ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 231 - (show annotations)
Tue Sep 11 11:15:33 2007 UTC (13 years, 5 months ago) by ph10
File size: 88821 byte(s)
Error occurred while calculating annotation data.
Add facility to make \R match only CR, LF, or CRLF.
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. Perl's regular expressions are described in its own documentation, and
13 regular expressions in general are covered in a number of books, some of which
14 have copious examples. Jeffrey Friedl's "Mastering Regular Expressions",
15 published by O'Reilly, covers regular expressions in great detail. This
16 description of PCRE's regular expressions is intended as reference material.
17 .P
18 The original operation of PCRE was on strings of one-byte characters. However,
19 there is now also support for UTF-8 character strings. To use this, you must
20 build PCRE to include UTF-8 support, and then call \fBpcre_compile()\fP with
21 the PCRE_UTF8 option. How this affects pattern matching is mentioned in several
22 places below. There is also a summary of UTF-8 features in the
23 .\" HTML <a href="pcre.html#utf8support">
24 .\" </a>
25 section on UTF-8 support
26 .\"
27 in the main
28 .\" HREF
29 \fBpcre\fP
30 .\"
31 page.
32 .P
33 The remainder of this document discusses the patterns that are supported by
34 PCRE when its main matching function, \fBpcre_exec()\fP, is used.
35 From release 6.0, PCRE offers a second matching function,
36 \fBpcre_dfa_exec()\fP, which matches using a different algorithm that is not
37 Perl-compatible. Some of the features discussed below are not available when
38 \fBpcre_dfa_exec()\fP is used. The advantages and disadvantages of the
39 alternative function, and how it differs from the normal function, are
40 discussed in the
41 .\" HREF
42 \fBpcrematching\fP
43 .\"
44 page.
45 .
46 .
48 .rs
49 .sp
50 PCRE supports five different conventions for indicating line breaks in
51 strings: a single CR (carriage return) character, a single LF (linefeed)
52 character, the two-character sequence CRLF, any of the three preceding, or any
53 Unicode newline sequence. The
54 .\" HREF
55 \fBpcreapi\fP
56 .\"
57 page has
58 .\" HTML <a href="pcreapi.html#newlines">
59 .\" </a>
60 further discussion
61 .\"
62 about newlines, and shows how to set the newline convention in the
63 \fIoptions\fP arguments for the compiling and matching functions.
64 .P
65 It is also possible to specify a newline convention by starting a pattern
66 string with one of the following five sequences:
67 .sp
68 (*CR) carriage return
69 (*LF) linefeed
70 (*CRLF) carriage return, followed by linefeed
71 (*ANYCRLF) any of the three above
72 (*ANY) all Unicode newline sequences
73 .sp
74 These override the default and the options given to \fBpcre_compile()\fP. For
75 example, on a Unix system where LF is the default newline sequence, the pattern
76 .sp
77 (*CR)a.b
78 .sp
79 changes the convention to CR. That pattern matches "a\enb" because LF is no
80 longer a newline. Note that these special settings, which are not
81 Perl-compatible, are recognized only at the very start of a pattern, and that
82 they must be in upper case. If more than one of them is present, the last one
83 is used.
84 .P
85 The newline convention does not affect what the \eR escape sequence matches. By
86 default, this is any Unicode newline sequence, for Perl compatibility. However,
87 this can be changed; see the description of \eR in the section entitled
88 .\" HTML <a href="#newlineseq">
89 .\" </a>
90 "Newline sequences"
91 .\"
92 below.
93 .
94 .
96 .rs
97 .sp
98 A regular expression is a pattern that is matched against a subject string from
99 left to right. Most characters stand for themselves in a pattern, and match the
100 corresponding characters in the subject. As a trivial example, the pattern
101 .sp
102 The quick brown fox
103 .sp
104 matches a portion of a subject string that is identical to itself. When
105 caseless matching is specified (the PCRE_CASELESS option), letters are matched
106 independently of case. In UTF-8 mode, PCRE always understands the concept of
107 case for characters whose values are less than 128, so caseless matching is
108 always possible. For characters with higher values, the concept of case is
109 supported if PCRE is compiled with Unicode property support, but not otherwise.
110 If you want to use caseless matching for characters 128 and above, you must
111 ensure that PCRE is compiled with Unicode property support as well as with
112 UTF-8 support.
113 .P
114 The power of regular expressions comes from the ability to include alternatives
115 and repetitions in the pattern. These are encoded in the pattern by the use of
116 \fImetacharacters\fP, which do not stand for themselves but instead are
117 interpreted in some special way.
118 .P
119 There are two different sets of metacharacters: those that are recognized
120 anywhere in the pattern except within square brackets, and those that are
121 recognized within square brackets. Outside square brackets, the metacharacters
122 are as follows:
123 .sp
124 \e general escape character with several uses
125 ^ assert start of string (or line, in multiline mode)
126 $ assert end of string (or line, in multiline mode)
127 . match any character except newline (by default)
128 [ start character class definition
129 | start of alternative branch
130 ( start subpattern
131 ) end subpattern
132 ? extends the meaning of (
133 also 0 or 1 quantifier
134 also quantifier minimizer
135 * 0 or more quantifier
136 + 1 or more quantifier
137 also "possessive quantifier"
138 { start min/max quantifier
139 .sp
140 Part of a pattern that is in square brackets is called a "character class". In
141 a character class the only metacharacters are:
142 .sp
143 \e general escape character
144 ^ negate the class, but only if the first character
145 - indicates character range
146 .\" JOIN
147 [ POSIX character class (only if followed by POSIX
148 syntax)
149 ] terminates the character class
150 .sp
151 The following sections describe the use of each of the metacharacters.
152 .
153 .
155 .rs
156 .sp
157 The backslash character has several uses. Firstly, if it is followed by a
158 non-alphanumeric character, it takes away any special meaning that character
159 may have. This use of backslash as an escape character applies both inside and
160 outside character classes.
161 .P
162 For example, if you want to match a * character, you write \e* in the pattern.
163 This escaping action applies whether or not the following character would
164 otherwise be interpreted as a metacharacter, so it is always safe to precede a
165 non-alphanumeric with backslash to specify that it stands for itself. In
166 particular, if you want to match a backslash, you write \e\e.
167 .P
168 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
169 pattern (other than in a character class) and characters between a # outside
170 a character class and the next newline are ignored. An escaping backslash can
171 be used to include a whitespace or # character as part of the pattern.
172 .P
173 If you want to remove the special meaning from a sequence of characters, you
174 can do so by putting them between \eQ and \eE. This is different from Perl in
175 that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
176 Perl, $ and @ cause variable interpolation. Note the following examples:
177 .sp
178 Pattern PCRE matches Perl matches
179 .sp
180 .\" JOIN
181 \eQabc$xyz\eE abc$xyz abc followed by the
182 contents of $xyz
183 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
184 \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
185 .sp
186 The \eQ...\eE sequence is recognized both inside and outside character classes.
187 .
188 .
189 .\" HTML <a name="digitsafterbackslash"></a>
190 .SS "Non-printing characters"
191 .rs
192 .sp
193 A second use of backslash provides a way of encoding non-printing characters
194 in patterns in a visible manner. There is no restriction on the appearance of
195 non-printing characters, apart from the binary zero that terminates a pattern,
196 but when a pattern is being prepared by text editing, it is usually easier to
197 use one of the following escape sequences than the binary character it
198 represents:
199 .sp
200 \ea alarm, that is, the BEL character (hex 07)
201 \ecx "control-x", where x is any character
202 \ee escape (hex 1B)
203 \ef formfeed (hex 0C)
204 \en linefeed (hex 0A)
205 \er carriage return (hex 0D)
206 \et tab (hex 09)
207 \eddd character with octal code ddd, or backreference
208 \exhh character with hex code hh
209 \ex{hhh..} character with hex code hhh..
210 .sp
211 The precise effect of \ecx is as follows: if x is a lower case letter, it
212 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
213 Thus \ecz becomes hex 1A, but \ec{ becomes hex 3B, while \ec; becomes hex
214 7B.
215 .P
216 After \ex, from zero to two hexadecimal digits are read (letters can be in
217 upper or lower case). Any number of hexadecimal digits may appear between \ex{
218 and }, but the value of the character code must be less than 256 in non-UTF-8
219 mode, and less than 2**31 in UTF-8 mode. That is, the maximum value in
220 hexadecimal is 7FFFFFFF. Note that this is bigger than the largest Unicode code
221 point, which is 10FFFF.
222 .P
223 If characters other than hexadecimal digits appear between \ex{ and }, or if
224 there is no terminating }, this form of escape is not recognized. Instead, the
225 initial \ex will be interpreted as a basic hexadecimal escape, with no
226 following digits, giving a character whose value is zero.
227 .P
228 Characters whose value is less than 256 can be defined by either of the two
229 syntaxes for \ex. There is no difference in the way they are handled. For
230 example, \exdc is exactly the same as \ex{dc}.
231 .P
232 After \e0 up to two further octal digits are read. If there are fewer than two
233 digits, just those that are present are used. Thus the sequence \e0\ex\e07
234 specifies two binary zeros followed by a BEL character (code value 7). Make
235 sure you supply two digits after the initial zero if the pattern character that
236 follows is itself an octal digit.
237 .P
238 The handling of a backslash followed by a digit other than 0 is complicated.
239 Outside a character class, PCRE reads it and any following digits as a decimal
240 number. If the number is less than 10, or if there have been at least that many
241 previous capturing left parentheses in the expression, the entire sequence is
242 taken as a \fIback reference\fP. A description of how this works is given
243 .\" HTML <a href="#backreferences">
244 .\" </a>
245 later,
246 .\"
247 following the discussion of
248 .\" HTML <a href="#subpattern">
249 .\" </a>
250 parenthesized subpatterns.
251 .\"
252 .P
253 Inside a character class, or if the decimal number is greater than 9 and there
254 have not been that many capturing subpatterns, PCRE re-reads up to three octal
255 digits following the backslash, and uses them to generate a data character. Any
256 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
257 character specified in octal must be less than \e400. In UTF-8 mode, values up
258 to \e777 are permitted. For example:
259 .sp
260 \e040 is another way of writing a space
261 .\" JOIN
262 \e40 is the same, provided there are fewer than 40
263 previous capturing subpatterns
264 \e7 is always a back reference
265 .\" JOIN
266 \e11 might be a back reference, or another way of
267 writing a tab
268 \e011 is always a tab
269 \e0113 is a tab followed by the character "3"
270 .\" JOIN
271 \e113 might be a back reference, otherwise the
272 character with octal code 113
273 .\" JOIN
274 \e377 might be a back reference, otherwise
275 the byte consisting entirely of 1 bits
276 .\" JOIN
277 \e81 is either a back reference, or a binary zero
278 followed by the two characters "8" and "1"
279 .sp
280 Note that octal values of 100 or greater must not be introduced by a leading
281 zero, because no more than three octal digits are ever read.
