ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 211 - (show annotations)
Thu Aug 9 09:52:43 2007 UTC (13 years, 6 months ago) by ph10
File MIME type: text/html
File size: 89848 byte(s)
Update UTF-8 validity check and documentation.
1 <html>
2 <head>
3 <title>pcrepattern specification</title>
4 </head>
5 <body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB">
6 <h1>pcrepattern man page</h1>
7 <p>
8 Return to the <a href="index.html">PCRE index page</a>.
9 </p>
10 <p>
11 This page is part of the PCRE HTML documentation. It was generated automatically
12 from the original man page. If there is any nonsense in it, please consult the
13 man page, in case the conversion went wrong.
14 <br>
15 <ul>
16 <li><a name="TOC1" href="#SEC1">PCRE REGULAR EXPRESSION DETAILS</a>
17 <li><a name="TOC2" href="#SEC2">CHARACTERS AND METACHARACTERS</a>
18 <li><a name="TOC3" href="#SEC3">BACKSLASH</a>
19 <li><a name="TOC4" href="#SEC4">CIRCUMFLEX AND DOLLAR</a>
20 <li><a name="TOC5" href="#SEC5">FULL STOP (PERIOD, DOT)</a>
21 <li><a name="TOC6" href="#SEC6">MATCHING A SINGLE BYTE</a>
23 <li><a name="TOC8" href="#SEC8">POSIX CHARACTER CLASSES</a>
24 <li><a name="TOC9" href="#SEC9">VERTICAL BAR</a>
25 <li><a name="TOC10" href="#SEC10">INTERNAL OPTION SETTING</a>
26 <li><a name="TOC11" href="#SEC11">SUBPATTERNS</a>
27 <li><a name="TOC12" href="#SEC12">DUPLICATE SUBPATTERN NUMBERS</a>
28 <li><a name="TOC13" href="#SEC13">NAMED SUBPATTERNS</a>
29 <li><a name="TOC14" href="#SEC14">REPETITION</a>
31 <li><a name="TOC16" href="#SEC16">BACK REFERENCES</a>
32 <li><a name="TOC17" href="#SEC17">ASSERTIONS</a>
33 <li><a name="TOC18" href="#SEC18">CONDITIONAL SUBPATTERNS</a>
34 <li><a name="TOC19" href="#SEC19">COMMENTS</a>
35 <li><a name="TOC20" href="#SEC20">RECURSIVE PATTERNS</a>
36 <li><a name="TOC21" href="#SEC21">SUBPATTERNS AS SUBROUTINES</a>
37 <li><a name="TOC22" href="#SEC22">CALLOUTS</a>
38 <li><a name="TOC23" href="#SEC23">BACTRACKING CONTROL</a>
39 <li><a name="TOC24" href="#SEC24">SEE ALSO</a>
40 <li><a name="TOC25" href="#SEC25">AUTHOR</a>
41 <li><a name="TOC26" href="#SEC26">REVISION</a>
42 </ul>
43 <br><a name="SEC1" href="#TOC1">PCRE REGULAR EXPRESSION DETAILS</a><br>
44 <P>
45 The syntax and semantics of the regular expressions that are supported by PCRE
46 are described in detail below. There is a quick-reference syntax summary in the
47 <a href="pcresyntax.html"><b>pcresyntax</b></a>
48 page. Perl's regular expressions are described in its own documentation, and
49 regular expressions in general are covered in a number of books, some of which
50 have copious examples. Jeffrey Friedl's "Mastering Regular Expressions",
51 published by O'Reilly, covers regular expressions in great detail. This
52 description of PCRE's regular expressions is intended as reference material.
53 </P>
54 <P>
55 The original operation of PCRE was on strings of one-byte characters. However,
56 there is now also support for UTF-8 character strings. To use this, you must
57 build PCRE to include UTF-8 support, and then call <b>pcre_compile()</b> with
58 the PCRE_UTF8 option. How this affects pattern matching is mentioned in several
59 places below. There is also a summary of UTF-8 features in the
60 <a href="pcre.html#utf8support">section on UTF-8 support</a>
61 in the main
62 <a href="pcre.html"><b>pcre</b></a>
63 page.
64 </P>
65 <P>
66 The remainder of this document discusses the patterns that are supported by
67 PCRE when its main matching function, <b>pcre_exec()</b>, is used.
68 From release 6.0, PCRE offers a second matching function,
69 <b>pcre_dfa_exec()</b>, which matches using a different algorithm that is not
70 Perl-compatible. Some of the features discussed below are not available when
71 <b>pcre_dfa_exec()</b> is used. The advantages and disadvantages of the
72 alternative function, and how it differs from the normal function, are
73 discussed in the
74 <a href="pcrematching.html"><b>pcrematching</b></a>
75 page.
76 </P>
77 <br><a name="SEC2" href="#TOC1">CHARACTERS AND METACHARACTERS</a><br>
78 <P>
79 A regular expression is a pattern that is matched against a subject string from
80 left to right. Most characters stand for themselves in a pattern, and match the
81 corresponding characters in the subject. As a trivial example, the pattern
82 <pre>
83 The quick brown fox
84 </pre>
85 matches a portion of a subject string that is identical to itself. When
86 caseless matching is specified (the PCRE_CASELESS option), letters are matched
87 independently of case. In UTF-8 mode, PCRE always understands the concept of
88 case for characters whose values are less than 128, so caseless matching is
89 always possible. For characters with higher values, the concept of case is
90 supported if PCRE is compiled with Unicode property support, but not otherwise.
91 If you want to use caseless matching for characters 128 and above, you must
92 ensure that PCRE is compiled with Unicode property support as well as with
93 UTF-8 support.
94 </P>
95 <P>
96 The power of regular expressions comes from the ability to include alternatives
97 and repetitions in the pattern. These are encoded in the pattern by the use of
98 <i>metacharacters</i>, which do not stand for themselves but instead are
99 interpreted in some special way.
100 </P>
101 <P>
102 There are two different sets of metacharacters: those that are recognized
103 anywhere in the pattern except within square brackets, and those that are
104 recognized within square brackets. Outside square brackets, the metacharacters
105 are as follows:
106 <pre>
107 \ general escape character with several uses
108 ^ assert start of string (or line, in multiline mode)
109 $ assert end of string (or line, in multiline mode)
110 . match any character except newline (by default)
111 [ start character class definition
112 | start of alternative branch
113 ( start subpattern
114 ) end subpattern
115 ? extends the meaning of (
116 also 0 or 1 quantifier
117 also quantifier minimizer
118 * 0 or more quantifier
119 + 1 or more quantifier
120 also "possessive quantifier"
121 { start min/max quantifier
122 </pre>
123 Part of a pattern that is in square brackets is called a "character class". In
124 a character class the only metacharacters are:
125 <pre>
126 \ general escape character
127 ^ negate the class, but only if the first character
128 - indicates character range
129 [ POSIX character class (only if followed by POSIX syntax)
130 ] terminates the character class
131 </pre>
132 The following sections describe the use of each of the metacharacters.
133 </P>
134 <br><a name="SEC3" href="#TOC1">BACKSLASH</a><br>
135 <P>
136 The backslash character has several uses. Firstly, if it is followed by a
137 non-alphanumeric character, it takes away any special meaning that character
138 may have. This use of backslash as an escape character applies both inside and
139 outside character classes.
140 </P>
141 <P>
142 For example, if you want to match a * character, you write \* in the pattern.
143 This escaping action applies whether or not the following character would
144 otherwise be interpreted as a metacharacter, so it is always safe to precede a
145 non-alphanumeric with backslash to specify that it stands for itself. In
146 particular, if you want to match a backslash, you write \\.
147 </P>
148 <P>
149 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
150 pattern (other than in a character class) and characters between a # outside
151 a character class and the next newline are ignored. An escaping backslash can
152 be used to include a whitespace or # character as part of the pattern.
153 </P>
154 <P>
155 If you want to remove the special meaning from a sequence of characters, you
156 can do so by putting them between \Q and \E. This is different from Perl in
157 that $ and @ are handled as literals in \Q...\E sequences in PCRE, whereas in
158 Perl, $ and @ cause variable interpolation. Note the following examples:
159 <pre>
160 Pattern PCRE matches Perl matches
162 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
163 \Qabc\$xyz\E abc\$xyz abc\$xyz
164 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
165 </pre>
166 The \Q...\E sequence is recognized both inside and outside character classes.
167 <a name="digitsafterbackslash"></a></P>
168 <br><b>
169 Non-printing characters
170 </b><br>
171 <P>
172 A second use of backslash provides a way of encoding non-printing characters
173 in patterns in a visible manner. There is no restriction on the appearance of
174 non-printing characters, apart from the binary zero that terminates a pattern,
175 but when a pattern is being prepared by text editing, it is usually easier to
176 use one of the following escape sequences than the binary character it
177 represents:
178 <pre>
179 \a alarm, that is, the BEL character (hex 07)
180 \cx "control-x", where x is any character
181 \e escape (hex 1B)
182 \f formfeed (hex 0C)
183 \n newline (hex 0A)
184 \r carriage return (hex 0D)
185 \t tab (hex 09)
186 \ddd character with octal code ddd, or backreference
187 \xhh character with hex code hh
188 \x{hhh..} character with hex code hhh..
