258 action of the matcher is to try again with only 5 digits matching the \ed+
1259 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1260 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1261 that once a subpattern has matched, it is not to be re-evaluated in this way.
1262 .P
1263 If we use atomic grouping for the previous example, the matcher gives up
1264 immediately on failing to match "foo" the first time. The notation is a kind of
1265 special parenthesis, starting with (?> as in this example:
1266 .sp
1267 (?>\ed+)foo
1268 .sp
1269 This kind of parenthesis "locks up" the part of the pattern it contains once
1270 it has matched, and a failure further into the pattern is prevented from
1271 backtracking into it. Backtracking past it to previous items, however, works as
1272 normal.
1273 .P
1274 An alternative description is that a subpattern of this type matches the string
1275 of characters that an identical standalone pattern would match, if anchored at
1276 the current point in the subject string.
1277 .P
1278 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1279 the above example can be thought of as a maximizing repeat that must swallow
1280 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1281 number of digits they match in order to make the rest of the pattern match,
1282 (?>\ed+) can only match an entire sequence of digits.
1283 .P
1284 Atomic groups in general can of course contain arbitrarily complicated
1285 subpatterns, and can be nested. However, when the subpattern for an atomic
1286 group is just a single repeated item, as in the example above, a simpler
1287 notation, called a "possessive quantifier" can be used. This consists of an
1288 additional + character following a quantifier. Using this notation, the
1289 previous example can be rewritten as
1290 .sp
1291 \ed++foo
1292 .sp
1293 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1294 option is ignored. They are a convenient notation for the simpler forms of
1295 atomic group. However, there is no difference in the meaning of a possessive
1296 quantifier and the equivalent atomic group, though there may be a performance
1297 difference; possessive quantifiers should be slightly faster.
1298 .P
1299 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1300 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1301 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1302 package, and PCRE copied it from there. It ultimately found its way into Perl
1303 at release 5.10.
1304 .P
1305 PCRE has an optimization that automatically "possessifies" certain simple
1306 pattern constructs. For example, the sequence A+B is treated as A++B because
1307 there is no point in backtracking into a sequence of A's when B must follow.
1308 .P
1309 When a pattern contains an unlimited repeat inside a subpattern that can itself
1310 be repeated an unlimited number of times, the use of an atomic group is the
1311 only way to avoid some failing matches taking a very long time indeed. The
1312 pattern
1313 .sp
1314 (\eD+|<\ed+>)*[!?]
1315 .sp
1316 matches an unlimited number of substrings that either consist of non-digits, or
1317 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1318 quickly. However, if it is applied to
1319 .sp
1320 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1321 .sp
1322 it takes a long time before reporting failure. This is because the string can
1323 be divided between the internal \eD+ repeat and the external * repeat in a
1324 large number of ways, and all have to be tried. (The example uses [!?] rather
1325 than a single character at the end, because both PCRE and Perl have an
1326 optimization that allows for fast failure when a single character is used. They
1327 remember the last single character that is required for a match, and fail early
1328 if it is not present in the string.) If the pattern is changed so that it uses
1329 an atomic group, like this:
1330 .sp
1331 ((?>\eD+)|<\ed+>)*[!?]
1332 .sp
1333 sequences of non-digits cannot be broken, and failure happens quickly.
1334 .
1335 .
1336 .\" HTML <a name="backreferences"></a>
1338 .rs
1339 .sp
1340 Outside a character class, a backslash followed by a digit greater than 0 (and
1341 possibly further digits) is a back reference to a capturing subpattern earlier
1342 (that is, to its left) in the pattern, provided there have been that many
1343 previous capturing left parentheses.
1344 .P
1345 However, if the decimal number following the backslash is less than 10, it is
1346 always taken as a back reference, and causes an error only if there are not
1347 that many capturing left parentheses in the entire pattern. In other words, the
1348 parentheses that are referenced need not be to the left of the reference for
1349 numbers less than 10. A "forward back reference" of this type can make sense
1350 when a repetition is involved and the subpattern to the right has participated
1351 in an earlier iteration.