282 .P
283 All the sequences that define a single character value can be used both inside
284 and outside character classes. In addition, inside a character class, the
285 sequence \eb is interpreted as the backspace character (hex 08), and the
286 sequences \eR and \eX are interpreted as the characters "R" and "X",
287 respectively. Outside a character class, these sequences have different
288 meanings
289 .\" HTML <a href="#uniextseq">
290 .\" </a>
291 (see below).
292 .\"
293 .
294 .
295 .SS "Absolute and relative back references"
296 .rs
297 .sp
298 The sequence \eg followed by an unsigned or a negative number, optionally
299 enclosed in braces, is an absolute or relative back reference. A named back
300 reference can be coded as \eg{name}. Back references are discussed
301 .\" HTML <a href="#backreferences">
302 .\" </a>
303 later,
304 .\"
305 following the discussion of
306 .\" HTML <a href="#subpattern">
307 .\" </a>
308 parenthesized subpatterns.
309 .\"
310 .
311 .
312 .SS "Generic character types"
313 .rs
314 .sp
315 Another use of backslash is for specifying generic character types. The
316 following are always recognized:
317 .sp
318 \ed any decimal digit
319 \eD any character that is not a decimal digit
320 \eh any horizontal whitespace character
321 \eH any character that is not a horizontal whitespace character
322 \es any whitespace character
323 \eS any character that is not a whitespace character
324 \ev any vertical whitespace character
325 \eV any character that is not a vertical whitespace character
326 \ew any "word" character
327 \eW any "non-word" character
328 .sp
329 Each pair of escape sequences partitions the complete set of characters into
330 two disjoint sets. Any given character matches one, and only one, of each pair.
331 .P
332 These character type sequences can appear both inside and outside character
333 classes. They each match one character of the appropriate type. If the current
334 matching point is at the end of the subject string, all of them fail, since
335 there is no character to match.
336 .P
337 For compatibility with Perl, \es does not match the VT character (code 11).
338 This makes it different from the the POSIX "space" class. The \es characters
339 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
340 included in a Perl script, \es may match the VT character. In PCRE, it never
341 does.
342 .P
343 In UTF-8 mode, characters with values greater than 128 never match \ed, \es, or
344 \ew, and always match \eD, \eS, and \eW. This is true even when Unicode
345 character property support is available. These sequences retain their original
346 meanings from before UTF-8 support was available, mainly for efficiency
347 reasons.
348 .P
349 The sequences \eh, \eH, \ev, and \eV are Perl 5.10 features. In contrast to the
350 other sequences, these do match certain high-valued codepoints in UTF-8 mode.
351 The horizontal space characters are:
352 .sp
353 U+0009 Horizontal tab
354 U+0020 Space
355 U+00A0 Non-break space
356 U+1680 Ogham space mark
357 U+180E Mongolian vowel separator
358 U+2000 En quad
359 U+2001 Em quad
360 U+2002 En space
361 U+2003 Em space
362 U+2004 Three-per-em space
363 U+2005 Four-per-em space
364 U+2006 Six-per-em space
365 U+2007 Figure space
366 U+2008 Punctuation space
367 U+2009 Thin space
368 U+200A Hair space
369 U+202F Narrow no-break space
370 U+205F Medium mathematical space
371 U+3000 Ideographic space
372 .sp
373 The vertical space characters are:
374 .sp
375 U+000A Linefeed
376 U+000B Vertical tab
377 U+000C Formfeed
378 U+000D Carriage return
379 U+0085 Next line
380 U+2028 Line separator
381 U+2029 Paragraph separator
382 .P
383 A "word" character is an underscore or any character less than 256 that is a
384 letter or digit. The definition of letters and digits is controlled by PCRE's
385 low-valued character tables, and may vary if locale-specific matching is taking
386 place (see
387 .\" HTML <a href="pcreapi.html#localesupport">
388 .\" </a>
389 "Locale support"
390 .\"
391 in the
392 .\" HREF
393 \fBpcreapi\fP
394 .\"
395 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
396 or "french" in Windows, some character codes greater than 128 are used for
397 accented letters, and these are matched by \ew. The use of locales with Unicode
398 is discouraged.
399 .
400 .
401 .\" HTML <a name="newlineseq"></a>
402 .SS "Newline sequences"
403 .rs
404 .sp
405 Outside a character class, by default, the escape sequence \eR matches any
406 Unicode newline sequence. This is a Perl 5.10 feature. In non-UTF-8 mode \eR is
407 equivalent to the following:
408 .sp
409 (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
410 .sp
411 This is an example of an "atomic group", details of which are given
412 .\" HTML <a href="#atomicgroup">
413 .\" </a>
414 below.
415 .\"
416 This particular group matches either the two-character sequence CR followed by
417 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
418 U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
419 line, U+0085). The two-character sequence is treated as a single unit that
420 cannot be split.
421 .P
422 In UTF-8 mode, two additional characters whose codepoints are greater than 255
423 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
424 Unicode character property support is not needed for these characters to be
425 recognized.
426 .P
427 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
428 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
429 either at compile time or when the pattern is matched. This can be made the
430 default when PCRE is built; if this is the case, the other behaviour can be
431 requested via the PCRE_BSR_UNICODE option. It is also possible to specify these
432 settings by starting a pattern string with one of the following sequences:
433 .sp
434 (*BSR_ANYCRLF) CR, LF, or CRLF only
435 (*BSR_UNICODE) any Unicode newline sequence
436 .sp
437 These override the default and the options given to \fBpcre_compile()\fP, but
438 they can be overridden by options given to \fBpcre_exec()\fP. Note that these
439 special settings, which are not Perl-compatible, are recognized only at the
440 very start of a pattern, and that they must be in upper case. If more than one
441 of them is present, the last one is used.
442 .P
443 Inside a character class, \eR matches the letter "R".
444 .
445 .
446 .\" HTML <a name="uniextseq"></a>
447 .SS Unicode character properties
448 .rs
449 .sp
450 When PCRE is built with Unicode character property support, three additional
451 escape sequences that match characters with specific properties are available.
452 When not in UTF-8 mode, these sequences are of course limited to testing
453 characters whose codepoints are less than 256, but they do work in this mode.
454 The extra escape sequences are:
455 .sp
456 \ep{\fIxx\fP} a character with the \fIxx\fP property
457 \eP{\fIxx\fP} a character without the \fIxx\fP property
458 \eX an extended Unicode sequence
459 .sp
460 The property names represented by \fIxx\fP above are limited to the Unicode
461 script names, the general category properties, and "Any", which matches any
462 character (including newline). Other properties such as "InMusicalSymbols" are
463 not currently supported by PCRE. Note that \eP{Any} does not match any
464 characters, so always causes a match failure.
465 .P
466 Sets of Unicode characters are defined as belonging to certain scripts. A
467 character from one of these sets can be matched using a script name. For
468 example:
469 .sp
470 \ep{Greek}
471 \eP{Han}
472 .sp
473 Those that are not part of an identified script are lumped together as
474 "Common". The current list of scripts is:
475 .P
476 Arabic,
477 Armenian,
478 Balinese,
479 Bengali,
480 Bopomofo,
481 Braille,
482 Buginese,
483 Buhid,
484 Canadian_Aboriginal,
485 Cherokee,
486 Common,
487 Coptic,
488 Cuneiform,
489 Cypriot,
490 Cyrillic,
491 Deseret,
492 Devanagari,
493 Ethiopic,
494 Georgian,
495 Glagolitic,
496 Gothic,
497 Greek,
498 Gujarati,
499 Gurmukhi,
500 Han,
501 Hangul,
502 Hanunoo,
503 Hebrew,
504 Hiragana,
505 Inherited,
506 Kannada,
507 Katakana,
508 Kharoshthi,
509 Khmer,
510 Lao,
511 Latin,
512 Limbu,
513 Linear_B,
514 Malayalam,
515 Mongolian,
516 Myanmar,
517 New_Tai_Lue,
518 Nko,
519 Ogham,
520 Old_Italic,
521 Old_Persian,
522 Oriya,
523 Osmanya,
524 Phags_Pa,
525 Phoenician,
526 Runic,
527 Shavian,
528 Sinhala,
529 Syloti_Nagri,
530 Syriac,
531 Tagalog,
532 Tagbanwa,
533 Tai_Le,
534 Tamil,
535 Telugu,
536 Thaana,
537 Thai,
538 Tibetan,
539 Tifinagh,
540 Ugaritic,
541 Yi.
542 .P
543 Each character has exactly one general category property, specified by a
544 two-letter abbreviation. For compatibility with Perl, negation can be specified
545 by including a circumflex between the opening brace and the property name. For
546 example, \ep{^Lu} is the same as \eP{Lu}.
547 .P
548 If only one letter is specified with \ep or \eP, it includes all the general
549 category properties that start with that letter. In this case, in the absence
550 of negation, the curly brackets in the escape sequence are optional; these two
551 examples have the same effect:
552 .sp
553 \ep{L}
554 \epL
555 .sp
556 The following general category property codes are supported:
557 .sp
558 C Other
559 Cc Control
560 Cf Format
561 Cn Unassigned
562 Co Private use
563 Cs Surrogate
564 .sp
565 L Letter
566 Ll Lower case letter
567 Lm Modifier letter
568 Lo Other letter
569 Lt Title case letter
570 Lu Upper case letter
571 .sp
572 M Mark
573 Mc Spacing mark
574 Me Enclosing mark
575 Mn Non-spacing mark
576 .sp
577 N Number
578 Nd Decimal number
579 Nl Letter number
580 No Other number
581 .sp
582 P Punctuation
583 Pc Connector punctuation
584 Pd Dash punctuation
585 Pe Close punctuation
586 Pf Final punctuation
587 Pi Initial punctuation
588 Po Other punctuation
589 Ps Open punctuation
590 .sp
591 S Symbol
592 Sc Currency symbol
593 Sk Modifier symbol
594 Sm Mathematical symbol
595 So Other symbol
596 .sp
597 Z Separator
598 Zl Line separator
599 Zp Paragraph separator
600 Zs Space separator
601 .sp
602 The special property L& is also supported: it matches a character that has
603 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
604 a modifier or "other".