189 </pre>
190 The precise effect of \cx is as follows: if x is a lower case letter, it
191 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
192 Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c; becomes hex
193 7B.
194 </P>
195 <P>
196 After \x, from zero to two hexadecimal digits are read (letters can be in
197 upper or lower case). Any number of hexadecimal digits may appear between \x{
198 and }, but the value of the character code must be less than 256 in non-UTF-8
199 mode, and less than 2**31 in UTF-8 mode. That is, the maximum value in
200 hexadecimal is 7FFFFFFF. Note that this is bigger than the largest Unicode code
201 point, which is 10FFFF.
202 </P>
203 <P>
204 If characters other than hexadecimal digits appear between \x{ and }, or if
205 there is no terminating }, this form of escape is not recognized. Instead, the
206 initial \x will be interpreted as a basic hexadecimal escape, with no
207 following digits, giving a character whose value is zero.
208 </P>
209 <P>
210 Characters whose value is less than 256 can be defined by either of the two
211 syntaxes for \x. There is no difference in the way they are handled. For
212 example, \xdc is exactly the same as \x{dc}.
213 </P>
214 <P>
215 After \0 up to two further octal digits are read. If there are fewer than two
216 digits, just those that are present are used. Thus the sequence \0\x\07
217 specifies two binary zeros followed by a BEL character (code value 7). Make
218 sure you supply two digits after the initial zero if the pattern character that
219 follows is itself an octal digit.
220 </P>
221 <P>
222 The handling of a backslash followed by a digit other than 0 is complicated.
223 Outside a character class, PCRE reads it and any following digits as a decimal
224 number. If the number is less than 10, or if there have been at least that many
225 previous capturing left parentheses in the expression, the entire sequence is
226 taken as a <i>back reference</i>. A description of how this works is given
227 <a href="#backreferences">later,</a>
228 following the discussion of
229 <a href="#subpattern">parenthesized subpatterns.</a>
230 </P>
231 <P>
232 Inside a character class, or if the decimal number is greater than 9 and there
233 have not been that many capturing subpatterns, PCRE re-reads up to three octal
234 digits following the backslash, and uses them to generate a data character. Any
235 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
236 character specified in octal must be less than \400. In UTF-8 mode, values up
237 to \777 are permitted. For example:
238 <pre>
239 \040 is another way of writing a space
240 \40 is the same, provided there are fewer than 40 previous capturing subpatterns
241 \7 is always a back reference
242 \11 might be a back reference, or another way of writing a tab
243 \011 is always a tab
244 \0113 is a tab followed by the character "3"
245 \113 might be a back reference, otherwise the character with octal code 113
246 \377 might be a back reference, otherwise the byte consisting entirely of 1 bits
247 \81 is either a back reference, or a binary zero followed by the two characters "8" and "1"
248 </pre>
249 Note that octal values of 100 or greater must not be introduced by a leading
250 zero, because no more than three octal digits are ever read.
251 </P>
252 <P>
253 All the sequences that define a single character value can be used both inside
254 and outside character classes. In addition, inside a character class, the
255 sequence \b is interpreted as the backspace character (hex 08), and the
256 sequences \R and \X are interpreted as the characters "R" and "X",
257 respectively. Outside a character class, these sequences have different
258 meanings
259 <a href="#uniextseq">(see below).</a>
260 </P>
261 <br><b>
262 Absolute and relative back references
263 </b><br>
264 <P>
265 The sequence \g followed by an unsigned or a negative number, optionally
266 enclosed in braces, is an absolute or relative back reference. A named back
267 reference can be coded as \g{name}. Back references are discussed
268 <a href="#backreferences">later,</a>
269 following the discussion of
270 <a href="#subpattern">parenthesized subpatterns.</a>
271 </P>
272 <br><b>
273 Generic character types
274 </b><br>
275 <P>
276 Another use of backslash is for specifying generic character types. The
277 following are always recognized:
278 <pre>
279 \d any decimal digit
280 \D any character that is not a decimal digit
281 \h any horizontal whitespace character
282 \H any character that is not a horizontal whitespace character
283 \s any whitespace character
284 \S any character that is not a whitespace character
285 \v any vertical whitespace character
286 \V any character that is not a vertical whitespace character
287 \w any "word" character
288 \W any "non-word" character
289 </pre>
290 Each pair of escape sequences partitions the complete set of characters into
291 two disjoint sets. Any given character matches one, and only one, of each pair.
292 </P>
293 <P>
294 These character type sequences can appear both inside and outside character
295 classes. They each match one character of the appropriate type. If the current
296 matching point is at the end of the subject string, all of them fail, since
297 there is no character to match.
298 </P>
299 <P>
300 For compatibility with Perl, \s does not match the VT character (code 11).
301 This makes it different from the the POSIX "space" class. The \s characters
302 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
303 included in a Perl script, \s may match the VT character. In PCRE, it never
304 does.
305 </P>
306 <P>
307 In UTF-8 mode, characters with values greater than 128 never match \d, \s, or
308 \w, and always match \D, \S, and \W. This is true even when Unicode
309 character property support is available. These sequences retain their original
310 meanings from before UTF-8 support was available, mainly for efficiency
311 reasons.
312 </P>
313 <P>
314 The sequences \h, \H, \v, and \V are Perl 5.10 features. In contrast to the
315 other sequences, these do match certain high-valued codepoints in UTF-8 mode.
316 The horizontal space characters are:
317 <pre>
318 U+0009 Horizontal tab
319 U+0020 Space
320 U+00A0 Non-break space
321 U+1680 Ogham space mark
322 U+180E Mongolian vowel separator
323 U+2000 En quad
324 U+2001 Em quad
325 U+2002 En space
326 U+2003 Em space
327 U+2004 Three-per-em space
328 U+2005 Four-per-em space
329 U+2006 Six-per-em space
330 U+2007 Figure space
331 U+2008 Punctuation space
332 U+2009 Thin space
333 U+200A Hair space
334 U+202F Narrow no-break space
335 U+205F Medium mathematical space
336 U+3000 Ideographic space
337 </pre>
338 The vertical space characters are:
339 <pre>
340 U+000A Linefeed
341 U+000B Vertical tab
342 U+000C Formfeed
343 U+000D Carriage return
344 U+0085 Next line
345 U+2028 Line separator
346 U+2029 Paragraph separator
347 </PRE>
348 </P>
349 <P>
350 A "word" character is an underscore or any character less than 256 that is a
351 letter or digit. The definition of letters and digits is controlled by PCRE's
352 low-valued character tables, and may vary if locale-specific matching is taking
353 place (see
354 <a href="pcreapi.html#localesupport">"Locale support"</a>
355 in the
356 <a href="pcreapi.html"><b>pcreapi</b></a>
357 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
358 or "french" in Windows, some character codes greater than 128 are used for
359 accented letters, and these are matched by \w. The use of locales with Unicode
360 is discouraged.
361 </P>
362 <br><b>
363 Newline sequences
364 </b><br>
365 <P>
366 Outside a character class, the escape sequence \R matches any Unicode newline
367 sequence. This is a Perl 5.10 feature. In non-UTF-8 mode \R is equivalent to
368 the following:
369 <pre>
370 (?&#62;\r\n|\n|\x0b|\f|\r|\x85)
371 </pre>
372 This is an example of an "atomic group", details of which are given
373 <a href="#atomicgroup">below.</a>
374 This particular group matches either the two-character sequence CR followed by
375 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
376 U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
377 line, U+0085). The two-character sequence is treated as a single unit that
378 cannot be split.
379 </P>
380 <P>
381 In UTF-8 mode, two additional characters whose codepoints are greater than 255
382 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
383 Unicode character property support is not needed for these characters to be
384 recognized.
385 </P>
386 <P>
387 Inside a character class, \R matches the letter "R".
388 <a name="uniextseq"></a></P>
389 <br><b>
390 Unicode character properties
391 </b><br>
392 <P>
393 When PCRE is built with Unicode character property support, three additional
394 escape sequences that match characters with specific properties are available.
395 When not in UTF-8 mode, these sequences are of course limited to testing
396 characters whose codepoints are less than 256, but they do work in this mode.
397 The extra escape sequences are:
398 <pre>
399 \p{<i>xx</i>} a character with the <i>xx</i> property
400 \P{<i>xx</i>} a character without the <i>xx</i> property
401 \X an extended Unicode sequence
402 </pre>
403 The property names represented by <i>xx</i> above are limited to the Unicode
404 script names, the general category properties, and "Any", which matches any
405 character (including newline). Other properties such as "InMusicalSymbols" are
406 not currently supported by PCRE. Note that \P{Any} does not match any
407 characters, so always causes a match failure.