1352 .P
1353 It is not possible to have a numerical "forward back reference" to a subpattern
1354 whose number is 10 or more using this syntax because a sequence such as \e50 is
1355 interpreted as a character defined in octal. See the subsection entitled
1356 "Non-printing characters"
1357 .\" HTML <a href="#digitsafterbackslash">
1358 .\" </a>
1359 above
1360 .\"
1361 for further details of the handling of digits following a backslash. There is
1362 no such problem when named parentheses are used. A back reference to any
1363 subpattern is possible using named parentheses (see below).
1364 .P
1365 Another way of avoiding the ambiguity inherent in the use of digits following a
1366 backslash is to use the \eg escape sequence, which is a feature introduced in
1367 Perl 5.10. This escape must be followed by a positive or a negative number,
1368 optionally enclosed in braces. These examples are all identical:
1369 .sp
1370 (ring), \e1
1371 (ring), \eg1
1372 (ring), \eg{1}
1373 .sp
1374 A positive number specifies an absolute reference without the ambiguity that is
1375 present in the older syntax. It is also useful when literal digits follow the
1376 reference. A negative number is a relative reference. Consider this example:
1377 .sp
1378 (abc(def)ghi)\eg{-1}
1379 .sp
1380 The sequence \eg{-1} is a reference to the most recently started capturing
1381 subpattern before \eg, that is, is it equivalent to \e2. Similarly, \eg{-2}
1382 would be equivalent to \e1. The use of relative references can be helpful in
1383 long patterns, and also in patterns that are created by joining together
1384 fragments that contain references within themselves.
1385 .P
1386 A back reference matches whatever actually matched the capturing subpattern in
1387 the current subject string, rather than anything matching the subpattern
1388 itself (see
1389 .\" HTML <a href="#subpatternsassubroutines">
1390 .\" </a>
1391 "Subpatterns as subroutines"
1392 .\"
1393 below for a way of doing that). So the pattern
1394 .sp
1395 (sens|respons)e and \e1ibility
1396 .sp
1397 matches "sense and sensibility" and "response and responsibility", but not
1398 "sense and responsibility". If caseful matching is in force at the time of the
1399 back reference, the case of letters is relevant. For example,
1400 .sp
1401 ((?i)rah)\es+\e1
1402 .sp
1403 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1404 capturing subpattern is matched caselessly.
1405 .P
1406 There are several different ways of writing back references to named
1407 subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1408 \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1409 back reference syntax, in which \eg can be used for both numeric and named
1410 references, is also supported. We could rewrite the above example in any of
1411 the following ways:
1412 .sp
1413 (?<p1>(?i)rah)\es+\ek<p1>
1414 (?'p1'(?i)rah)\es+\ek{p1}
1415 (?P<p1>(?i)rah)\es+(?P=p1)
1416 (?<p1>(?i)rah)\es+\eg{p1}
1417 .sp
1418 A subpattern that is referenced by name may appear in the pattern before or
1419 after the reference.
1420 .P
1421 There may be more than one back reference to the same subpattern. If a
1422 subpattern has not actually been used in a particular match, any back
1423 references to it always fail. For example, the pattern
1424 .sp
1425 (a|(bc))\e2
1426 .sp
1427 always fails if it starts to match "a" rather than "bc". Because there may be
1428 many capturing parentheses in a pattern, all digits following the backslash are
1429 taken as part of a potential back reference number. If the pattern continues
1430 with a digit character, some delimiter must be used to terminate the back
1431 reference. If the PCRE_EXTENDED option is set, this can be whitespace.
1432 Otherwise an empty comment (see
1433 .\" HTML <a href="#comments">
1434 .\" </a>
1435 "Comments"
1436 .\"
1437 below) can be used.
1438 .P
1439 A back reference that occurs inside the parentheses to which it refers fails
1440 when the subpattern is first used, so, for example, (a\e1) never matches.
1441 However, such references can be useful inside repeated subpatterns. For
1442 example, the pattern
1443 .sp
1444 (a|b\e1)+
1445 .sp
1446 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1447 the subpattern, the back reference matches the character string corresponding
1448 to the previous iteration. In order for this to work, the pattern must be such
1449 that the first iteration does not need to match the back reference. This can be
1450 done using alternation, as in the example above, or by a quantifier with a
1451 minimum of zero.
1452 .
1453 .
1454 .\" HTML <a name="bigassertions"></a>
1456 .rs
1457 .sp
1458 An assertion is a test on the characters following or preceding the current
1459 matching point that does not actually consume any characters. The simple
1460 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1461 .\" HTML <a href="#smallassertions">
1462 .\" </a>
1463 above.