605 .P
606 The Cs (Surrogate) property applies only to characters in the range U+D800 to
607 U+DFFF. Such characters are not valid in UTF-8 strings (see RFC 3629) and so
608 cannot be tested by PCRE, unless UTF-8 validity checking has been turned off
609 (see the discussion of PCRE_NO_UTF8_CHECK in the
610 .\" HREF
611 \fBpcreapi\fP
612 .\"
613 page).
614 .P
615 The long synonyms for these properties that Perl supports (such as \ep{Letter})
616 are not supported by PCRE, nor is it permitted to prefix any of these
617 properties with "Is".
618 .P
619 No character that is in the Unicode table has the Cn (unassigned) property.
620 Instead, this property is assumed for any code point that is not in the
621 Unicode table.
622 .P
623 Specifying caseless matching does not affect these escape sequences. For
624 example, \ep{Lu} always matches only upper case letters.
625 .P
626 The \eX escape matches any number of Unicode characters that form an extended
627 Unicode sequence. \eX is equivalent to
628 .sp
629 (?>\ePM\epM*)
630 .sp
631 That is, it matches a character without the "mark" property, followed by zero
632 or more characters with the "mark" property, and treats the sequence as an
633 atomic group
634 .\" HTML <a href="#atomicgroup">
635 .\" </a>
636 (see below).
637 .\"
638 Characters with the "mark" property are typically accents that affect the
639 preceding character. None of them have codepoints less than 256, so in
640 non-UTF-8 mode \eX matches any one character.
641 .P
642 Matching characters by Unicode property is not fast, because PCRE has to search
643 a structure that contains data for over fifteen thousand characters. That is
644 why the traditional escape sequences such as \ed and \ew do not use Unicode
645 properties in PCRE.
646 .
647 .
648 .\" HTML <a name="resetmatchstart"></a>
649 .SS "Resetting the match start"
650 .rs
651 .sp
652 The escape sequence \eK, which is a Perl 5.10 feature, causes any previously
653 matched characters not to be included in the final matched sequence. For
654 example, the pattern:
655 .sp
656 foo\eKbar
657 .sp
658 matches "foobar", but reports that it has matched "bar". This feature is
659 similar to a lookbehind assertion
660 .\" HTML <a href="#lookbehind">
661 .\" </a>
662 (described below).
663 .\"
664 However, in this case, the part of the subject before the real match does not
665 have to be of fixed length, as lookbehind assertions do. The use of \eK does
666 not interfere with the setting of
667 .\" HTML <a href="#subpattern">
668 .\" </a>
669 captured substrings.
670 .\"
671 For example, when the pattern
672 .sp
673 (foo)\eKbar
674 .sp
675 matches "foobar", the first substring is still set to "foo".
676 .
677 .
678 .\" HTML <a name="smallassertions"></a>
679 .SS "Simple assertions"
680 .rs
681 .sp
682 The final use of backslash is for certain simple assertions. An assertion
683 specifies a condition that has to be met at a particular point in a match,
684 without consuming any characters from the subject string. The use of
685 subpatterns for more complicated assertions is described
686 .\" HTML <a href="#bigassertions">
687 .\" </a>
688 below.
689 .\"
690 The backslashed assertions are:
691 .sp
692 \eb matches at a word boundary
693 \eB matches when not at a word boundary
694 \eA matches at the start of the subject
695 \eZ matches at the end of the subject
696 also matches before a newline at the end of the subject
697 \ez matches only at the end of the subject
698 \eG matches at the first matching position in the subject
699 .sp
700 These assertions may not appear in character classes (but note that \eb has a
701 different meaning, namely the backspace character, inside a character class).
702 .P
703 A word boundary is a position in the subject string where the current character
704 and the previous character do not both match \ew or \eW (i.e. one matches
705 \ew and the other matches \eW), or the start or end of the string if the
706 first or last character matches \ew, respectively.
707 .P
708 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
709 dollar (described in the next section) in that they only ever match at the very
710 start and end of the subject string, whatever options are set. Thus, they are
711 independent of multiline mode. These three assertions are not affected by the
712 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
713 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
714 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
715 at a point other than the beginning of the subject, \eA can never match. The
716 difference between \eZ and \ez is that \eZ matches before a newline at the end
717 of the string as well as at the very end, whereas \ez matches only at the end.
718 .P
719 The \eG assertion is true only when the current matching position is at the
720 start point of the match, as specified by the \fIstartoffset\fP argument of
721 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
722 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
723 arguments, you can mimic Perl's /g option, and it is in this kind of
724 implementation where \eG can be useful.
725 .P
726 Note, however, that PCRE's interpretation of \eG, as the start of the current
727 match, is subtly different from Perl's, which defines it as the end of the
728 previous match. In Perl, these can be different when the previously matched
729 string was empty. Because PCRE does just one match at a time, it cannot
730 reproduce this behaviour.
731 .P
732 If all the alternatives of a pattern begin with \eG, the expression is anchored
733 to the starting match position, and the "anchored" flag is set in the compiled
734 regular expression.
735 .
736 .
738 .rs
739 .sp
740 Outside a character class, in the default matching mode, the circumflex
741 character is an assertion that is true only if the current matching point is
742 at the start of the subject string. If the \fIstartoffset\fP argument of
743 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
744 option is unset. Inside a character class, circumflex has an entirely different
745 meaning
746 .\" HTML <a href="#characterclass">
747 .\" </a>
748 (see below).
749 .\"
750 .P
751 Circumflex need not be the first character of the pattern if a number of
752 alternatives are involved, but it should be the first thing in each alternative
753 in which it appears if the pattern is ever to match that branch. If all
754 possible alternatives start with a circumflex, that is, if the pattern is
755 constrained to match only at the start of the subject, it is said to be an
756 "anchored" pattern. (There are also other constructs that can cause a pattern
757 to be anchored.)
758 .P
759 A dollar character is an assertion that is true only if the current matching
760 point is at the end of the subject string, or immediately before a newline
761 at the end of the string (by default). Dollar need not be the last character of
762 the pattern if a number of alternatives are involved, but it should be the last
763 item in any branch in which it appears. Dollar has no special meaning in a
764 character class.
765 .P
766 The meaning of dollar can be changed so that it matches only at the very end of
767 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
768 does not affect the \eZ assertion.
769 .P
770 The meanings of the circumflex and dollar characters are changed if the
771 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
772 immediately after internal newlines as well as at the start of the subject
773 string. It does not match after a newline that ends the string. A dollar
774 matches before any newlines in the string, as well as at the very end, when
775 PCRE_MULTILINE is set. When newline is specified as the two-character
776 sequence CRLF, isolated CR and LF characters do not indicate newlines.
777 .P
778 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
779 \en represents a newline) in multiline mode, but not otherwise. Consequently,
780 patterns that are anchored in single line mode because all branches start with
781 ^ are not anchored in multiline mode, and a match for circumflex is possible
782 when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
783 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
784 .P
785 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
786 end of the subject in both modes, and if all branches of a pattern start with
787 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
788 .
789 .
791 .rs
792 .sp
793 Outside a character class, a dot in the pattern matches any one character in
794 the subject string except (by default) a character that signifies the end of a
795 line. In UTF-8 mode, the matched character may be more than one byte long.
796 .P
797 When a line ending is defined as a single character, dot never matches that
798 character; when the two-character sequence CRLF is used, dot does not match CR
799 if it is immediately followed by LF, but otherwise it matches all characters
800 (including isolated CRs and LFs). When any Unicode line endings are being
801 recognized, dot does not match CR or LF or any of the other line ending
802 characters.
803 .P
804 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
805 option is set, a dot matches any one character, without exception. If the
806 two-character sequence CRLF is present in the subject string, it takes two dots
807 to match it.
808 .P
809 The handling of dot is entirely independent of the handling of circumflex and
810 dollar, the only relationship being that they both involve newlines. Dot has no
811 special meaning in a character class.
812 .
813 .
815 .rs
816 .sp
817 Outside a character class, the escape sequence \eC matches any one byte, both
818 in and out of UTF-8 mode. Unlike a dot, it always matches any line-ending
819 characters. The feature is provided in Perl in order to match individual bytes
820 in UTF-8 mode. Because it breaks up UTF-8 characters into individual bytes,
821 what remains in the string may be a malformed UTF-8 string. For this reason,
822 the \eC escape sequence is best avoided.
823 .P
824 PCRE does not allow \eC to appear in lookbehind assertions
825 .\" HTML <a href="#lookbehind">
826 .\" </a>
827 (described below),
828 .\"
829 because in UTF-8 mode this would make it impossible to calculate the length of
830 the lookbehind.
831 .
832 .
833 .\" HTML <a name="characterclass"></a>
835 .rs
836 .sp
837 An opening square bracket introduces a character class, terminated by a closing
838 square bracket. A closing square bracket on its own is not special. If a
839 closing square bracket is required as a member of the class, it should be the
840 first data character in the class (after an initial circumflex, if present) or
841 escaped with a backslash.
842 .P
843 A character class matches a single character in the subject. In UTF-8 mode, the
844 character may occupy more than one byte. A matched character must be in the set
845 of characters defined by the class, unless the first character in the class
846 definition is a circumflex, in which case the subject character must not be in
847 the set defined by the class. If a circumflex is actually required as a member
848 of the class, ensure it is not the first character, or escape it with a
849 backslash.
850 .P
851 For example, the character class [aeiou] matches any lower case vowel, while
852 [^aeiou] matches any character that is not a lower case vowel. Note that a
853 circumflex is just a convenient notation for specifying the characters that
854 are in the class by enumerating those that are not. A class that starts with a
855 circumflex is not an assertion: it still consumes a character from the subject
856 string, and therefore it fails if the current pointer is at the end of the
857 string.
858 .P
859 In UTF-8 mode, characters with values greater than 255 can be included in a
860 class as a literal string of bytes, or by using the \ex{ escaping mechanism.
861 .P
862 When caseless matching is set, any letters in a class represent both their
863 upper case and lower case versions, so for example, a caseless [aeiou] matches
864 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
865 caseful version would. In UTF-8 mode, PCRE always understands the concept of
866 case for characters whose values are less than 128, so caseless matching is
867 always possible. For characters with higher values, the concept of case is
868 supported if PCRE is compiled with Unicode property support, but not otherwise.