408 </P>
409 <P>
410 Sets of Unicode characters are defined as belonging to certain scripts. A
411 character from one of these sets can be matched using a script name. For
412 example:
413 <pre>
414 \p{Greek}
415 \P{Han}
416 </pre>
417 Those that are not part of an identified script are lumped together as
418 "Common". The current list of scripts is:
419 </P>
420 <P>
421 Arabic,
422 Armenian,
423 Balinese,
424 Bengali,
425 Bopomofo,
426 Braille,
427 Buginese,
428 Buhid,
429 Canadian_Aboriginal,
430 Cherokee,
431 Common,
432 Coptic,
433 Cuneiform,
434 Cypriot,
435 Cyrillic,
436 Deseret,
437 Devanagari,
438 Ethiopic,
439 Georgian,
440 Glagolitic,
441 Gothic,
442 Greek,
443 Gujarati,
444 Gurmukhi,
445 Han,
446 Hangul,
447 Hanunoo,
448 Hebrew,
449 Hiragana,
450 Inherited,
451 Kannada,
452 Katakana,
453 Kharoshthi,
454 Khmer,
455 Lao,
456 Latin,
457 Limbu,
458 Linear_B,
459 Malayalam,
460 Mongolian,
461 Myanmar,
462 New_Tai_Lue,
463 Nko,
464 Ogham,
465 Old_Italic,
466 Old_Persian,
467 Oriya,
468 Osmanya,
469 Phags_Pa,
470 Phoenician,
471 Runic,
472 Shavian,
473 Sinhala,
474 Syloti_Nagri,
475 Syriac,
476 Tagalog,
477 Tagbanwa,
478 Tai_Le,
479 Tamil,
480 Telugu,
481 Thaana,
482 Thai,
483 Tibetan,
484 Tifinagh,
485 Ugaritic,
486 Yi.
487 </P>
488 <P>
489 Each character has exactly one general category property, specified by a
490 two-letter abbreviation. For compatibility with Perl, negation can be specified
491 by including a circumflex between the opening brace and the property name. For
492 example, \p{^Lu} is the same as \P{Lu}.
493 </P>
494 <P>
495 If only one letter is specified with \p or \P, it includes all the general
496 category properties that start with that letter. In this case, in the absence
497 of negation, the curly brackets in the escape sequence are optional; these two
498 examples have the same effect:
499 <pre>
500 \p{L}
501 \pL
502 </pre>
503 The following general category property codes are supported:
504 <pre>
505 C Other
506 Cc Control
507 Cf Format
508 Cn Unassigned
509 Co Private use
510 Cs Surrogate
512 L Letter
513 Ll Lower case letter
514 Lm Modifier letter
515 Lo Other letter
516 Lt Title case letter
517 Lu Upper case letter
519 M Mark
520 Mc Spacing mark
521 Me Enclosing mark
522 Mn Non-spacing mark
524 N Number
525 Nd Decimal number
526 Nl Letter number
527 No Other number
529 P Punctuation
530 Pc Connector punctuation
531 Pd Dash punctuation
532 Pe Close punctuation
533 Pf Final punctuation
534 Pi Initial punctuation
535 Po Other punctuation
536 Ps Open punctuation
538 S Symbol
539 Sc Currency symbol
540 Sk Modifier symbol
541 Sm Mathematical symbol
542 So Other symbol
544 Z Separator
545 Zl Line separator
546 Zp Paragraph separator
547 Zs Space separator
548 </pre>
549 The special property L& is also supported: it matches a character that has
550 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
551 a modifier or "other".
552 </P>
553 <P>
554 The Cs (Surrogate) property applies only to characters in the range U+D800 to
555 U+DFFF. Such characters are not valid in UTF-8 strings (see RFC 3629) and so
556 cannot be tested by PCRE, unless UTF-8 validity checking has been turned off
557 (see the discussion of PCRE_NO_UTF8_CHECK in the
558 <a href="pcreapi.html"><b>pcreapi</b></a>
559 page).
560 </P>
561 <P>
562 The long synonyms for these properties that Perl supports (such as \p{Letter})
563 are not supported by PCRE, nor is it permitted to prefix any of these
564 properties with "Is".
565 </P>
566 <P>
567 No character that is in the Unicode table has the Cn (unassigned) property.
568 Instead, this property is assumed for any code point that is not in the
569 Unicode table.
570 </P>
571 <P>
572 Specifying caseless matching does not affect these escape sequences. For
573 example, \p{Lu} always matches only upper case letters.
574 </P>
575 <P>
576 The \X escape matches any number of Unicode characters that form an extended
577 Unicode sequence. \X is equivalent to
578 <pre>
579 (?&#62;\PM\pM*)
580 </pre>
581 That is, it matches a character without the "mark" property, followed by zero
582 or more characters with the "mark" property, and treats the sequence as an
583 atomic group
584 <a href="#atomicgroup">(see below).</a>
585 Characters with the "mark" property are typically accents that affect the
586 preceding character. None of them have codepoints less than 256, so in
587 non-UTF-8 mode \X matches any one character.
588 </P>
589 <P>
590 Matching characters by Unicode property is not fast, because PCRE has to search
591 a structure that contains data for over fifteen thousand characters. That is
592 why the traditional escape sequences such as \d and \w do not use Unicode
593 properties in PCRE.
594 <a name="resetmatchstart"></a></P>
595 <br><b>
596 Resetting the match start
597 </b><br>
598 <P>
599 The escape sequence \K, which is a Perl 5.10 feature, causes any previously
600 matched characters not to be included in the final matched sequence. For
601 example, the pattern:
602 <pre>
603 foo\Kbar
604 </pre>
605 matches "foobar", but reports that it has matched "bar". This feature is
606 similar to a lookbehind assertion
607 <a href="#lookbehind">(described below).</a>
608 However, in this case, the part of the subject before the real match does not
609 have to be of fixed length, as lookbehind assertions do. The use of \K does
610 not interfere with the setting of
611 <a href="#subpattern">captured substrings.</a>
612 For example, when the pattern
613 <pre>
614 (foo)\Kbar
615 </pre>
616 matches "foobar", the first substring is still set to "foo".
617 <a name="smallassertions"></a></P>
618 <br><b>
619 Simple assertions
620 </b><br>
621 <P>
622 The final use of backslash is for certain simple assertions. An assertion
623 specifies a condition that has to be met at a particular point in a match,
624 without consuming any characters from the subject string. The use of
625 subpatterns for more complicated assertions is described
626 <a href="#bigassertions">below.</a>
627 The backslashed assertions are:
628 <pre>
629 \b matches at a word boundary
630 \B matches when not at a word boundary
631 \A matches at the start of the subject
632 \Z matches at the end of the subject
633 also matches before a newline at the end of the subject
634 \z matches only at the end of the subject
635 \G matches at the first matching position in the subject
636 </pre>
637 These assertions may not appear in character classes (but note that \b has a
638 different meaning, namely the backspace character, inside a character class).
639 </P>
640 <P>
641 A word boundary is a position in the subject string where the current character
642 and the previous character do not both match \w or \W (i.e. one matches
643 \w and the other matches \W), or the start or end of the string if the
644 first or last character matches \w, respectively.
645 </P>
646 <P>
647 The \A, \Z, and \z assertions differ from the traditional circumflex and
648 dollar (described in the next section) in that they only ever match at the very
649 start and end of the subject string, whatever options are set. Thus, they are
650 independent of multiline mode. These three assertions are not affected by the
651 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
652 circumflex and dollar metacharacters. However, if the <i>startoffset</i>
653 argument of <b>pcre_exec()</b> is non-zero, indicating that matching is to start
654 at a point other than the beginning of the subject, \A can never match. The
655 difference between \Z and \z is that \Z matches before a newline at the end
656 of the string as well as at the very end, whereas \z matches only at the end.
657 </P>
658 <P>
659 The \G assertion is true only when the current matching position is at the
660 start point of the match, as specified by the <i>startoffset</i> argument of
661 <b>pcre_exec()</b>. It differs from \A when the value of <i>startoffset</i> is
662 non-zero. By calling <b>pcre_exec()</b> multiple times with appropriate
663 arguments, you can mimic Perl's /g option, and it is in this kind of
664 implementation where \G can be useful.
665 </P>
666 <P>
667 Note, however, that PCRE's interpretation of \G, as the start of the current
668 match, is subtly different from Perl's, which defines it as the end of the
669 previous match. In Perl, these can be different when the previously matched
670 string was empty. Because PCRE does just one match at a time, it cannot
671 reproduce this behaviour.
672 </P>
673 <P>
674 If all the alternatives of a pattern begin with \G, the expression is anchored
675 to the starting match position, and the "anchored" flag is set in the compiled
676 regular expression.