1464 .\"
1465 .P
1466 More complicated assertions are coded as subpatterns. There are two kinds:
1467 those that look ahead of the current position in the subject string, and those
1468 that look behind it. An assertion subpattern is matched in the normal way,
1469 except that it does not cause the current matching position to be changed.
1470 .P
1471 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1472 because it makes no sense to assert the same thing several times. If any kind
1473 of assertion contains capturing subpatterns within it, these are counted for
1474 the purposes of numbering the capturing subpatterns in the whole pattern.
1475 However, substring capturing is carried out only for positive assertions,
1476 because it does not make sense for negative assertions.
1477 .
1478 .
1479 .SS "Lookahead assertions"
1480 .rs
1481 .sp
1482 Lookahead assertions start with (?= for positive assertions and (?! for
1483 negative assertions. For example,
1484 .sp
1485 \ew+(?=;)
1486 .sp
1487 matches a word followed by a semicolon, but does not include the semicolon in
1488 the match, and
1489 .sp
1490 foo(?!bar)
1491 .sp
1492 matches any occurrence of "foo" that is not followed by "bar". Note that the
1493 apparently similar pattern
1494 .sp
1495 (?!foo)bar
1496 .sp
1497 does not find an occurrence of "bar" that is preceded by something other than
1498 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1499 (?!foo) is always true when the next three characters are "bar". A
1500 lookbehind assertion is needed to achieve the other effect.
1501 .P
1502 If you want to force a matching failure at some point in a pattern, the most
1503 convenient way to do it is with (?!) because an empty string always matches, so
1504 an assertion that requires there not to be an empty string must always fail.
1505 .
1506 .
1507 .\" HTML <a name="lookbehind"></a>
1508 .SS "Lookbehind assertions"
1509 .rs
1510 .sp
1511 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1512 negative assertions. For example,
1513 .sp
1514 (?<!foo)bar
1515 .sp
1516 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1517 a lookbehind assertion are restricted such that all the strings it matches must
1518 have a fixed length. However, if there are several top-level alternatives, they
1519 do not all have to have the same fixed length. Thus
1520 .sp
1521 (?<=bullock|donkey)
1522 .sp
1523 is permitted, but
1524 .sp
1525 (?<!dogs?|cats?)
1526 .sp
1527 causes an error at compile time. Branches that match different length strings
1528 are permitted only at the top level of a lookbehind assertion. This is an
1529 extension compared with Perl (at least for 5.8), which requires all branches to
1530 match the same length of string. An assertion such as
1531 .sp
1532 (?<=ab(c|de))
1533 .sp
1534 is not permitted, because its single top-level branch can match two different
1535 lengths, but it is acceptable if rewritten to use two top-level branches:
1536 .sp
1537 (?<=abc|abde)
1538 .sp
1539 In some cases, the Perl 5.10 escape sequence \eK
1540 .\" HTML <a href="#resetmatchstart">
1541 .\" </a>
1542 (see above)
1543 .\"
1544 can be used instead of a lookbehind assertion; this is not restricted to a
1545 fixed-length.
1546 .P
1547 The implementation of lookbehind assertions is, for each alternative, to
1548 temporarily move the current position back by the fixed length and then try to
1549 match. If there are insufficient characters before the current position, the
1550 assertion fails.
1551 .P
1552 PCRE does not allow the \eC escape (which matches a single byte in UTF-8 mode)
1553 to appear in lookbehind assertions, because it makes it impossible to calculate
1554 the length of the lookbehind. The \eX and \eR escapes, which can match
1555 different numbers of bytes, are also not permitted.
1556 .P
1557 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1558 specify efficient matching at the end of the subject string. Consider a simple
1559 pattern such as
1560 .sp
1561 abcd$
1562 .sp
1563 when applied to a long string that does not match. Because matching proceeds
1564 from left to right, PCRE will look for each "a" in the subject and then see if
1565 what follows matches the rest of the pattern. If the pattern is specified as
1566 .sp
1567 ^.*abcd$
1568 .sp
1569 the initial .* matches the entire string at first, but when this fails (because
1570 there is no following "a"), it backtracks to match all but the last character,
1571 then all but the last two characters, and so on. Once again the search for "a"
1572 covers the entire string, from right to left, so we are no better off. However,
1573 if the pattern is written as
1574 .sp
1575 ^.*+(?<=abcd)
1576 .sp
1577 there can be no backtracking for the .*+ item; it can match only the entire
1578 string. The subsequent lookbehind assertion does a single test on the last four
1579 characters. If it fails, the match fails immediately. For long strings, this
1580 approach makes a significant difference to the processing time.