869 If you want to use caseless matching for characters 128 and above, you must
870 ensure that PCRE is compiled with Unicode property support as well as with
871 UTF-8 support.
872 .P
873 Characters that might indicate line breaks are never treated in any special way
874 when matching character classes, whatever line-ending sequence is in use, and
875 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
876 such as [^a] always matches one of these characters.
877 .P
878 The minus (hyphen) character can be used to specify a range of characters in a
879 character class. For example, [d-m] matches any letter between d and m,
880 inclusive. If a minus character is required in a class, it must be escaped with
881 a backslash or appear in a position where it cannot be interpreted as
882 indicating a range, typically as the first or last character in the class.
883 .P
884 It is not possible to have the literal character "]" as the end character of a
885 range. A pattern such as [W-]46] is interpreted as a class of two characters
886 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
887 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
888 the end of range, so [W-\e]46] is interpreted as a class containing a range
889 followed by two other characters. The octal or hexadecimal representation of
890 "]" can also be used to end a range.
891 .P
892 Ranges operate in the collating sequence of character values. They can also be
893 used for characters specified numerically, for example [\e000-\e037]. In UTF-8
894 mode, ranges can include characters whose values are greater than 255, for
895 example [\ex{100}-\ex{2ff}].
896 .P
897 If a range that includes letters is used when caseless matching is set, it
898 matches the letters in either case. For example, [W-c] is equivalent to
899 [][\e\e^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
900 tables for a French locale are in use, [\exc8-\excb] matches accented E
901 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
902 characters with values greater than 128 only when it is compiled with Unicode
903 property support.
904 .P
905 The character types \ed, \eD, \ep, \eP, \es, \eS, \ew, and \eW may also appear
906 in a character class, and add the characters that they match to the class. For
907 example, [\edABCDEF] matches any hexadecimal digit. A circumflex can
908 conveniently be used with the upper case character types to specify a more
909 restricted set of characters than the matching lower case type. For example,
910 the class [^\eW_] matches any letter or digit, but not underscore.
911 .P
912 The only metacharacters that are recognized in character classes are backslash,
913 hyphen (only where it can be interpreted as specifying a range), circumflex
914 (only at the start), opening square bracket (only when it can be interpreted as
915 introducing a POSIX class name - see the next section), and the terminating
916 closing square bracket. However, escaping other non-alphanumeric characters
917 does no harm.
918 .
919 .
921 .rs
922 .sp
923 Perl supports the POSIX notation for character classes. This uses names
924 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
925 this notation. For example,
926 .sp
927 [01[:alpha:]%]
928 .sp
929 matches "0", "1", any alphabetic character, or "%". The supported class names
930 are
931 .sp
932 alnum letters and digits
933 alpha letters
934 ascii character codes 0 - 127
935 blank space or tab only
936 cntrl control characters
937 digit decimal digits (same as \ed)
938 graph printing characters, excluding space
939 lower lower case letters
940 print printing characters, including space
941 punct printing characters, excluding letters and digits
942 space white space (not quite the same as \es)
943 upper upper case letters
944 word "word" characters (same as \ew)
945 xdigit hexadecimal digits
946 .sp
947 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
948 space (32). Notice that this list includes the VT character (code 11). This
949 makes "space" different to \es, which does not include VT (for Perl
950 compatibility).
951 .P
952 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
953 5.8. Another Perl extension is negation, which is indicated by a ^ character
954 after the colon. For example,
955 .sp
956 [12[:^digit:]]
957 .sp
958 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
959 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
960 supported, and an error is given if they are encountered.
961 .P
962 In UTF-8 mode, characters with values greater than 128 do not match any of
963 the POSIX character classes.
964 .
965 .
967 .rs
968 .sp
969 Vertical bar characters are used to separate alternative patterns. For example,
970 the pattern
971 .sp
972 gilbert|sullivan
973 .sp
974 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
975 and an empty alternative is permitted (matching the empty string). The matching
976 process tries each alternative in turn, from left to right, and the first one
977 that succeeds is used. If the alternatives are within a subpattern
978 .\" HTML <a href="#subpattern">
979 .\" </a>
980 (defined below),
981 .\"
982 "succeeds" means matching the rest of the main pattern as well as the
983 alternative in the subpattern.
984 .
985 .
987 .rs
988 .sp
990 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
991 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
992 The option letters are
993 .sp
996 s for PCRE_DOTALL
998 .sp
999 For example, (?im) sets caseless, multiline matching. It is also possible to
1000 unset these options by preceding the letter with a hyphen, and a combined
1001 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1002 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1003 permitted. If a letter appears both before and after the hyphen, the option is
1004 unset.
1005 .P
1006 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1007 changed in the same way as the Perl-compatible options by using the characters
1008 J, U and X respectively.
1009 .P
1010 When an option change occurs at top level (that is, not inside subpattern
1011 parentheses), the change applies to the remainder of the pattern that follows.
1012 If the change is placed right at the start of a pattern, PCRE extracts it into
1013 the global options (and it will therefore show up in data extracted by the
1014 \fBpcre_fullinfo()\fP function).
1015 .P
1016 An option change within a subpattern (see below for a description of
1017 subpatterns) affects only that part of the current pattern that follows it, so
1018 .sp
1019 (a(?i)b)c
1020 .sp
1021 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1022 By this means, options can be made to have different settings in different
1023 parts of the pattern. Any changes made in one alternative do carry on
1024 into subsequent branches within the same subpattern. For example,
1025 .sp
1026 (a(?i)b|c)
1027 .sp
1028 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1029 branch is abandoned before the option setting. This is because the effects of
1030 option settings happen at compile time. There would be some very weird
1031 behaviour otherwise.
1032 .
1033 .
1034 .\" HTML <a name="subpattern"></a>
1036 .rs
1037 .sp
1038 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1039 Turning part of a pattern into a subpattern does two things:
1040 .sp
1041 1. It localizes a set of alternatives. For example, the pattern
1042 .sp
1043 cat(aract|erpillar|)
1044 .sp
1045 matches one of the words "cat", "cataract", or "caterpillar". Without the
1046 parentheses, it would match "cataract", "erpillar" or an empty string.
1047 .sp
1048 2. It sets up the subpattern as a capturing subpattern. This means that, when
1049 the whole pattern matches, that portion of the subject string that matched the
1050 subpattern is passed back to the caller via the \fIovector\fP argument of
1051 \fBpcre_exec()\fP. Opening parentheses are counted from left to right (starting
1052 from 1) to obtain numbers for the capturing subpatterns.
1053 .P
1054 For example, if the string "the red king" is matched against the pattern
1055 .sp
1056 the ((red|white) (king|queen))
1057 .sp
1058 the captured substrings are "red king", "red", and "king", and are numbered 1,
1059 2, and 3, respectively.
1060 .P
1061 The fact that plain parentheses fulfil two functions is not always helpful.
1062 There are often times when a grouping subpattern is required without a
1063 capturing requirement. If an opening parenthesis is followed by a question mark
1064 and a colon, the subpattern does not do any capturing, and is not counted when
1065 computing the number of any subsequent capturing subpatterns. For example, if
1066 the string "the white queen" is matched against the pattern
1067 .sp
1068 the ((?:red|white) (king|queen))
1069 .sp
1070 the captured substrings are "white queen" and "queen", and are numbered 1 and
1071 2. The maximum number of capturing subpatterns is 65535.
1072 .P
1073 As a convenient shorthand, if any option settings are required at the start of
1074 a non-capturing subpattern, the option letters may appear between the "?" and
1075 the ":". Thus the two patterns
1076 .sp
1077 (?i:saturday|sunday)
1078 (?:(?i)saturday|sunday)
1079 .sp
1080 match exactly the same set of strings. Because alternative branches are tried
1081 from left to right, and options are not reset until the end of the subpattern
1082 is reached, an option setting in one branch does affect subsequent branches, so
1083 the above patterns match "SUNDAY" as well as "Saturday".
1084 .
1085 .
1087 .rs
1088 .sp
1089 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1090 the same numbers for its capturing parentheses. Such a subpattern starts with
1091 (?| and is itself a non-capturing subpattern. For example, consider this
1092 pattern:
1093 .sp
1094 (?|(Sat)ur|(Sun))day
1095 .sp
1096 Because the two alternatives are inside a (?| group, both sets of capturing
1097 parentheses are numbered one. Thus, when the pattern matches, you can look
1098 at captured substring number one, whichever alternative matched. This construct
1099 is useful when you want to capture part, but not all, of one of a number of
1100 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1101 number is reset at the start of each branch. The numbers of any capturing
1102 buffers that follow the subpattern start after the highest number used in any
1103 branch. The following example is taken from the Perl documentation.
1104 The numbers underneath show in which buffer the captured content will be
1105 stored.
1106 .sp
1107 # before ---------------branch-reset----------- after
1108 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1109 # 1 2 2 3 2 3 4
1110 .sp
1111 A backreference or a recursive call to a numbered subpattern always refers to
1112 the first one in the pattern with the given number.
1113 .P
1114 An alternative approach to using this "branch reset" feature is to use
1115 duplicate named subpatterns, as described in the next section.
1116 .
1117 .
1119 .rs
1120 .sp
1121 Identifying capturing parentheses by number is simple, but it can be very hard
1122 to keep track of the numbers in complicated regular expressions. Furthermore,
1123 if an expression is modified, the numbers may change. To help with this
1124 difficulty, PCRE supports the naming of subpatterns. This feature was not
1125 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1126 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1127 the Perl and the Python syntax.
1128 .P
1129 In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1130 (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1131 parentheses from other parts of the pattern, such as
1132 .\" HTML <a href="#backreferences">
1133 .\" </a>
1134 backreferences,
1135 .\"
1136 .\" HTML <a href="#recursion">
1137 .\" </a>
1138 recursion,
1139 .\"
1140 and
1141 .\" HTML <a href="#conditions">
1142 .\" </a>
1143 conditions,
1144 .\"
1145 can be made by name as well as by number.
1146 .P
1147 Names consist of up to 32 alphanumeric characters and underscores. Named
1148 capturing parentheses are still allocated numbers as well as names, exactly as
1149 if the names were not present. The PCRE API provides function calls for
1150 extracting the name-to-number translation table from a compiled pattern. There
1151 is also a convenience function for extracting a captured substring by name.