677 </P>
678 <br><a name="SEC4" href="#TOC1">CIRCUMFLEX AND DOLLAR</a><br>
679 <P>
680 Outside a character class, in the default matching mode, the circumflex
681 character is an assertion that is true only if the current matching point is
682 at the start of the subject string. If the <i>startoffset</i> argument of
683 <b>pcre_exec()</b> is non-zero, circumflex can never match if the PCRE_MULTILINE
684 option is unset. Inside a character class, circumflex has an entirely different
685 meaning
686 <a href="#characterclass">(see below).</a>
687 </P>
688 <P>
689 Circumflex need not be the first character of the pattern if a number of
690 alternatives are involved, but it should be the first thing in each alternative
691 in which it appears if the pattern is ever to match that branch. If all
692 possible alternatives start with a circumflex, that is, if the pattern is
693 constrained to match only at the start of the subject, it is said to be an
694 "anchored" pattern. (There are also other constructs that can cause a pattern
695 to be anchored.)
696 </P>
697 <P>
698 A dollar character is an assertion that is true only if the current matching
699 point is at the end of the subject string, or immediately before a newline
700 at the end of the string (by default). Dollar need not be the last character of
701 the pattern if a number of alternatives are involved, but it should be the last
702 item in any branch in which it appears. Dollar has no special meaning in a
703 character class.
704 </P>
705 <P>
706 The meaning of dollar can be changed so that it matches only at the very end of
707 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
708 does not affect the \Z assertion.
709 </P>
710 <P>
711 The meanings of the circumflex and dollar characters are changed if the
712 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
713 immediately after internal newlines as well as at the start of the subject
714 string. It does not match after a newline that ends the string. A dollar
715 matches before any newlines in the string, as well as at the very end, when
716 PCRE_MULTILINE is set. When newline is specified as the two-character
717 sequence CRLF, isolated CR and LF characters do not indicate newlines.
718 </P>
719 <P>
720 For example, the pattern /^abc$/ matches the subject string "def\nabc" (where
721 \n represents a newline) in multiline mode, but not otherwise. Consequently,
722 patterns that are anchored in single line mode because all branches start with
723 ^ are not anchored in multiline mode, and a match for circumflex is possible
724 when the <i>startoffset</i> argument of <b>pcre_exec()</b> is non-zero. The
725 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
726 </P>
727 <P>
728 Note that the sequences \A, \Z, and \z can be used to match the start and
729 end of the subject in both modes, and if all branches of a pattern start with
730 \A it is always anchored, whether or not PCRE_MULTILINE is set.
731 </P>
732 <br><a name="SEC5" href="#TOC1">FULL STOP (PERIOD, DOT)</a><br>
733 <P>
734 Outside a character class, a dot in the pattern matches any one character in
735 the subject string except (by default) a character that signifies the end of a
736 line. In UTF-8 mode, the matched character may be more than one byte long.
737 </P>
738 <P>
739 When a line ending is defined as a single character, dot never matches that
740 character; when the two-character sequence CRLF is used, dot does not match CR
741 if it is immediately followed by LF, but otherwise it matches all characters
742 (including isolated CRs and LFs). When any Unicode line endings are being
743 recognized, dot does not match CR or LF or any of the other line ending
744 characters.
745 </P>
746 <P>
747 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
748 option is set, a dot matches any one character, without exception. If the
749 two-character sequence CRLF is present in the subject string, it takes two dots
750 to match it.
751 </P>
752 <P>
753 The handling of dot is entirely independent of the handling of circumflex and
754 dollar, the only relationship being that they both involve newlines. Dot has no
755 special meaning in a character class.
756 </P>
757 <br><a name="SEC6" href="#TOC1">MATCHING A SINGLE BYTE</a><br>
758 <P>
759 Outside a character class, the escape sequence \C matches any one byte, both
760 in and out of UTF-8 mode. Unlike a dot, it always matches any line-ending
761 characters. The feature is provided in Perl in order to match individual bytes
762 in UTF-8 mode. Because it breaks up UTF-8 characters into individual bytes,
763 what remains in the string may be a malformed UTF-8 string. For this reason,
764 the \C escape sequence is best avoided.
765 </P>
766 <P>
767 PCRE does not allow \C to appear in lookbehind assertions
768 <a href="#lookbehind">(described below),</a>
769 because in UTF-8 mode this would make it impossible to calculate the length of
770 the lookbehind.
771 <a name="characterclass"></a></P>
772 <br><a name="SEC7" href="#TOC1">SQUARE BRACKETS AND CHARACTER CLASSES</a><br>
773 <P>
774 An opening square bracket introduces a character class, terminated by a closing
775 square bracket. A closing square bracket on its own is not special. If a
776 closing square bracket is required as a member of the class, it should be the
777 first data character in the class (after an initial circumflex, if present) or
778 escaped with a backslash.
779 </P>
780 <P>
781 A character class matches a single character in the subject. In UTF-8 mode, the
782 character may occupy more than one byte. A matched character must be in the set
783 of characters defined by the class, unless the first character in the class
784 definition is a circumflex, in which case the subject character must not be in
785 the set defined by the class. If a circumflex is actually required as a member
786 of the class, ensure it is not the first character, or escape it with a
787 backslash.
788 </P>
789 <P>
790 For example, the character class [aeiou] matches any lower case vowel, while
791 [^aeiou] matches any character that is not a lower case vowel. Note that a
792 circumflex is just a convenient notation for specifying the characters that
793 are in the class by enumerating those that are not. A class that starts with a
794 circumflex is not an assertion: it still consumes a character from the subject
795 string, and therefore it fails if the current pointer is at the end of the
796 string.
797 </P>
798 <P>
799 In UTF-8 mode, characters with values greater than 255 can be included in a
800 class as a literal string of bytes, or by using the \x{ escaping mechanism.
801 </P>
802 <P>
803 When caseless matching is set, any letters in a class represent both their
804 upper case and lower case versions, so for example, a caseless [aeiou] matches
805 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
806 caseful version would. In UTF-8 mode, PCRE always understands the concept of
807 case for characters whose values are less than 128, so caseless matching is
808 always possible. For characters with higher values, the concept of case is
809 supported if PCRE is compiled with Unicode property support, but not otherwise.
810 If you want to use caseless matching for characters 128 and above, you must
811 ensure that PCRE is compiled with Unicode property support as well as with
812 UTF-8 support.
813 </P>
814 <P>
815 Characters that might indicate line breaks are never treated in any special way
816 when matching character classes, whatever line-ending sequence is in use, and
817 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
818 such as [^a] always matches one of these characters.
819 </P>
820 <P>
821 The minus (hyphen) character can be used to specify a range of characters in a
822 character class. For example, [d-m] matches any letter between d and m,
823 inclusive. If a minus character is required in a class, it must be escaped with
824 a backslash or appear in a position where it cannot be interpreted as
825 indicating a range, typically as the first or last character in the class.
826 </P>
827 <P>
828 It is not possible to have the literal character "]" as the end character of a
829 range. A pattern such as [W-]46] is interpreted as a class of two characters
830 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
831 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
832 the end of range, so [W-\]46] is interpreted as a class containing a range
833 followed by two other characters. The octal or hexadecimal representation of
834 "]" can also be used to end a range.
835 </P>
836 <P>
837 Ranges operate in the collating sequence of character values. They can also be
838 used for characters specified numerically, for example [\000-\037]. In UTF-8
839 mode, ranges can include characters whose values are greater than 255, for
840 example [\x{100}-\x{2ff}].
841 </P>
842 <P>
843 If a range that includes letters is used when caseless matching is set, it
844 matches the letters in either case. For example, [W-c] is equivalent to
845 [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
846 tables for a French locale are in use, [\xc8-\xcb] matches accented E
847 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
848 characters with values greater than 128 only when it is compiled with Unicode
849 property support.
850 </P>
851 <P>
852 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
853 in a character class, and add the characters that they match to the class. For
854 example, [\dABCDEF] matches any hexadecimal digit. A circumflex can
855 conveniently be used with the upper case character types to specify a more
856 restricted set of characters than the matching lower case type. For example,
857 the class [^\W_] matches any letter or digit, but not underscore.
858 </P>
859 <P>
860 The only metacharacters that are recognized in character classes are backslash,
861 hyphen (only where it can be interpreted as specifying a range), circumflex
862 (only at the start), opening square bracket (only when it can be interpreted as
863 introducing a POSIX class name - see the next section), and the terminating
864 closing square bracket. However, escaping other non-alphanumeric characters
865 does no harm.
866 </P>
867 <br><a name="SEC8" href="#TOC1">POSIX CHARACTER CLASSES</a><br>
868 <P>
869 Perl supports the POSIX notation for character classes. This uses names
870 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
871 this notation. For example,
872 <pre>
873 [01[:alpha:]%]
874 </pre>
875 matches "0", "1", any alphabetic character, or "%". The supported class names
876 are
877 <pre>
878 alnum letters and digits
879 alpha letters
880 ascii character codes 0 - 127
881 blank space or tab only
882 cntrl control characters
883 digit decimal digits (same as \d)
884 graph printing characters, excluding space
885 lower lower case letters
886 print printing characters, including space
887 punct printing characters, excluding letters and digits
888 space white space (not quite the same as \s)
889 upper upper case letters
890 word "word" characters (same as \w)
891 xdigit hexadecimal digits
892 </pre>
893 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
894 space (32). Notice that this list includes the VT character (code 11). This
895 makes "space" different to \s, which does not include VT (for Perl
896 compatibility).