1581 .
1582 .
1583 .SS "Using multiple assertions"
1584 .rs
1585 .sp
1586 Several assertions (of any sort) may occur in succession. For example,
1587 .sp
1588 (?<=\ed{3})(?<!999)foo
1589 .sp
1590 matches "foo" preceded by three digits that are not "999". Notice that each of
1591 the assertions is applied independently at the same point in the subject
1592 string. First there is a check that the previous three characters are all
1593 digits, and then there is a check that the same three characters are not "999".
1594 This pattern does \fInot\fP match "foo" preceded by six characters, the first
1595 of which are digits and the last three of which are not "999". For example, it
1596 doesn't match "123abcfoo". A pattern to do that is
1597 .sp
1598 (?<=\ed{3}...)(?<!999)foo
1599 .sp
1600 This time the first assertion looks at the preceding six characters, checking
1601 that the first three are digits, and then the second assertion checks that the
1602 preceding three characters are not "999".
1603 .P
1604 Assertions can be nested in any combination. For example,
1605 .sp
1606 (?<=(?<!foo)bar)baz
1607 .sp
1608 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1609 preceded by "foo", while
1610 .sp
1611 (?<=\ed{3}(?!999)...)foo
1612 .sp
1613 is another pattern that matches "foo" preceded by three digits and any three
1614 characters that are not "999".
1615 .
1616 .
1617 .\" HTML <a name="conditions"></a>
1619 .rs
1620 .sp
1621 It is possible to cause the matching process to obey a subpattern
1622 conditionally or to choose between two alternative subpatterns, depending on
1623 the result of an assertion, or whether a previous capturing subpattern matched
1624 or not. The two possible forms of conditional subpattern are
1625 .sp
1626 (?(condition)yes-pattern)
1627 (?(condition)yes-pattern|no-pattern)
1628 .sp
1629 If the condition is satisfied, the yes-pattern is used; otherwise the
1630 no-pattern (if present) is used. If there are more than two alternatives in the
1631 subpattern, a compile-time error occurs.
1632 .P
1633 There are four kinds of condition: references to subpatterns, references to
1634 recursion, a pseudo-condition called DEFINE, and assertions.
1635 .
1636 .SS "Checking for a used subpattern by number"
1637 .rs
1638 .sp
1639 If the text between the parentheses consists of a sequence of digits, the
1640 condition is true if the capturing subpattern of that number has previously
1641 matched. An alternative notation is to precede the digits with a plus or minus
1642 sign. In this case, the subpattern number is relative rather than absolute.
1643 The most recently opened parentheses can be referenced by (?(-1), the next most
1644 recent by (?(-2), and so on. In looping constructs it can also make sense to
1645 refer to subsequent groups with constructs such as (?(+2).
1646 .P
1647 Consider the following pattern, which contains non-significant white space to
1648 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
1649 three parts for ease of discussion:
1650 .sp
1651 ( \e( )? [^()]+ (?(1) \e) )
1652 .sp
1653 The first part matches an optional opening parenthesis, and if that
1654 character is present, sets it as the first captured substring. The second part
1655 matches one or more characters that are not parentheses. The third part is a
1656 conditional subpattern that tests whether the first set of parentheses matched
1657 or not. If they did, that is, if subject started with an opening parenthesis,
1658 the condition is true, and so the yes-pattern is executed and a closing
1659 parenthesis is required. Otherwise, since no-pattern is not present, the
1660 subpattern matches nothing. In other words, this pattern matches a sequence of
1661 non-parentheses, optionally enclosed in parentheses.
1662 .P
1663 If you were embedding this pattern in a larger one, you could use a relative
1664 reference:
1665 .sp
1666 ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
1667 .sp
1668 This makes the fragment independent of the parentheses in the larger pattern.
1669 .