1152 .P
1153 By default, a name must be unique within a pattern, but it is possible to relax
1154 this constraint by setting the PCRE_DUPNAMES option at compile time. This can
1155 be useful for patterns where only one instance of the named parentheses can
1156 match. Suppose you want to match the name of a weekday, either as a 3-letter
1157 abbreviation or as the full name, and in both cases you want to extract the
1158 abbreviation. This pattern (ignoring the line breaks) does the job:
1159 .sp
1160 (?<DN>Mon|Fri|Sun)(?:day)?|
1161 (?<DN>Tue)(?:sday)?|
1162 (?<DN>Wed)(?:nesday)?|
1163 (?<DN>Thu)(?:rsday)?|
1164 (?<DN>Sat)(?:urday)?
1165 .sp
1166 There are five capturing substrings, but only one is ever set after a match.
1167 (An alternative way of solving this problem is to use a "branch reset"
1168 subpattern, as described in the previous section.)
1169 .P
1170 The convenience function for extracting the data by name returns the substring
1171 for the first (and in this example, the only) subpattern of that name that
1172 matched. This saves searching to find which numbered subpattern it was. If you
1173 make a reference to a non-unique named subpattern from elsewhere in the
1174 pattern, the one that corresponds to the lowest number is used. For further
1175 details of the interfaces for handling named subpatterns, see the
1176 .\" HREF
1177 \fBpcreapi\fP
1178 .\"
1179 documentation.
1180 .
1181 .
1183 .rs
1184 .sp
1185 Repetition is specified by quantifiers, which can follow any of the following
1186 items:
1187 .sp
1188 a literal data character
1189 the dot metacharacter
1190 the \eC escape sequence
1191 the \eX escape sequence (in UTF-8 mode with Unicode properties)
1192 the \eR escape sequence
1193 an escape such as \ed that matches a single character
1194 a character class
1195 a back reference (see next section)
1196 a parenthesized subpattern (unless it is an assertion)
1197 .sp
1198 The general repetition quantifier specifies a minimum and maximum number of
1199 permitted matches, by giving the two numbers in curly brackets (braces),
1200 separated by a comma. The numbers must be less than 65536, and the first must
1201 be less than or equal to the second. For example:
1202 .sp
1203 z{2,4}
1204 .sp
1205 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1206 character. If the second number is omitted, but the comma is present, there is
1207 no upper limit; if the second number and the comma are both omitted, the
1208 quantifier specifies an exact number of required matches. Thus
1209 .sp
1210 [aeiou]{3,}
1211 .sp
1212 matches at least 3 successive vowels, but may match many more, while
1213 .sp
1214 \ed{8}
1215 .sp
1216 matches exactly 8 digits. An opening curly bracket that appears in a position
1217 where a quantifier is not allowed, or one that does not match the syntax of a
1218 quantifier, is taken as a literal character. For example, {,6} is not a
1219 quantifier, but a literal string of four characters.
1220 .P
1221 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1222 bytes. Thus, for example, \ex{100}{2} matches two UTF-8 characters, each of
1223 which is represented by a two-byte sequence. Similarly, when Unicode property
1224 support is available, \eX{3} matches three Unicode extended sequences, each of
1225 which may be several bytes long (and they may be of different lengths).
1226 .P
1227 The quantifier {0} is permitted, causing the expression to behave as if the
1228 previous item and the quantifier were not present.
1229 .P
1230 For convenience, the three most common quantifiers have single-character
1231 abbreviations:
1232 .sp
1233 * is equivalent to {0,}
1234 + is equivalent to {1,}
1235 ? is equivalent to {0,1}
1236 .sp
1237 It is possible to construct infinite loops by following a subpattern that can
1238 match no characters with a quantifier that has no upper limit, for example:
1239 .sp
1240 (a?)*
1241 .sp
1242 Earlier versions of Perl and PCRE used to give an error at compile time for
1243 such patterns. However, because there are cases where this can be useful, such
1244 patterns are now accepted, but if any repetition of the subpattern does in fact
1245 match no characters, the loop is forcibly broken.
1246 .P
1247 By default, the quantifiers are "greedy", that is, they match as much as
1248 possible (up to the maximum number of permitted times), without causing the
1249 rest of the pattern to fail. The classic example of where this gives problems
1250 is in trying to match comments in C programs. These appear between /* and */
1251 and within the comment, individual * and / characters may appear. An attempt to
1252 match C comments by applying the pattern
1253 .sp
1254 /\e*.*\e*/
1255 .sp
1256 to the string
1257 .sp
1258 /* first comment */ not comment /* second comment */
1259 .sp
1260 fails, because it matches the entire string owing to the greediness of the .*
1261 item.
1262 .P
1263 However, if a quantifier is followed by a question mark, it ceases to be
1264 greedy, and instead matches the minimum number of times possible, so the
1265 pattern
1266 .sp
1267 /\e*.*?\e*/
1268 .sp
1269 does the right thing with the C comments. The meaning of the various
1270 quantifiers is not otherwise changed, just the preferred number of matches.
1271 Do not confuse this use of question mark with its use as a quantifier in its
1272 own right. Because it has two uses, it can sometimes appear doubled, as in
1273 .sp
1274 \ed??\ed
1275 .sp
1276 which matches one digit by preference, but can match two if that is the only
1277 way the rest of the pattern matches.
1278 .P
1279 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1280 the quantifiers are not greedy by default, but individual ones can be made
1281 greedy by following them with a question mark. In other words, it inverts the
1282 default behaviour.
1283 .P
1284 When a parenthesized subpattern is quantified with a minimum repeat count that
1285 is greater than 1 or with a limited maximum, more memory is required for the
1286 compiled pattern, in proportion to the size of the minimum or maximum.
1287 .P
1288 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1289 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1290 implicitly anchored, because whatever follows will be tried against every
1291 character position in the subject string, so there is no point in retrying the
1292 overall match at any position after the first. PCRE normally treats such a
1293 pattern as though it were preceded by \eA.
1294 .P
1295 In cases where it is known that the subject string contains no newlines, it is
1296 worth setting PCRE_DOTALL in order to obtain this optimization, or
1297 alternatively using ^ to indicate anchoring explicitly.
1298 .P
1299 However, there is one situation where the optimization cannot be used. When .*
1300 is inside capturing parentheses that are the subject of a backreference
1301 elsewhere in the pattern, a match at the start may fail where a later one
1302 succeeds. Consider, for example:
1303 .sp
1304 (.*)abc\e1
1305 .sp
1306 If the subject is "xyz123abc123" the match point is the fourth character. For
1307 this reason, such a pattern is not implicitly anchored.
1308 .P
1309 When a capturing subpattern is repeated, the value captured is the substring
1310 that matched the final iteration. For example, after
1311 .sp
1312 (tweedle[dume]{3}\es*)+
1313 .sp
1314 has matched "tweedledum tweedledee" the value of the captured substring is
1315 "tweedledee". However, if there are nested capturing subpatterns, the
1316 corresponding captured values may have been set in previous iterations. For
1317 example, after
1318 .sp
1319 /(a|(b))+/
1320 .sp
1321 matches "aba" the value of the second captured substring is "b".
1322 .
1323 .
1324 .\" HTML <a name="atomicgroup"></a>
1326 .rs
1327 .sp
1328 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1329 repetition, failure of what follows normally causes the repeated item to be
1330 re-evaluated to see if a different number of repeats allows the rest of the
1331 pattern to match. Sometimes it is useful to prevent this, either to change the
1332 nature of the match, or to cause it fail earlier than it otherwise might, when
1333 the author of the pattern knows there is no point in carrying on.
1334 .P
1335 Consider, for example, the pattern \ed+foo when applied to the subject line
1336 .sp
1337 123456bar
1338 .sp
1339 After matching all 6 digits and then failing to match "foo", the normal
1340 action of the matcher is to try again with only 5 digits matching the \ed+
1341 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1342 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1343 that once a subpattern has matched, it is not to be re-evaluated in this way.
1344 .P
1345 If we use atomic grouping for the previous example, the matcher gives up
1346 immediately on failing to match "foo" the first time. The notation is a kind of
1347 special parenthesis, starting with (?> as in this example:
1348 .sp
1349 (?>\ed+)foo
1350 .sp
1351 This kind of parenthesis "locks up" the part of the pattern it contains once
1352 it has matched, and a failure further into the pattern is prevented from
1353 backtracking into it. Backtracking past it to previous items, however, works as
1354 normal.
1355 .P
1356 An alternative description is that a subpattern of this type matches the string
1357 of characters that an identical standalone pattern would match, if anchored at
1358 the current point in the subject string.
1359 .P
1360 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1361 the above example can be thought of as a maximizing repeat that must swallow
1362 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1363 number of digits they match in order to make the rest of the pattern match,
1364 (?>\ed+) can only match an entire sequence of digits.
1365 .P
1366 Atomic groups in general can of course contain arbitrarily complicated
1367 subpatterns, and can be nested. However, when the subpattern for an atomic
1368 group is just a single repeated item, as in the example above, a simpler
1369 notation, called a "possessive quantifier" can be used. This consists of an
1370 additional + character following a quantifier. Using this notation, the
1371 previous example can be rewritten as
1372 .sp
1373 \ed++foo
1374 .sp
1375 Note that a possessive quantifier can be used with an entire group, for
1376 example:
1377 .sp
1378 (abc|xyz){2,3}+
1379 .sp
1380 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1381 option is ignored. They are a convenient notation for the simpler forms of
1382 atomic group. However, there is no difference in the meaning of a possessive
1383 quantifier and the equivalent atomic group, though there may be a performance
1384 difference; possessive quantifiers should be slightly faster.
1385 .P
1386 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1387 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1388 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1389 package, and PCRE copied it from there. It ultimately found its way into Perl
1390 at release 5.10.
1391 .P
1392 PCRE has an optimization that automatically "possessifies" certain simple
1393 pattern constructs. For example, the sequence A+B is treated as A++B because
1394 there is no point in backtracking into a sequence of A's when B must follow.
1395 .P
1396 When a pattern contains an unlimited repeat inside a subpattern that can itself
1397 be repeated an unlimited number of times, the use of an atomic group is the
1398 only way to avoid some failing matches taking a very long time indeed. The
1399 pattern
1400 .sp
1401 (\eD+|<\ed+>)*[!?]
1402 .sp
1403 matches an unlimited number of substrings that either consist of non-digits, or
1404 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1405 quickly. However, if it is applied to
1406 .sp
1407 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1408 .sp
1409 it takes a long time before reporting failure. This is because the string can
1410 be divided between the internal \eD+ repeat and the external * repeat in a
1411 large number of ways, and all have to be tried. (The example uses [!?] rather
1412 than a single character at the end, because both PCRE and Perl have an
1413 optimization that allows for fast failure when a single character is used. They
1414 remember the last single character that is required for a match, and fail early
1415 if it is not present in the string.) If the pattern is changed so that it uses
1416 an atomic group, like this:
1417 .sp
1418 ((?>\eD+)|<\ed+>)*[!?]