897 </P>
898 <P>
899 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
900 5.8. Another Perl extension is negation, which is indicated by a ^ character
901 after the colon. For example,
902 <pre>
903 [12[:^digit:]]
904 </pre>
905 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
906 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
907 supported, and an error is given if they are encountered.
908 </P>
909 <P>
910 In UTF-8 mode, characters with values greater than 128 do not match any of
911 the POSIX character classes.
912 </P>
913 <br><a name="SEC9" href="#TOC1">VERTICAL BAR</a><br>
914 <P>
915 Vertical bar characters are used to separate alternative patterns. For example,
916 the pattern
917 <pre>
918 gilbert|sullivan
919 </pre>
920 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
921 and an empty alternative is permitted (matching the empty string). The matching
922 process tries each alternative in turn, from left to right, and the first one
923 that succeeds is used. If the alternatives are within a subpattern
924 <a href="#subpattern">(defined below),</a>
925 "succeeds" means matching the rest of the main pattern as well as the
926 alternative in the subpattern.
927 </P>
928 <br><a name="SEC10" href="#TOC1">INTERNAL OPTION SETTING</a><br>
929 <P>
931 PCRE_EXTENDED options can be changed from within the pattern by a sequence of
932 Perl option letters enclosed between "(?" and ")". The option letters are
933 <pre>
936 s for PCRE_DOTALL
938 </pre>
939 For example, (?im) sets caseless, multiline matching. It is also possible to
940 unset these options by preceding the letter with a hyphen, and a combined
941 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
942 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
943 permitted. If a letter appears both before and after the hyphen, the option is
944 unset.
945 </P>
946 <P>
947 When an option change occurs at top level (that is, not inside subpattern
948 parentheses), the change applies to the remainder of the pattern that follows.
949 If the change is placed right at the start of a pattern, PCRE extracts it into
950 the global options (and it will therefore show up in data extracted by the
951 <b>pcre_fullinfo()</b> function).
952 </P>
953 <P>
954 An option change within a subpattern (see below for a description of
955 subpatterns) affects only that part of the current pattern that follows it, so
956 <pre>
957 (a(?i)b)c
958 </pre>
959 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
960 By this means, options can be made to have different settings in different
961 parts of the pattern. Any changes made in one alternative do carry on
962 into subsequent branches within the same subpattern. For example,
963 <pre>
964 (a(?i)b|c)
965 </pre>
966 matches "ab", "aB", "c", and "C", even though when matching "C" the first
967 branch is abandoned before the option setting. This is because the effects of
968 option settings happen at compile time. There would be some very weird
969 behaviour otherwise.
970 </P>
971 <P>
972 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
973 changed in the same way as the Perl-compatible options by using the characters
974 J, U and X respectively.
975 <a name="subpattern"></a></P>
976 <br><a name="SEC11" href="#TOC1">SUBPATTERNS</a><br>
977 <P>
978 Subpatterns are delimited by parentheses (round brackets), which can be nested.
979 Turning part of a pattern into a subpattern does two things:
980 <br>
981 <br>
982 1. It localizes a set of alternatives. For example, the pattern
983 <pre>
984 cat(aract|erpillar|)
985 </pre>
986 matches one of the words "cat", "cataract", or "caterpillar". Without the
987 parentheses, it would match "cataract", "erpillar" or an empty string.
988 <br>
989 <br>
990 2. It sets up the subpattern as a capturing subpattern. This means that, when
991 the whole pattern matches, that portion of the subject string that matched the
992 subpattern is passed back to the caller via the <i>ovector</i> argument of
993 <b>pcre_exec()</b>. Opening parentheses are counted from left to right (starting
994 from 1) to obtain numbers for the capturing subpatterns.
995 </P>
996 <P>
997 For example, if the string "the red king" is matched against the pattern
998 <pre>
999 the ((red|white) (king|queen))
1000 </pre>
1001 the captured substrings are "red king", "red", and "king", and are numbered 1,
1002 2, and 3, respectively.
1003 </P>
1004 <P>
1005 The fact that plain parentheses fulfil two functions is not always helpful.
1006 There are often times when a grouping subpattern is required without a
1007 capturing requirement. If an opening parenthesis is followed by a question mark
1008 and a colon, the subpattern does not do any capturing, and is not counted when
1009 computing the number of any subsequent capturing subpatterns. For example, if
1010 the string "the white queen" is matched against the pattern
1011 <pre>
1012 the ((?:red|white) (king|queen))
1013 </pre>
1014 the captured substrings are "white queen" and "queen", and are numbered 1 and
1015 2. The maximum number of capturing subpatterns is 65535.
1016 </P>
1017 <P>
1018 As a convenient shorthand, if any option settings are required at the start of
1019 a non-capturing subpattern, the option letters may appear between the "?" and
1020 the ":". Thus the two patterns
1021 <pre>
1022 (?i:saturday|sunday)
1023 (?:(?i)saturday|sunday)
1024 </pre>
1025 match exactly the same set of strings. Because alternative branches are tried
1026 from left to right, and options are not reset until the end of the subpattern
1027 is reached, an option setting in one branch does affect subsequent branches, so
1028 the above patterns match "SUNDAY" as well as "Saturday".
1029 </P>
1030 <br><a name="SEC12" href="#TOC1">DUPLICATE SUBPATTERN NUMBERS</a><br>
1031 <P>
1032 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1033 the same numbers for its capturing parentheses. Such a subpattern starts with
1034 (?| and is itself a non-capturing subpattern. For example, consider this
1035 pattern:
1036 <pre>
1037 (?|(Sat)ur|(Sun))day
1038 </pre>
1039 Because the two alternatives are inside a (?| group, both sets of capturing
1040 parentheses are numbered one. Thus, when the pattern matches, you can look
1041 at captured substring number one, whichever alternative matched. This construct
1042 is useful when you want to capture part, but not all, of one of a number of
1043 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1044 number is reset at the start of each branch. The numbers of any capturing
1045 buffers that follow the subpattern start after the highest number used in any
1046 branch. The following example is taken from the Perl documentation.
1047 The numbers underneath show in which buffer the captured content will be
1048 stored.
1049 <pre>
1050 # before ---------------branch-reset----------- after
1051 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1052 # 1 2 2 3 2 3 4
1053 </pre>
1054 A backreference or a recursive call to a numbered subpattern always refers to
1055 the first one in the pattern with the given number.
1056 </P>
1057 <P>
1058 An alternative approach to using this "branch reset" feature is to use
1059 duplicate named subpatterns, as described in the next section.
1060 </P>
1061 <br><a name="SEC13" href="#TOC1">NAMED SUBPATTERNS</a><br>
1062 <P>
1063 Identifying capturing parentheses by number is simple, but it can be very hard
1064 to keep track of the numbers in complicated regular expressions. Furthermore,
1065 if an expression is modified, the numbers may change. To help with this
1066 difficulty, PCRE supports the naming of subpatterns. This feature was not
1067 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1068 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1069 the Perl and the Python syntax.
1070 </P>
1071 <P>
1072 In PCRE, a subpattern can be named in one of three ways: (?&#60;name&#62;...) or
1073 (?'name'...) as in Perl, or (?P&#60;name&#62;...) as in Python. References to capturing
1074 parentheses from other parts of the pattern, such as
1075 <a href="#backreferences">backreferences,</a>
1076 <a href="#recursion">recursion,</a>
1077 and
1078 <a href="#conditions">conditions,</a>
1079 can be made by name as well as by number.
1080 </P>
1081 <P>
1082 Names consist of up to 32 alphanumeric characters and underscores. Named
1083 capturing parentheses are still allocated numbers as well as names, exactly as
1084 if the names were not present. The PCRE API provides function calls for
1085 extracting the name-to-number translation table from a compiled pattern. There
1086 is also a convenience function for extracting a captured substring by name.
1087 </P>
1088 <P>
1089 By default, a name must be unique within a pattern, but it is possible to relax
1090 this constraint by setting the PCRE_DUPNAMES option at compile time. This can
1091 be useful for patterns where only one instance of the named parentheses can
1092 match. Suppose you want to match the name of a weekday, either as a 3-letter
1093 abbreviation or as the full name, and in both cases you want to extract the
1094 abbreviation. This pattern (ignoring the line breaks) does the job:
1095 <pre>
1096 (?&#60;DN&#62;Mon|Fri|Sun)(?:day)?|
1097 (?&#60;DN&#62;Tue)(?:sday)?|
1098 (?&#60;DN&#62;Wed)(?:nesday)?|
1099 (?&#60;DN&#62;Thu)(?:rsday)?|
1100 (?&#60;DN&#62;Sat)(?:urday)?