1670 .SS "Checking for a used subpattern by name"
1671 .rs
1672 .sp
1673 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
1674 subpattern by name. For compatibility with earlier versions of PCRE, which had
1675 this facility before Perl, the syntax (?(name)...) is also recognized. However,
1676 there is a possible ambiguity with this syntax, because subpattern names may
1677 consist entirely of digits. PCRE looks first for a named subpattern; if it
1678 cannot find one and the name consists entirely of digits, PCRE looks for a
1679 subpattern of that number, which must be greater than zero. Using subpattern
1680 names that consist entirely of digits is not recommended.
1681 .P
1682 Rewriting the above example to use a named subpattern gives this:
1683 .sp
1684 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
1685 .sp
1686 .
1687 .SS "Checking for pattern recursion"
1688 .rs
1689 .sp
1690 If the condition is the string (R), and there is no subpattern with the name R,
1691 the condition is true if a recursive call to the whole pattern or any
1692 subpattern has been made. If digits or a name preceded by ampersand follow the
1693 letter R, for example:
1694 .sp
1695 (?(R3)...) or (?(R&name)...)
1696 .sp
1697 the condition is true if the most recent recursion is into the subpattern whose
1698 number or name is given. This condition does not check the entire recursion
1699 stack.
1700 .P
1701 At "top level", all these recursion test conditions are false. Recursive
1702 patterns are described below.
1703 .
1704 .SS "Defining subpatterns for use by reference only"
1705 .rs
1706 .sp
1707 If the condition is the string (DEFINE), and there is no subpattern with the
1708 name DEFINE, the condition is always false. In this case, there may be only one
1709 alternative in the subpattern. It is always skipped if control reaches this
1710 point in the pattern; the idea of DEFINE is that it can be used to define
1711 "subroutines" that can be referenced from elsewhere. (The use of "subroutines"
1712 is described below.) For example, a pattern to match an IPv4 address could be
1713 written like this (ignore whitespace and line breaks):
1714 .sp
1715 (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
1716 \eb (?&byte) (\e.(?&byte)){3} \eb
1717 .sp
1718 The first part of the pattern is a DEFINE group inside which a another group
1719 named "byte" is defined. This matches an individual component of an IPv4
1720 address (a number less than 256). When matching takes place, this part of the
1721 pattern is skipped because DEFINE acts like a false condition.
1722 .P
1723 The rest of the pattern uses references to the named group to match the four
1724 dot-separated components of an IPv4 address, insisting on a word boundary at
1725 each end.
1726 .
1727 .SS "Assertion conditions"
1728 .rs
1729 .sp
1730 If the condition is not in any of the above formats, it must be an assertion.
1731 This may be a positive or negative lookahead or lookbehind assertion. Consider
1732 this pattern, again containing non-significant white space, and with the two
1733 alternatives on the second line:
1734 .sp
1735 (?(?=[^a-z]*[a-z])
1736 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
1737 .sp
1738 The condition is a positive lookahead assertion that matches an optional
1739 sequence of non-letters followed by a letter. In other words, it tests for the
1740 presence of at least one letter in the subject. If a letter is found, the
1741 subject is matched against the first alternative; otherwise it is matched
1742 against the second. This pattern matches strings in one of the two forms
1743 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
1744 .
1745 .
1746 .\" HTML <a name="comments"></a>
1748 .rs
1749 .sp
1750 The sequence (?# marks the start of a comment that continues up to the next
1751 closing parenthesis. Nested parentheses are not permitted. The characters
1752 that make up a comment play no part in the pattern matching at all.
1753 .P
1754 If the PCRE_EXTENDED option is set, an unescaped # character outside a
1755 character class introduces a comment that continues to immediately after the
1756 next newline in the pattern.
1757 .
1758 .
1759 .\" HTML <a name="recursion"></a>
1761 .rs
1762 .sp
1763 Consider the problem of matching a string in parentheses, allowing for
1764 unlimited nested parentheses. Without the use of recursion, the best that can
1765 be done is to use a pattern that matches up to some fixed depth of nesting. It
1766 is not possible to handle an arbitrary nesting depth.
1767 .P
1768 For some time, Perl has provided a facility that allows regular expressions to
1769 recurse (amongst other things). It does this by interpolating Perl code in the
1770 expression at run time, and the code can refer to the expression itself. A Perl
1771 pattern using code interpolation to solve the parentheses problem can be
1772 created like this:
1773 .sp
1774 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
1775 .sp
1776 The (?p{...}) item interpolates Perl code at run time, and in this case refers
1777 recursively to the pattern in which it appears.