1419 .sp
1420 sequences of non-digits cannot be broken, and failure happens quickly.
1421 .
1422 .
1423 .\" HTML <a name="backreferences"></a>
1425 .rs
1426 .sp
1427 Outside a character class, a backslash followed by a digit greater than 0 (and
1428 possibly further digits) is a back reference to a capturing subpattern earlier
1429 (that is, to its left) in the pattern, provided there have been that many
1430 previous capturing left parentheses.
1431 .P
1432 However, if the decimal number following the backslash is less than 10, it is
1433 always taken as a back reference, and causes an error only if there are not
1434 that many capturing left parentheses in the entire pattern. In other words, the
1435 parentheses that are referenced need not be to the left of the reference for
1436 numbers less than 10. A "forward back reference" of this type can make sense
1437 when a repetition is involved and the subpattern to the right has participated
1438 in an earlier iteration.
1439 .P
1440 It is not possible to have a numerical "forward back reference" to a subpattern
1441 whose number is 10 or more using this syntax because a sequence such as \e50 is
1442 interpreted as a character defined in octal. See the subsection entitled
1443 "Non-printing characters"
1444 .\" HTML <a href="#digitsafterbackslash">
1445 .\" </a>
1446 above
1447 .\"
1448 for further details of the handling of digits following a backslash. There is
1449 no such problem when named parentheses are used. A back reference to any
1450 subpattern is possible using named parentheses (see below).
1451 .P
1452 Another way of avoiding the ambiguity inherent in the use of digits following a
1453 backslash is to use the \eg escape sequence, which is a feature introduced in
1454 Perl 5.10. This escape must be followed by an unsigned number or a negative
1455 number, optionally enclosed in braces. These examples are all identical:
1456 .sp
1457 (ring), \e1
1458 (ring), \eg1
1459 (ring), \eg{1}
1460 .sp
1461 An unsigned number specifies an absolute reference without the ambiguity that
1462 is present in the older syntax. It is also useful when literal digits follow
1463 the reference. A negative number is a relative reference. Consider this
1464 example:
1465 .sp
1466 (abc(def)ghi)\eg{-1}
1467 .sp
1468 The sequence \eg{-1} is a reference to the most recently started capturing
1469 subpattern before \eg, that is, is it equivalent to \e2. Similarly, \eg{-2}
1470 would be equivalent to \e1. The use of relative references can be helpful in
1471 long patterns, and also in patterns that are created by joining together
1472 fragments that contain references within themselves.
1473 .P
1474 A back reference matches whatever actually matched the capturing subpattern in
1475 the current subject string, rather than anything matching the subpattern
1476 itself (see
1477 .\" HTML <a href="#subpatternsassubroutines">
1478 .\" </a>
1479 "Subpatterns as subroutines"
1480 .\"
1481 below for a way of doing that). So the pattern
1482 .sp
1483 (sens|respons)e and \e1ibility
1484 .sp
1485 matches "sense and sensibility" and "response and responsibility", but not
1486 "sense and responsibility". If caseful matching is in force at the time of the
1487 back reference, the case of letters is relevant. For example,
1488 .sp
1489 ((?i)rah)\es+\e1
1490 .sp
1491 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1492 capturing subpattern is matched caselessly.
1493 .P
1494 There are several different ways of writing back references to named
1495 subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1496 \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1497 back reference syntax, in which \eg can be used for both numeric and named
1498 references, is also supported. We could rewrite the above example in any of
1499 the following ways:
1500 .sp
1501 (?<p1>(?i)rah)\es+\ek<p1>
1502 (?'p1'(?i)rah)\es+\ek{p1}
1503 (?P<p1>(?i)rah)\es+(?P=p1)
1504 (?<p1>(?i)rah)\es+\eg{p1}
1505 .sp
1506 A subpattern that is referenced by name may appear in the pattern before or
1507 after the reference.
1508 .P
1509 There may be more than one back reference to the same subpattern. If a
1510 subpattern has not actually been used in a particular match, any back
1511 references to it always fail. For example, the pattern
1512 .sp
1513 (a|(bc))\e2
1514 .sp
1515 always fails if it starts to match "a" rather than "bc". Because there may be
1516 many capturing parentheses in a pattern, all digits following the backslash are
1517 taken as part of a potential back reference number. If the pattern continues
1518 with a digit character, some delimiter must be used to terminate the back
1519 reference. If the PCRE_EXTENDED option is set, this can be whitespace.
1520 Otherwise an empty comment (see
1521 .\" HTML <a href="#comments">
1522 .\" </a>
1523 "Comments"
1524 .\"
1525 below) can be used.
1526 .P
1527 A back reference that occurs inside the parentheses to which it refers fails
1528 when the subpattern is first used, so, for example, (a\e1) never matches.
1529 However, such references can be useful inside repeated subpatterns. For
1530 example, the pattern
1531 .sp
1532 (a|b\e1)+
1533 .sp
1534 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1535 the subpattern, the back reference matches the character string corresponding
1536 to the previous iteration. In order for this to work, the pattern must be such
1537 that the first iteration does not need to match the back reference. This can be
1538 done using alternation, as in the example above, or by a quantifier with a
1539 minimum of zero.
1540 .
1541 .
1542 .\" HTML <a name="bigassertions"></a>
1544 .rs
1545 .sp
1546 An assertion is a test on the characters following or preceding the current
1547 matching point that does not actually consume any characters. The simple
1548 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1549 .\" HTML <a href="#smallassertions">
1550 .\" </a>
1551 above.
1552 .\"
1553 .P
1554 More complicated assertions are coded as subpatterns. There are two kinds:
1555 those that look ahead of the current position in the subject string, and those
1556 that look behind it. An assertion subpattern is matched in the normal way,
1557 except that it does not cause the current matching position to be changed.
1558 .P
1559 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1560 because it makes no sense to assert the same thing several times. If any kind
1561 of assertion contains capturing subpatterns within it, these are counted for
1562 the purposes of numbering the capturing subpatterns in the whole pattern.
1563 However, substring capturing is carried out only for positive assertions,
1564 because it does not make sense for negative assertions.
1565 .
1566 .
1567 .SS "Lookahead assertions"
1568 .rs
1569 .sp
1570 Lookahead assertions start with (?= for positive assertions and (?! for
1571 negative assertions. For example,
1572 .sp
1573 \ew+(?=;)
1574 .sp
1575 matches a word followed by a semicolon, but does not include the semicolon in
1576 the match, and
1577 .sp
1578 foo(?!bar)
1579 .sp
1580 matches any occurrence of "foo" that is not followed by "bar". Note that the
1581 apparently similar pattern
1582 .sp
1583 (?!foo)bar
1584 .sp
1585 does not find an occurrence of "bar" that is preceded by something other than
1586 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1587 (?!foo) is always true when the next three characters are "bar". A
1588 lookbehind assertion is needed to achieve the other effect.
1589 .P
1590 If you want to force a matching failure at some point in a pattern, the most
1591 convenient way to do it is with (?!) because an empty string always matches, so
1592 an assertion that requires there not to be an empty string must always fail.
1593 .
1594 .
1595 .\" HTML <a name="lookbehind"></a>
1596 .SS "Lookbehind assertions"
1597 .rs
1598 .sp
1599 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1600 negative assertions. For example,
1601 .sp
1602 (?<!foo)bar
1603 .sp
1604 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1605 a lookbehind assertion are restricted such that all the strings it matches must
1606 have a fixed length. However, if there are several top-level alternatives, they
1607 do not all have to have the same fixed length. Thus
1608 .sp
1609 (?<=bullock|donkey)
1610 .sp
1611 is permitted, but
1612 .sp
1613 (?<!dogs?|cats?)
1614 .sp
1615 causes an error at compile time. Branches that match different length strings
1616 are permitted only at the top level of a lookbehind assertion. This is an
1617 extension compared with Perl (at least for 5.8), which requires all branches to
1618 match the same length of string. An assertion such as
1619 .sp
1620 (?<=ab(c|de))
1621 .sp
1622 is not permitted, because its single top-level branch can match two different
1623 lengths, but it is acceptable if rewritten to use two top-level branches:
1624 .sp
1625 (?<=abc|abde)
1626 .sp
1627 In some cases, the Perl 5.10 escape sequence \eK
1628 .\" HTML <a href="#resetmatchstart">
1629 .\" </a>
1630 (see above)
1631 .\"
1632 can be used instead of a lookbehind assertion; this is not restricted to a
1633 fixed-length.
1634 .P
1635 The implementation of lookbehind assertions is, for each alternative, to
1636 temporarily move the current position back by the fixed length and then try to
1637 match. If there are insufficient characters before the current position, the
1638 assertion fails.
1639 .P
1640 PCRE does not allow the \eC escape (which matches a single byte in UTF-8 mode)
1641 to appear in lookbehind assertions, because it makes it impossible to calculate
1642 the length of the lookbehind. The \eX and \eR escapes, which can match
1643 different numbers of bytes, are also not permitted.
1644 .P
1645 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1646 specify efficient matching at the end of the subject string. Consider a simple
1647 pattern such as
1648 .sp
1649 abcd$
1650 .sp
1651 when applied to a long string that does not match. Because matching proceeds
1652 from left to right, PCRE will look for each "a" in the subject and then see if
1653 what follows matches the rest of the pattern. If the pattern is specified as
1654 .sp
1655 ^.*abcd$
1656 .sp
1657 the initial .* matches the entire string at first, but when this fails (because
1658 there is no following "a"), it backtracks to match all but the last character,
1659 then all but the last two characters, and so on. Once again the search for "a"
1660 covers the entire string, from right to left, so we are no better off. However,
1661 if the pattern is written as
1662 .sp
1663 ^.*+(?<=abcd)
1664 .sp
1665 there can be no backtracking for the .*+ item; it can match only the entire
1666 string. The subsequent lookbehind assertion does a single test on the last four
1667 characters. If it fails, the match fails immediately. For long strings, this
1668 approach makes a significant difference to the processing time.
1669 .
1670 .
1671 .SS "Using multiple assertions"
1672 .rs
1673 .sp
1674 Several assertions (of any sort) may occur in succession. For example,
1675 .sp
1676 (?<=\ed{3})(?<!999)foo
1677 .sp
1678 matches "foo" preceded by three digits that are not "999". Notice that each of
1679 the assertions is applied independently at the same point in the subject
1680 string. First there is a check that the previous three characters are all
1681 digits, and then there is a check that the same three characters are not "999".