1101 </pre>
1102 There are five capturing substrings, but only one is ever set after a match.
1103 (An alternative way of solving this problem is to use a "branch reset"
1104 subpattern, as described in the previous section.)
1105 </P>
1106 <P>
1107 The convenience function for extracting the data by name returns the substring
1108 for the first (and in this example, the only) subpattern of that name that
1109 matched. This saves searching to find which numbered subpattern it was. If you
1110 make a reference to a non-unique named subpattern from elsewhere in the
1111 pattern, the one that corresponds to the lowest number is used. For further
1112 details of the interfaces for handling named subpatterns, see the
1113 <a href="pcreapi.html"><b>pcreapi</b></a>
1114 documentation.
1115 </P>
1116 <br><a name="SEC14" href="#TOC1">REPETITION</a><br>
1117 <P>
1118 Repetition is specified by quantifiers, which can follow any of the following
1119 items:
1120 <pre>
1121 a literal data character
1122 the dot metacharacter
1123 the \C escape sequence
1124 the \X escape sequence (in UTF-8 mode with Unicode properties)
1125 the \R escape sequence
1126 an escape such as \d that matches a single character
1127 a character class
1128 a back reference (see next section)
1129 a parenthesized subpattern (unless it is an assertion)
1130 </pre>
1131 The general repetition quantifier specifies a minimum and maximum number of
1132 permitted matches, by giving the two numbers in curly brackets (braces),
1133 separated by a comma. The numbers must be less than 65536, and the first must
1134 be less than or equal to the second. For example:
1135 <pre>
1136 z{2,4}
1137 </pre>
1138 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1139 character. If the second number is omitted, but the comma is present, there is
1140 no upper limit; if the second number and the comma are both omitted, the
1141 quantifier specifies an exact number of required matches. Thus
1142 <pre>
1143 [aeiou]{3,}
1144 </pre>
1145 matches at least 3 successive vowels, but may match many more, while
1146 <pre>
1147 \d{8}
1148 </pre>
1149 matches exactly 8 digits. An opening curly bracket that appears in a position
1150 where a quantifier is not allowed, or one that does not match the syntax of a
1151 quantifier, is taken as a literal character. For example, {,6} is not a
1152 quantifier, but a literal string of four characters.
1153 </P>
1154 <P>
1155 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1156 bytes. Thus, for example, \x{100}{2} matches two UTF-8 characters, each of
1157 which is represented by a two-byte sequence. Similarly, when Unicode property
1158 support is available, \X{3} matches three Unicode extended sequences, each of
1159 which may be several bytes long (and they may be of different lengths).
1160 </P>
1161 <P>
1162 The quantifier {0} is permitted, causing the expression to behave as if the
1163 previous item and the quantifier were not present.
1164 </P>
1165 <P>
1166 For convenience, the three most common quantifiers have single-character
1167 abbreviations:
1168 <pre>
1169 * is equivalent to {0,}
1170 + is equivalent to {1,}
1171 ? is equivalent to {0,1}
1172 </pre>
1173 It is possible to construct infinite loops by following a subpattern that can
1174 match no characters with a quantifier that has no upper limit, for example:
1175 <pre>
1176 (a?)*
1177 </pre>
1178 Earlier versions of Perl and PCRE used to give an error at compile time for
1179 such patterns. However, because there are cases where this can be useful, such
1180 patterns are now accepted, but if any repetition of the subpattern does in fact
1181 match no characters, the loop is forcibly broken.
1182 </P>
1183 <P>
1184 By default, the quantifiers are "greedy", that is, they match as much as
1185 possible (up to the maximum number of permitted times), without causing the
1186 rest of the pattern to fail. The classic example of where this gives problems
1187 is in trying to match comments in C programs. These appear between /* and */
1188 and within the comment, individual * and / characters may appear. An attempt to
1189 match C comments by applying the pattern
1190 <pre>
1191 /\*.*\*/
1192 </pre>
1193 to the string
1194 <pre>
1195 /* first comment */ not comment /* second comment */
1196 </pre>
1197 fails, because it matches the entire string owing to the greediness of the .*
1198 item.
1199 </P>
1200 <P>
1201 However, if a quantifier is followed by a question mark, it ceases to be
1202 greedy, and instead matches the minimum number of times possible, so the
1203 pattern
1204 <pre>
1205 /\*.*?\*/
1206 </pre>
1207 does the right thing with the C comments. The meaning of the various
1208 quantifiers is not otherwise changed, just the preferred number of matches.
1209 Do not confuse this use of question mark with its use as a quantifier in its
1210 own right. Because it has two uses, it can sometimes appear doubled, as in
1211 <pre>
1212 \d??\d
1213 </pre>
1214 which matches one digit by preference, but can match two if that is the only
1215 way the rest of the pattern matches.
1216 </P>
1217 <P>
1218 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1219 the quantifiers are not greedy by default, but individual ones can be made
1220 greedy by following them with a question mark. In other words, it inverts the
1221 default behaviour.
1222 </P>
1223 <P>
1224 When a parenthesized subpattern is quantified with a minimum repeat count that
1225 is greater than 1 or with a limited maximum, more memory is required for the
1226 compiled pattern, in proportion to the size of the minimum or maximum.
1227 </P>
1228 <P>
1229 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1230 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1231 implicitly anchored, because whatever follows will be tried against every
1232 character position in the subject string, so there is no point in retrying the
1233 overall match at any position after the first. PCRE normally treats such a
1234 pattern as though it were preceded by \A.
1235 </P>
1236 <P>
1237 In cases where it is known that the subject string contains no newlines, it is
1238 worth setting PCRE_DOTALL in order to obtain this optimization, or
1239 alternatively using ^ to indicate anchoring explicitly.
1240 </P>
1241 <P>
1242 However, there is one situation where the optimization cannot be used. When .*
1243 is inside capturing parentheses that are the subject of a backreference
1244 elsewhere in the pattern, a match at the start may fail where a later one
1245 succeeds. Consider, for example:
1246 <pre>
1247 (.*)abc\1
1248 </pre>
1249 If the subject is "xyz123abc123" the match point is the fourth character. For
1250 this reason, such a pattern is not implicitly anchored.
1251 </P>
1252 <P>
1253 When a capturing subpattern is repeated, the value captured is the substring
1254 that matched the final iteration. For example, after
1255 <pre>
1256 (tweedle[dume]{3}\s*)+
1257 </pre>
1258 has matched "tweedledum tweedledee" the value of the captured substring is
1259 "tweedledee". However, if there are nested capturing subpatterns, the
1260 corresponding captured values may have been set in previous iterations. For
1261 example, after
1262 <pre>
1263 /(a|(b))+/
1264 </pre>
1265 matches "aba" the value of the second captured substring is "b".
1266 <a name="atomicgroup"></a></P>
1267 <br><a name="SEC15" href="#TOC1">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a><br>
1268 <P>
1269 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1270 repetition, failure of what follows normally causes the repeated item to be
1271 re-evaluated to see if a different number of repeats allows the rest of the
1272 pattern to match. Sometimes it is useful to prevent this, either to change the
1273 nature of the match, or to cause it fail earlier than it otherwise might, when
1274 the author of the pattern knows there is no point in carrying on.
1275 </P>
1276 <P>
1277 Consider, for example, the pattern \d+foo when applied to the subject line
1278 <pre>
1279 123456bar
1280 </pre>
1281 After matching all 6 digits and then failing to match "foo", the normal
1282 action of the matcher is to try again with only 5 digits matching the \d+
1283 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1284 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1285 that once a subpattern has matched, it is not to be re-evaluated in this way.
1286 </P>
1287 <P>
1288 If we use atomic grouping for the previous example, the matcher gives up
1289 immediately on failing to match "foo" the first time. The notation is a kind of
1290 special parenthesis, starting with (?&#62; as in this example:
1291 <pre>
1292 (?&#62;\d+)foo
1293 </pre>
1294 This kind of parenthesis "locks up" the part of the pattern it contains once
1295 it has matched, and a failure further into the pattern is prevented from
1296 backtracking into it. Backtracking past it to previous items, however, works as
1297 normal.
1298 </P>
1299 <P>
1300 An alternative description is that a subpattern of this type matches the string
1301 of characters that an identical standalone pattern would match, if anchored at
1302 the current point in the subject string.
1303 </P>
1304 <P>
1305 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1306 the above example can be thought of as a maximizing repeat that must swallow
1307 everything it can. So, while both \d+ and \d+? are prepared to adjust the
1308 number of digits they match in order to make the rest of the pattern match,
1309 (?&#62;\d+) can only match an entire sequence of digits.