1778 .P
1779 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
1780 supports special syntax for recursion of the entire pattern, and also for
1781 individual subpattern recursion. After its introduction in PCRE and Python,
1782 this kind of recursion was introduced into Perl at release 5.10.
1783 .P
1784 A special item that consists of (? followed by a number greater than zero and a
1785 closing parenthesis is a recursive call of the subpattern of the given number,
1786 provided that it occurs inside that subpattern. (If not, it is a "subroutine"
1787 call, which is described in the next section.) The special item (?R) or (?0) is
1788 a recursive call of the entire regular expression.
1789 .P
1790 In PCRE (like Python, but unlike Perl), a recursive subpattern call is always
1791 treated as an atomic group. That is, once it has matched some of the subject
1792 string, it is never re-entered, even if it contains untried alternatives and
1793 there is a subsequent matching failure.
1794 .P
1795 This PCRE pattern solves the nested parentheses problem (assume the
1796 PCRE_EXTENDED option is set so that white space is ignored):
1797 .sp
1798 \e( ( (?>[^()]+) | (?R) )* \e)
1799 .sp
1800 First it matches an opening parenthesis. Then it matches any number of
1801 substrings which can either be a sequence of non-parentheses, or a recursive
1802 match of the pattern itself (that is, a correctly parenthesized substring).
1803 Finally there is a closing parenthesis.
1804 .P
1805 If this were part of a larger pattern, you would not want to recurse the entire
1806 pattern, so instead you could use this:
1807 .sp
1808 ( \e( ( (?>[^()]+) | (?1) )* \e) )
1809 .sp
1810 We have put the pattern into parentheses, and caused the recursion to refer to
1811 them instead of the whole pattern.
1812 .P
1813 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
1814 is made easier by the use of relative references. (A Perl 5.10 feature.)
1815 Instead of (?1) in the pattern above you can write (?-2) to refer to the second
1816 most recently opened parentheses preceding the recursion. In other words, a
1817 negative number counts capturing parentheses leftwards from the point at which
1818 it is encountered.
1819 .P
1820 It is also possible to refer to subsequently opened parentheses, by writing
1821 references such as (?+2). However, these cannot be recursive because the
1822 reference is not inside the parentheses that are referenced. They are always
1823 "subroutine" calls, as described in the next section.
1824 .P
1825 An alternative approach is to use named parentheses instead. The Perl syntax
1826 for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
1827 could rewrite the above example as follows:
1828 .sp
1829 (?<pn> \e( ( (?>[^()]+) | (?&pn) )* \e) )
1830 .sp
1831 If there is more than one subpattern with the same name, the earliest one is
1832 used.
1833 .P
1834 This particular example pattern that we have been looking at contains nested
1835 unlimited repeats, and so the use of atomic grouping for matching strings of
1836 non-parentheses is important when applying the pattern to strings that do not
1837 match. For example, when this pattern is applied to
1838 .sp
1839 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
1840 .sp
1841 it yields "no match" quickly. However, if atomic grouping is not used,
1842 the match runs for a very long time indeed because there are so many different
1843 ways the + and * repeats can carve up the subject, and all have to be tested
1844 before failure can be reported.
1845 .P
1846 At the end of a match, the values set for any capturing subpatterns are those
1847 from the outermost level of the recursion at which the subpattern value is set.
1848 If you want to obtain intermediate values, a callout function can be used (see
1849 below and the
1850 .\" HREF
1851 \fBpcrecallout\fP
1852 .\"
1853 documentation). If the pattern above is matched against
1854 .sp
1855 (ab(cd)ef)
1856 .sp
1857 the value for the capturing parentheses is "ef", which is the last value taken
1858 on at the top level. If additional parentheses are added, giving
1859 .sp
1860 \e( ( ( (?>[^()]+) | (?R) )* ) \e)
1861 ^ ^
1862 ^ ^
1863 .sp
1864 the string they capture is "ab(cd)ef", the contents of the top level
1865 parentheses. If there are more than 15 capturing parentheses in a pattern, PCRE
1866 has to obtain extra memory to store data during a recursion, which it does by
1867 using \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no
1868 memory can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
1869 .P
1870 Do not confuse the (?R) item with the condition (R), which tests for recursion.
1871 Consider this pattern, which matches text in angle brackets, allowing for
1872 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
1873 recursing), whereas any characters are permitted at the outer level.