1682 This pattern does \fInot\fP match "foo" preceded by six characters, the first
1683 of which are digits and the last three of which are not "999". For example, it
1684 doesn't match "123abcfoo". A pattern to do that is
1685 .sp
1686 (?<=\ed{3}...)(?<!999)foo
1687 .sp
1688 This time the first assertion looks at the preceding six characters, checking
1689 that the first three are digits, and then the second assertion checks that the
1690 preceding three characters are not "999".
1691 .P
1692 Assertions can be nested in any combination. For example,
1693 .sp
1694 (?<=(?<!foo)bar)baz
1695 .sp
1696 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1697 preceded by "foo", while
1698 .sp
1699 (?<=\ed{3}(?!999)...)foo
1700 .sp
1701 is another pattern that matches "foo" preceded by three digits and any three
1702 characters that are not "999".
1703 .
1704 .
1705 .\" HTML <a name="conditions"></a>
1707 .rs
1708 .sp
1709 It is possible to cause the matching process to obey a subpattern
1710 conditionally or to choose between two alternative subpatterns, depending on
1711 the result of an assertion, or whether a previous capturing subpattern matched
1712 or not. The two possible forms of conditional subpattern are
1713 .sp
1714 (?(condition)yes-pattern)
1715 (?(condition)yes-pattern|no-pattern)
1716 .sp
1717 If the condition is satisfied, the yes-pattern is used; otherwise the
1718 no-pattern (if present) is used. If there are more than two alternatives in the
1719 subpattern, a compile-time error occurs.
1720 .P
1721 There are four kinds of condition: references to subpatterns, references to
1722 recursion, a pseudo-condition called DEFINE, and assertions.
1723 .
1724 .SS "Checking for a used subpattern by number"
1725 .rs
1726 .sp
1727 If the text between the parentheses consists of a sequence of digits, the
1728 condition is true if the capturing subpattern of that number has previously
1729 matched. An alternative notation is to precede the digits with a plus or minus
1730 sign. In this case, the subpattern number is relative rather than absolute.
1731 The most recently opened parentheses can be referenced by (?(-1), the next most
1732 recent by (?(-2), and so on. In looping constructs it can also make sense to
1733 refer to subsequent groups with constructs such as (?(+2).
1734 .P
1735 Consider the following pattern, which contains non-significant white space to
1736 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
1737 three parts for ease of discussion:
1738 .sp
1739 ( \e( )? [^()]+ (?(1) \e) )
1740 .sp
1741 The first part matches an optional opening parenthesis, and if that
1742 character is present, sets it as the first captured substring. The second part
1743 matches one or more characters that are not parentheses. The third part is a
1744 conditional subpattern that tests whether the first set of parentheses matched
1745 or not. If they did, that is, if subject started with an opening parenthesis,
1746 the condition is true, and so the yes-pattern is executed and a closing
1747 parenthesis is required. Otherwise, since no-pattern is not present, the
1748 subpattern matches nothing. In other words, this pattern matches a sequence of
1749 non-parentheses, optionally enclosed in parentheses.
1750 .P
1751 If you were embedding this pattern in a larger one, you could use a relative
1752 reference:
1753 .sp
1754 ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
1755 .sp
1756 This makes the fragment independent of the parentheses in the larger pattern.
1757 .
1758 .SS "Checking for a used subpattern by name"
1759 .rs
1760 .sp
1761 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
1762 subpattern by name. For compatibility with earlier versions of PCRE, which had
1763 this facility before Perl, the syntax (?(name)...) is also recognized. However,
1764 there is a possible ambiguity with this syntax, because subpattern names may
1765 consist entirely of digits. PCRE looks first for a named subpattern; if it
1766 cannot find one and the name consists entirely of digits, PCRE looks for a
1767 subpattern of that number, which must be greater than zero. Using subpattern
1768 names that consist entirely of digits is not recommended.
1769 .P
1770 Rewriting the above example to use a named subpattern gives this:
1771 .sp
1772 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
1773 .sp
1774 .
1775 .SS "Checking for pattern recursion"
1776 .rs
1777 .sp
1778 If the condition is the string (R), and there is no subpattern with the name R,
1779 the condition is true if a recursive call to the whole pattern or any
1780 subpattern has been made. If digits or a name preceded by ampersand follow the
1781 letter R, for example:
1782 .sp
1783 (?(R3)...) or (?(R&name)...)
1784 .sp
1785 the condition is true if the most recent recursion is into the subpattern whose
1786 number or name is given. This condition does not check the entire recursion
1787 stack.
1788 .P
1789 At "top level", all these recursion test conditions are false. Recursive
1790 patterns are described below.
1791 .
1792 .SS "Defining subpatterns for use by reference only"
1793 .rs
1794 .sp
1795 If the condition is the string (DEFINE), and there is no subpattern with the
1796 name DEFINE, the condition is always false. In this case, there may be only one
1797 alternative in the subpattern. It is always skipped if control reaches this
1798 point in the pattern; the idea of DEFINE is that it can be used to define
1799 "subroutines" that can be referenced from elsewhere. (The use of "subroutines"
1800 is described below.) For example, a pattern to match an IPv4 address could be
1801 written like this (ignore whitespace and line breaks):
1802 .sp
1803 (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
1804 \eb (?&byte) (\e.(?&byte)){3} \eb
1805 .sp
1806 The first part of the pattern is a DEFINE group inside which a another group
1807 named "byte" is defined. This matches an individual component of an IPv4
1808 address (a number less than 256). When matching takes place, this part of the
1809 pattern is skipped because DEFINE acts like a false condition.
1810 .P
1811 The rest of the pattern uses references to the named group to match the four
1812 dot-separated components of an IPv4 address, insisting on a word boundary at
1813 each end.
1814 .
1815 .SS "Assertion conditions"
1816 .rs
1817 .sp
1818 If the condition is not in any of the above formats, it must be an assertion.
1819 This may be a positive or negative lookahead or lookbehind assertion. Consider
1820 this pattern, again containing non-significant white space, and with the two
1821 alternatives on the second line:
1822 .sp
1823 (?(?=[^a-z]*[a-z])
1824 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
1825 .sp
1826 The condition is a positive lookahead assertion that matches an optional
1827 sequence of non-letters followed by a letter. In other words, it tests for the
1828 presence of at least one letter in the subject. If a letter is found, the
1829 subject is matched against the first alternative; otherwise it is matched
1830 against the second. This pattern matches strings in one of the two forms
1831 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
1832 .
1833 .
1834 .\" HTML <a name="comments"></a>
1836 .rs
1837 .sp
1838 The sequence (?# marks the start of a comment that continues up to the next
1839 closing parenthesis. Nested parentheses are not permitted. The characters
1840 that make up a comment play no part in the pattern matching at all.
1841 .P
1842 If the PCRE_EXTENDED option is set, an unescaped # character outside a
1843 character class introduces a comment that continues to immediately after the
1844 next newline in the pattern.
1845 .
1846 .
1847 .\" HTML <a name="recursion"></a>
1849 .rs
1850 .sp
1851 Consider the problem of matching a string in parentheses, allowing for
1852 unlimited nested parentheses. Without the use of recursion, the best that can
1853 be done is to use a pattern that matches up to some fixed depth of nesting. It
1854 is not possible to handle an arbitrary nesting depth.
1855 .P
1856 For some time, Perl has provided a facility that allows regular expressions to
1857 recurse (amongst other things). It does this by interpolating Perl code in the
1858 expression at run time, and the code can refer to the expression itself. A Perl
1859 pattern using code interpolation to solve the parentheses problem can be
1860 created like this:
1861 .sp
1862 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
1863 .sp
1864 The (?p{...}) item interpolates Perl code at run time, and in this case refers
1865 recursively to the pattern in which it appears.
1866 .P
1867 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
1868 supports special syntax for recursion of the entire pattern, and also for
1869 individual subpattern recursion. After its introduction in PCRE and Python,
1870 this kind of recursion was introduced into Perl at release 5.10.
1871 .P
1872 A special item that consists of (? followed by a number greater than zero and a
1873 closing parenthesis is a recursive call of the subpattern of the given number,
1874 provided that it occurs inside that subpattern. (If not, it is a "subroutine"
1875 call, which is described in the next section.) The special item (?R) or (?0) is
1876 a recursive call of the entire regular expression.
1877 .P
1878 In PCRE (like Python, but unlike Perl), a recursive subpattern call is always
1879 treated as an atomic group. That is, once it has matched some of the subject
1880 string, it is never re-entered, even if it contains untried alternatives and
1881 there is a subsequent matching failure.
1882 .P
1883 This PCRE pattern solves the nested parentheses problem (assume the
1884 PCRE_EXTENDED option is set so that white space is ignored):
1885 .sp
1886 \e( ( (?>[^()]+) | (?R) )* \e)
1887 .sp
1888 First it matches an opening parenthesis. Then it matches any number of
1889 substrings which can either be a sequence of non-parentheses, or a recursive
1890 match of the pattern itself (that is, a correctly parenthesized substring).
1891 Finally there is a closing parenthesis.
1892 .P
1893 If this were part of a larger pattern, you would not want to recurse the entire
1894 pattern, so instead you could use this:
1895 .sp
1896 ( \e( ( (?>[^()]+) | (?1) )* \e) )
1897 .sp
1898 We have put the pattern into parentheses, and caused the recursion to refer to
1899 them instead of the whole pattern.
1900 .P
1901 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
1902 is made easier by the use of relative references. (A Perl 5.10 feature.)
1903 Instead of (?1) in the pattern above you can write (?-2) to refer to the second
1904 most recently opened parentheses preceding the recursion. In other words, a
1905 negative number counts capturing parentheses leftwards from the point at which
1906 it is encountered.
1907 .P
1908 It is also possible to refer to subsequently opened parentheses, by writing
1909 references such as (?+2). However, these cannot be recursive because the
1910 reference is not inside the parentheses that are referenced. They are always
1911 "subroutine" calls, as described in the next section.
1912 .P
1913 An alternative approach is to use named parentheses instead. The Perl syntax
1914 for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
1915 could rewrite the above example as follows:
1916 .sp
1917 (?<pn> \e( ( (?>[^()]+) | (?&pn) )* \e) )
1918 .sp
1919 If there is more than one subpattern with the same name, the earliest one is
1920 used.