1310 </P>
1311 <P>
1312 Atomic groups in general can of course contain arbitrarily complicated
1313 subpatterns, and can be nested. However, when the subpattern for an atomic
1314 group is just a single repeated item, as in the example above, a simpler
1315 notation, called a "possessive quantifier" can be used. This consists of an
1316 additional + character following a quantifier. Using this notation, the
1317 previous example can be rewritten as
1318 <pre>
1319 \d++foo
1320 </pre>
1321 Note that a possessive quantifier can be used with an entire group, for
1322 example:
1323 <pre>
1324 (abc|xyz){2,3}+
1325 </pre>
1326 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1327 option is ignored. They are a convenient notation for the simpler forms of
1328 atomic group. However, there is no difference in the meaning of a possessive
1329 quantifier and the equivalent atomic group, though there may be a performance
1330 difference; possessive quantifiers should be slightly faster.
1331 </P>
1332 <P>
1333 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1334 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1335 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1336 package, and PCRE copied it from there. It ultimately found its way into Perl
1337 at release 5.10.
1338 </P>
1339 <P>
1340 PCRE has an optimization that automatically "possessifies" certain simple
1341 pattern constructs. For example, the sequence A+B is treated as A++B because
1342 there is no point in backtracking into a sequence of A's when B must follow.
1343 </P>
1344 <P>
1345 When a pattern contains an unlimited repeat inside a subpattern that can itself
1346 be repeated an unlimited number of times, the use of an atomic group is the
1347 only way to avoid some failing matches taking a very long time indeed. The
1348 pattern
1349 <pre>
1350 (\D+|&#60;\d+&#62;)*[!?]
1351 </pre>
1352 matches an unlimited number of substrings that either consist of non-digits, or
1353 digits enclosed in &#60;&#62;, followed by either ! or ?. When it matches, it runs
1354 quickly. However, if it is applied to
1355 <pre>
1356 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1357 </pre>
1358 it takes a long time before reporting failure. This is because the string can
1359 be divided between the internal \D+ repeat and the external * repeat in a
1360 large number of ways, and all have to be tried. (The example uses [!?] rather
1361 than a single character at the end, because both PCRE and Perl have an
1362 optimization that allows for fast failure when a single character is used. They
1363 remember the last single character that is required for a match, and fail early
1364 if it is not present in the string.) If the pattern is changed so that it uses
1365 an atomic group, like this:
1366 <pre>
1367 ((?&#62;\D+)|&#60;\d+&#62;)*[!?]
1368 </pre>
1369 sequences of non-digits cannot be broken, and failure happens quickly.
1370 <a name="backreferences"></a></P>
1371 <br><a name="SEC16" href="#TOC1">BACK REFERENCES</a><br>
1372 <P>
1373 Outside a character class, a backslash followed by a digit greater than 0 (and
1374 possibly further digits) is a back reference to a capturing subpattern earlier
1375 (that is, to its left) in the pattern, provided there have been that many
1376 previous capturing left parentheses.
1377 </P>
1378 <P>
1379 However, if the decimal number following the backslash is less than 10, it is
1380 always taken as a back reference, and causes an error only if there are not
1381 that many capturing left parentheses in the entire pattern. In other words, the
1382 parentheses that are referenced need not be to the left of the reference for
1383 numbers less than 10. A "forward back reference" of this type can make sense
1384 when a repetition is involved and the subpattern to the right has participated
1385 in an earlier iteration.
1386 </P>
1387 <P>
1388 It is not possible to have a numerical "forward back reference" to a subpattern
1389 whose number is 10 or more using this syntax because a sequence such as \50 is
1390 interpreted as a character defined in octal. See the subsection entitled
1391 "Non-printing characters"
1392 <a href="#digitsafterbackslash">above</a>
1393 for further details of the handling of digits following a backslash. There is
1394 no such problem when named parentheses are used. A back reference to any
1395 subpattern is possible using named parentheses (see below).
1396 </P>
1397 <P>
1398 Another way of avoiding the ambiguity inherent in the use of digits following a
1399 backslash is to use the \g escape sequence, which is a feature introduced in
1400 Perl 5.10. This escape must be followed by an unsigned number or a negative
1401 number, optionally enclosed in braces. These examples are all identical:
1402 <pre>
1403 (ring), \1
1404 (ring), \g1
1405 (ring), \g{1}
1406 </pre>
1407 An unsigned number specifies an absolute reference without the ambiguity that
1408 is present in the older syntax. It is also useful when literal digits follow
1409 the reference. A negative number is a relative reference. Consider this
1410 example:
1411 <pre>
1412 (abc(def)ghi)\g{-1}
1413 </pre>
1414 The sequence \g{-1} is a reference to the most recently started capturing
1415 subpattern before \g, that is, is it equivalent to \2. Similarly, \g{-2}
1416 would be equivalent to \1. The use of relative references can be helpful in
1417 long patterns, and also in patterns that are created by joining together
1418 fragments that contain references within themselves.
1419 </P>
1420 <P>
1421 A back reference matches whatever actually matched the capturing subpattern in
1422 the current subject string, rather than anything matching the subpattern
1423 itself (see
1424 <a href="#subpatternsassubroutines">"Subpatterns as subroutines"</a>
1425 below for a way of doing that). So the pattern
1426 <pre>
1427 (sens|respons)e and \1ibility
1428 </pre>
1429 matches "sense and sensibility" and "response and responsibility", but not
1430 "sense and responsibility". If caseful matching is in force at the time of the
1431 back reference, the case of letters is relevant. For example,
1432 <pre>
1433 ((?i)rah)\s+\1
1434 </pre>
1435 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1436 capturing subpattern is matched caselessly.
1437 </P>
1438 <P>
1439 There are several different ways of writing back references to named
1440 subpatterns. The .NET syntax \k{name} and the Perl syntax \k&#60;name&#62; or
1441 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1442 back reference syntax, in which \g can be used for both numeric and named
1443 references, is also supported. We could rewrite the above example in any of
1444 the following ways:
1445 <pre>
1446 (?&#60;p1&#62;(?i)rah)\s+\k&#60;p1&#62;
1447 (?'p1'(?i)rah)\s+\k{p1}
1448 (?P&#60;p1&#62;(?i)rah)\s+(?P=p1)
1449 (?&#60;p1&#62;(?i)rah)\s+\g{p1}
1450 </pre>
1451 A subpattern that is referenced by name may appear in the pattern before or
1452 after the reference.
1453 </P>
1454 <P>
1455 There may be more than one back reference to the same subpattern. If a
1456 subpattern has not actually been used in a particular match, any back
1457 references to it always fail. For example, the pattern
1458 <pre>
1459 (a|(bc))\2
1460 </pre>
1461 always fails if it starts to match "a" rather than "bc". Because there may be
1462 many capturing parentheses in a pattern, all digits following the backslash are
1463 taken as part of a potential back reference number. If the pattern continues
1464 with a digit character, some delimiter must be used to terminate the back
1465 reference. If the PCRE_EXTENDED option is set, this can be whitespace.
1466 Otherwise an empty comment (see
1467 <a href="#comments">"Comments"</a>
1468 below) can be used.
1469 </P>
1470 <P>
1471 A back reference that occurs inside the parentheses to which it refers fails
1472 when the subpattern is first used, so, for example, (a\1) never matches.
1473 However, such references can be useful inside repeated subpatterns. For
1474 example, the pattern
1475 <pre>
1476 (a|b\1)+
1477 </pre>
1478 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1479 the subpattern, the back reference matches the character string corresponding
1480 to the previous iteration. In order for this to work, the pattern must be such
1481 that the first iteration does not need to match the back reference. This can be
1482 done using alternation, as in the example above, or by a quantifier with a
1483 minimum of zero.
1484 <a name="bigassertions"></a></P>
1485 <br><a name="SEC17" href="#TOC1">ASSERTIONS</a><br>
1486 <P>
1487 An assertion is a test on the characters following or preceding the current
1488 matching point that does not actually consume any characters. The simple
1489 assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are described
1490 <a href="#smallassertions">above.</a>
1491 </P>
1492 <P>
1493 More complicated assertions are coded as subpatterns. There are two kinds:
1494 those that look ahead of the current position in the subject string, and those
1495 that look behind it. An assertion subpattern is matched in the normal way,
1496 except that it does not cause the current matching position to be changed.
1497 </P>
1498 <P>
1499 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1500 because it makes no sense to assert the same thing several times. If any kind
1501 of assertion contains capturing subpatterns within it, these are counted for
1502 the purposes of numbering the capturing subpatterns in the whole pattern.
1503 However, substring capturing is carried out only for positive assertions,
1504 because it does not make sense for negative assertions.
1505 </P>
1506 <br><b>
1507 Lookahead assertions
1508 </b><br>
1509 <P>
1510 Lookahead assertions start with (?= for positive assertions and (?! for
1511 negative assertions. For example,
1512 <pre>
1513 \w+(?=;)
1514 </pre>
1515 matches a word followed by a semicolon, but does not include the semicolon in
1516 the match, and
1517 <pre>
1518 foo(?!bar)
1519 </pre>
1520 matches any occurrence of "foo" that is not followed by "bar". Note that the
1521 apparently similar pattern
1522 <pre>
1523 (?!foo)bar
1524 </pre>
1525 does not find an occurrence of "bar" that is preceded by something other than
1526 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1527 (?!foo) is always true when the next three characters are "bar". A
1528 lookbehind assertion is needed to achieve the other effect.