1874 .sp
1875 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
1876 .sp
1877 In this pattern, (?(R) is the start of a conditional subpattern, with two
1878 different alternatives for the recursive and non-recursive cases. The (?R) item
1879 is the actual recursive call.
1880 .
1881 .
1882 .\" HTML <a name="subpatternsassubroutines"></a>
1884 .rs
1885 .sp
1886 If the syntax for a recursive subpattern reference (either by number or by
1887 name) is used outside the parentheses to which it refers, it operates like a
1888 subroutine in a programming language. The "called" subpattern may be defined
1889 before or after the reference. A numbered reference can be absolute or
1890 relative, as in these examples:
1891 .sp
1892 (...(absolute)...)...(?2)...
1893 (...(relative)...)...(?-1)...
1894 (...(?+1)...(relative)...
1895 .sp
1896 An earlier example pointed out that the pattern
1897 .sp
1898 (sens|respons)e and \e1ibility
1899 .sp
1900 matches "sense and sensibility" and "response and responsibility", but not
1901 "sense and responsibility". If instead the pattern
1902 .sp
1903 (sens|respons)e and (?1)ibility
1904 .sp
1905 is used, it does match "sense and responsibility" as well as the other two
1906 strings. Another example is given in the discussion of DEFINE above.
1907 .P
1908 Like recursive subpatterns, a "subroutine" call is always treated as an atomic
1909 group. That is, once it has matched some of the subject string, it is never
1910 re-entered, even if it contains untried alternatives and there is a subsequent
1911 matching failure.
1912 .P
1913 When a subpattern is used as a subroutine, processing options such as
1914 case-independence are fixed when the subpattern is defined. They cannot be
1915 changed for different calls. For example, consider this pattern:
1916 .sp
1917 (abc)(?i:(?-1))
1918 .sp
1919 It matches "abcabc". It does not match "abcABC" because the change of
1920 processing option does not affect the called subpattern.
1921 .
1922 .
1924 .rs
1925 .sp
1926 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
1927 code to be obeyed in the middle of matching a regular expression. This makes it
1928 possible, amongst other things, to extract different substrings that match the
1929 same pair of parentheses when there is a repetition.
1930 .P
1931 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
1932 code. The feature is called "callout". The caller of PCRE provides an external
1933 function by putting its entry point in the global variable \fIpcre_callout\fP.
1934 By default, this variable contains NULL, which disables all calling out.
1935 .P
1936 Within a regular expression, (?C) indicates the points at which the external
1937 function is to be called. If you want to identify different callout points, you
1938 can put a number less than 256 after the letter C. The default value is zero.
1939 For example, this pattern has two callout points:
1940 .sp
1941 (?C1)abc(?C2)def
1942 .sp
1943 If the PCRE_AUTO_CALLOUT flag is passed to \fBpcre_compile()\fP, callouts are
1944 automatically installed before each item in the pattern. They are all numbered
1945 255.
1946 .P
1947 During matching, when PCRE reaches a callout point (and \fIpcre_callout\fP is
1948 set), the external function is called. It is provided with the number of the
1949 callout, the position in the pattern, and, optionally, one item of data
1950 originally supplied by the caller of \fBpcre_exec()\fP. The callout function
1951 may cause matching to proceed, to backtrack, or to fail altogether. A complete
1952 description of the interface to the callout function is given in the
1953 .\" HREF
1954 \fBpcrecallout\fP
1955 .\"
1956 documentation.
1957 .
1958 .
1959 .SH "SEE ALSO"
1960 .rs
1961 .sp
1962 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3), \fBpcre\fP(3).
1963 .
1964 .
1966 .rs
1967 .sp
1968 .nf
1969 Philip Hazel
1970 University Computing Service
1971 Cambridge CB2 3QH, England.
1972 .fi
1973 .
1974 .
1976 .rs
1977 .sp
1978 .nf
1979 Last updated: 19 June 2007
1980 Copyright (c) 1997-2007 University of Cambridge.
1981 .fi


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

  ViewVC Help
Powered by ViewVC 1.1.5