1921 .P
1922 This particular example pattern that we have been looking at contains nested
1923 unlimited repeats, and so the use of atomic grouping for matching strings of
1924 non-parentheses is important when applying the pattern to strings that do not
1925 match. For example, when this pattern is applied to
1926 .sp
1927 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
1928 .sp
1929 it yields "no match" quickly. However, if atomic grouping is not used,
1930 the match runs for a very long time indeed because there are so many different
1931 ways the + and * repeats can carve up the subject, and all have to be tested
1932 before failure can be reported.
1933 .P
1934 At the end of a match, the values set for any capturing subpatterns are those
1935 from the outermost level of the recursion at which the subpattern value is set.
1936 If you want to obtain intermediate values, a callout function can be used (see
1937 below and the
1938 .\" HREF
1939 \fBpcrecallout\fP
1940 .\"
1941 documentation). If the pattern above is matched against
1942 .sp
1943 (ab(cd)ef)
1944 .sp
1945 the value for the capturing parentheses is "ef", which is the last value taken
1946 on at the top level. If additional parentheses are added, giving
1947 .sp
1948 \e( ( ( (?>[^()]+) | (?R) )* ) \e)
1949 ^ ^
1950 ^ ^
1951 .sp
1952 the string they capture is "ab(cd)ef", the contents of the top level
1953 parentheses. If there are more than 15 capturing parentheses in a pattern, PCRE
1954 has to obtain extra memory to store data during a recursion, which it does by
1955 using \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no
1956 memory can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
1957 .P
1958 Do not confuse the (?R) item with the condition (R), which tests for recursion.
1959 Consider this pattern, which matches text in angle brackets, allowing for
1960 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
1961 recursing), whereas any characters are permitted at the outer level.
1962 .sp
1963 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
1964 .sp
1965 In this pattern, (?(R) is the start of a conditional subpattern, with two
1966 different alternatives for the recursive and non-recursive cases. The (?R) item
1967 is the actual recursive call.
1968 .
1969 .
1970 .\" HTML <a name="subpatternsassubroutines"></a>
1972 .rs
1973 .sp
1974 If the syntax for a recursive subpattern reference (either by number or by
1975 name) is used outside the parentheses to which it refers, it operates like a
1976 subroutine in a programming language. The "called" subpattern may be defined
1977 before or after the reference. A numbered reference can be absolute or
1978 relative, as in these examples:
1979 .sp
1980 (...(absolute)...)...(?2)...
1981 (...(relative)...)...(?-1)...
1982 (...(?+1)...(relative)...
1983 .sp
1984 An earlier example pointed out that the pattern
1985 .sp
1986 (sens|respons)e and \e1ibility
1987 .sp
1988 matches "sense and sensibility" and "response and responsibility", but not
1989 "sense and responsibility". If instead the pattern
1990 .sp
1991 (sens|respons)e and (?1)ibility
1992 .sp
1993 is used, it does match "sense and responsibility" as well as the other two
1994 strings. Another example is given in the discussion of DEFINE above.
1995 .P
1996 Like recursive subpatterns, a "subroutine" call is always treated as an atomic
1997 group. That is, once it has matched some of the subject string, it is never
1998 re-entered, even if it contains untried alternatives and there is a subsequent
1999 matching failure.
2000 .P
2001 When a subpattern is used as a subroutine, processing options such as
2002 case-independence are fixed when the subpattern is defined. They cannot be
2003 changed for different calls. For example, consider this pattern:
2004 .sp
2005 (abc)(?i:(?-1))
2006 .sp
2007 It matches "abcabc". It does not match "abcABC" because the change of
2008 processing option does not affect the called subpattern.
2009 .
2010 .
2012 .rs
2013 .sp
2014 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2015 code to be obeyed in the middle of matching a regular expression. This makes it
2016 possible, amongst other things, to extract different substrings that match the
2017 same pair of parentheses when there is a repetition.
2018 .P
2019 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2020 code. The feature is called "callout". The caller of PCRE provides an external
2021 function by putting its entry point in the global variable \fIpcre_callout\fP.
2022 By default, this variable contains NULL, which disables all calling out.
2023 .P
2024 Within a regular expression, (?C) indicates the points at which the external
2025 function is to be called. If you want to identify different callout points, you
2026 can put a number less than 256 after the letter C. The default value is zero.
2027 For example, this pattern has two callout points:
2028 .sp
2029 (?C1)abc(?C2)def
2030 .sp
2031 If the PCRE_AUTO_CALLOUT flag is passed to \fBpcre_compile()\fP, callouts are
2032 automatically installed before each item in the pattern. They are all numbered
2033 255.
2034 .P
2035 During matching, when PCRE reaches a callout point (and \fIpcre_callout\fP is
2036 set), the external function is called. It is provided with the number of the
2037 callout, the position in the pattern, and, optionally, one item of data
2038 originally supplied by the caller of \fBpcre_exec()\fP. The callout function
2039 may cause matching to proceed, to backtrack, or to fail altogether. A complete
2040 description of the interface to the callout function is given in the
2041 .\" HREF
2042 \fBpcrecallout\fP
2043 .\"
2044 documentation.
2045 .
2046 .
2048 .rs
2049 .sp
2050 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2051 are described in the Perl documentation as "experimental and subject to change
2052 or removal in a future version of Perl". It goes on to say: "Their usage in
2053 production code should be noted to avoid problems during upgrades." The same
2054 remarks apply to the PCRE features described in this section.
2055 .P
2056 Since these verbs are specifically related to backtracking, they can be used
2057 only when the pattern is to be matched using \fBpcre_exec()\fP, which uses a
2058 backtracking algorithm. They cause an error if encountered by
2059 \fBpcre_dfa_exec()\fP.
2060 .P
2061 The new verbs make use of what was previously invalid syntax: an opening
2062 parenthesis followed by an asterisk. In Perl, they are generally of the form
2063 (*VERB:ARG) but PCRE does not support the use of arguments, so its general
2064 form is just (*VERB). Any number of these verbs may occur in a pattern. There
2065 are two kinds:
2066 .
2067 .SS "Verbs that act immediately"
2068 .rs
2069 .sp
2070 The following verbs act as soon as they are encountered:
2071 .sp
2072 (*ACCEPT)
2073 .sp
2074 This verb causes the match to end successfully, skipping the remainder of the
2075 pattern. When inside a recursion, only the innermost pattern is ended
2076 immediately. PCRE differs from Perl in what happens if the (*ACCEPT) is inside
2077 capturing parentheses. In Perl, the data so far is captured: in PCRE no data is
2078 captured. For example:
2079 .sp
2080 A(A|B(*ACCEPT)|C)D
2081 .sp
2082 This matches "AB", "AAD", or "ACD", but when it matches "AB", no data is
2083 captured.
2084 .sp
2085 (*FAIL) or (*F)
2086 .sp
2087 This verb causes the match to fail, forcing backtracking to occur. It is
2088 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2089 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2090 Perl features that are not present in PCRE. The nearest equivalent is the
2091 callout feature, as for example in this pattern:
2092 .sp
2093 a+(?C)(*FAIL)
2094 .sp
2095 A match with the string "aaaa" always fails, but the callout is taken before
2096 each backtrack happens (in this example, 10 times).
2097 .
2098 .SS "Verbs that act after backtracking"
2099 .rs
2100 .sp
2101 The following verbs do nothing when they are encountered. Matching continues
2102 with what follows, but if there is no subsequent match, a failure is forced.
2103 The verbs differ in exactly what kind of failure occurs.
2104 .sp
2105 (*COMMIT)
2106 .sp
2107 This verb causes the whole match to fail outright if the rest of the pattern
2108 does not match. Even if the pattern is unanchored, no further attempts to find
2109 a match by advancing the start point take place. Once (*COMMIT) has been
2110 passed, \fBpcre_exec()\fP is committed to finding a match at the current
2111 starting point, or not at all. For example:
2112 .sp
2113 a+(*COMMIT)b
2114 .sp
2115 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2116 dynamic anchor, or "I've started, so I must finish."
2117 .sp
2118 (*PRUNE)
2119 .sp
2120 This verb causes the match to fail at the current position if the rest of the
2121 pattern does not match. If the pattern is unanchored, the normal "bumpalong"
2122 advance to the next starting character then happens. Backtracking can occur as
2123 usual to the left of (*PRUNE), or when matching to the right of (*PRUNE), but
2124 if there is no match to the right, backtracking cannot cross (*PRUNE).
2125 In simple cases, the use of (*PRUNE) is just an alternative to an atomic
2126 group or possessive quantifier, but there are some uses of (*PRUNE) that cannot
2127 be expressed in any other way.
2128 .sp
2129 (*SKIP)
2130 .sp
2131 This verb is like (*PRUNE), except that if the pattern is unanchored, the
2132 "bumpalong" advance is not to the next character, but to the position in the
2133 subject where (*SKIP) was encountered. (*SKIP) signifies that whatever text
2134 was matched leading up to it cannot be part of a successful match. Consider:
2135 .sp
2136 a+(*SKIP)b
2137 .sp
2138 If the subject is "aaaac...", after the first match attempt fails (starting at
2139 the first character in the string), the starting point skips on to start the
2140 next attempt at "c". Note that a possessive quantifer does not have the same
2141 effect in this example; although it would suppress backtracking during the
2142 first match attempt, the second attempt would start at the second character
2143 instead of skipping on to "c".
2144 .sp
2145 (*THEN)
2146 .sp
2147 This verb causes a skip to the next alternation if the rest of the pattern does
2148 not match. That is, it cancels pending backtracking, but only within the
2149 current alternation. Its name comes from the observation that it can be used
2150 for a pattern-based if-then-else block:
2151 .sp
2152 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2153 .sp
2154 If the COND1 pattern matches, FOO is tried (and possibly further items after
2155 the end of the group if FOO succeeds); on failure the matcher skips to the
2156 second alternative and tries COND2, without backtracking into COND1. If (*THEN)
2157 is used outside of any alternation, it acts exactly like (*PRUNE).
2158 .
2159 .
2160 .SH "SEE ALSO"
2161 .rs
2162 .sp
2163 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3), \fBpcre\fP(3).
2164 .
2165 .
2167 .rs
2168 .sp
2169 .nf
2170 Philip Hazel
2171 University Computing Service
2172 Cambridge CB2 3QH, England.
2173 .fi
2174 .
2175 .
2177 .rs
2178 .sp
2179 .nf
2180 Last updated: 11 September 2007
2181 Copyright (c) 1997-2007 University of Cambridge.
2182 .fi


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

  ViewVC Help
Powered by ViewVC 1.1.5