1529 </P>
1530 <P>
1531 If you want to force a matching failure at some point in a pattern, the most
1532 convenient way to do it is with (?!) because an empty string always matches, so
1533 an assertion that requires there not to be an empty string must always fail.
1534 <a name="lookbehind"></a></P>
1535 <br><b>
1536 Lookbehind assertions
1537 </b><br>
1538 <P>
1539 Lookbehind assertions start with (?&#60;= for positive assertions and (?&#60;! for
1540 negative assertions. For example,
1541 <pre>
1542 (?&#60;!foo)bar
1543 </pre>
1544 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1545 a lookbehind assertion are restricted such that all the strings it matches must
1546 have a fixed length. However, if there are several top-level alternatives, they
1547 do not all have to have the same fixed length. Thus
1548 <pre>
1549 (?&#60;=bullock|donkey)
1550 </pre>
1551 is permitted, but
1552 <pre>
1553 (?&#60;!dogs?|cats?)
1554 </pre>
1555 causes an error at compile time. Branches that match different length strings
1556 are permitted only at the top level of a lookbehind assertion. This is an
1557 extension compared with Perl (at least for 5.8), which requires all branches to
1558 match the same length of string. An assertion such as
1559 <pre>
1560 (?&#60;=ab(c|de))
1561 </pre>
1562 is not permitted, because its single top-level branch can match two different
1563 lengths, but it is acceptable if rewritten to use two top-level branches:
1564 <pre>
1565 (?&#60;=abc|abde)
1566 </pre>
1567 In some cases, the Perl 5.10 escape sequence \K
1568 <a href="#resetmatchstart">(see above)</a>
1569 can be used instead of a lookbehind assertion; this is not restricted to a
1570 fixed-length.
1571 </P>
1572 <P>
1573 The implementation of lookbehind assertions is, for each alternative, to
1574 temporarily move the current position back by the fixed length and then try to
1575 match. If there are insufficient characters before the current position, the
1576 assertion fails.
1577 </P>
1578 <P>
1579 PCRE does not allow the \C escape (which matches a single byte in UTF-8 mode)
1580 to appear in lookbehind assertions, because it makes it impossible to calculate
1581 the length of the lookbehind. The \X and \R escapes, which can match
1582 different numbers of bytes, are also not permitted.
1583 </P>
1584 <P>
1585 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1586 specify efficient matching at the end of the subject string. Consider a simple
1587 pattern such as
1588 <pre>
1589 abcd$
1590 </pre>
1591 when applied to a long string that does not match. Because matching proceeds
1592 from left to right, PCRE will look for each "a" in the subject and then see if
1593 what follows matches the rest of the pattern. If the pattern is specified as
1594 <pre>
1595 ^.*abcd$
1596 </pre>
1597 the initial .* matches the entire string at first, but when this fails (because
1598 there is no following "a"), it backtracks to match all but the last character,
1599 then all but the last two characters, and so on. Once again the search for "a"
1600 covers the entire string, from right to left, so we are no better off. However,
1601 if the pattern is written as
1602 <pre>
1603 ^.*+(?&#60;=abcd)
1604 </pre>
1605 there can be no backtracking for the .*+ item; it can match only the entire
1606 string. The subsequent lookbehind assertion does a single test on the last four
1607 characters. If it fails, the match fails immediately. For long strings, this
1608 approach makes a significant difference to the processing time.
1609 </P>
1610 <br><b>
1611 Using multiple assertions
1612 </b><br>
1613 <P>
1614 Several assertions (of any sort) may occur in succession. For example,
1615 <pre>
1616 (?&#60;=\d{3})(?&#60;!999)foo
1617 </pre>
1618 matches "foo" preceded by three digits that are not "999". Notice that each of
1619 the assertions is applied independently at the same point in the subject
1620 string. First there is a check that the previous three characters are all
1621 digits, and then there is a check that the same three characters are not "999".
1622 This pattern does <i>not</i> match "foo" preceded by six characters, the first
1623 of which are digits and the last three of which are not "999". For example, it
1624 doesn't match "123abcfoo". A pattern to do that is
1625 <pre>
1626 (?&#60;=\d{3}...)(?&#60;!999)foo
1627 </pre>
1628 This time the first assertion looks at the preceding six characters, checking
1629 that the first three are digits, and then the second assertion checks that the
1630 preceding three characters are not "999".
1631 </P>
1632 <P>
1633 Assertions can be nested in any combination. For example,
1634 <pre>
1635 (?&#60;=(?&#60;!foo)bar)baz
1636 </pre>
1637 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1638 preceded by "foo", while
1639 <pre>
1640 (?&#60;=\d{3}(?!999)...)foo
1641 </pre>
1642 is another pattern that matches "foo" preceded by three digits and any three
1643 characters that are not "999".
1644 <a name="conditions"></a></P>
1645 <br><a name="SEC18" href="#TOC1">CONDITIONAL SUBPATTERNS</a><br>
1646 <P>
1647 It is possible to cause the matching process to obey a subpattern
1648 conditionally or to choose between two alternative subpatterns, depending on
1649 the result of an assertion, or whether a previous capturing subpattern matched
1650 or not. The two possible forms of conditional subpattern are
1651 <pre>
1652 (?(condition)yes-pattern)
1653 (?(condition)yes-pattern|no-pattern)
1654 </pre>
1655 If the condition is satisfied, the yes-pattern is used; otherwise the
1656 no-pattern (if present) is used. If there are more than two alternatives in the
1657 subpattern, a compile-time error occurs.
1658 </P>
1659 <P>
1660 There are four kinds of condition: references to subpatterns, references to
1661 recursion, a pseudo-condition called DEFINE, and assertions.
1662 </P>
1663 <br><b>
1664 Checking for a used subpattern by number
1665 </b><br>
1666 <P>
1667 If the text between the parentheses consists of a sequence of digits, the
1668 condition is true if the capturing subpattern of that number has previously
1669 matched. An alternative notation is to precede the digits with a plus or minus
1670 sign. In this case, the subpattern number is relative rather than absolute.
1671 The most recently opened parentheses can be referenced by (?(-1), the next most
1672 recent by (?(-2), and so on. In looping constructs it can also make sense to
1673 refer to subsequent groups with constructs such as (?(+2).
1674 </P>
1675 <P>
1676 Consider the following pattern, which contains non-significant white space to
1677 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
1678 three parts for ease of discussion:
1679 <pre>
1680 ( \( )? [^()]+ (?(1) \) )
1681 </pre>
1682 The first part matches an optional opening parenthesis, and if that
1683 character is present, sets it as the first captured substring. The second part
1684 matches one or more characters that are not parentheses. The third part is a
1685 conditional subpattern that tests whether the first set of parentheses matched
1686 or not. If they did, that is, if subject started with an opening parenthesis,
1687 the condition is true, and so the yes-pattern is executed and a closing
1688 parenthesis is required. Otherwise, since no-pattern is not present, the
1689 subpattern matches nothing. In other words, this pattern matches a sequence of
1690 non-parentheses, optionally enclosed in parentheses.
1691 </P>
1692 <P>
1693 If you were embedding this pattern in a larger one, you could use a relative
1694 reference:
1695 <pre>
1696 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
1697 </pre>
1698 This makes the fragment independent of the parentheses in the larger pattern.
1699 </P>
1700 <br><b>
1701 Checking for a used subpattern by name
1702 </b><br>
1703 <P>
1704 Perl uses the syntax (?(&#60;name&#62;)...) or (?('name')...) to test for a used
1705 subpattern by name. For compatibility with earlier versions of PCRE, which had
1706 this facility before Perl, the syntax (?(name)...) is also recognized. However,
1707 there is a possible ambiguity with this syntax, because subpattern names may
1708 consist entirely of digits. PCRE looks first for a named subpattern; if it
1709 cannot find one and the name consists entirely of digits, PCRE looks for a
1710 subpattern of that number, which must be greater than zero. Using subpattern
1711 names that consist entirely of digits is not recommended.
1712 </P>
1713 <P>
1714 Rewriting the above example to use a named subpattern gives this:
1715 <pre>
1716 (?&#60;OPEN&#62; \( )? [^()]+ (?(&#60;OPEN&#62;) \) )
1718 </PRE>
1719 </P>
1720 <br><b>
1721 Checking for pattern recursion
1722 </b><br>
1723 <P>
1724 If the condition is the string (R), and there is no subpattern with the name R,
1725 the condition is true if a recursive call to the whole pattern or any
1726 subpattern has been made. If digits or a name preceded by ampersand follow the
1727 letter R, for example:
1728 <pre>
1729 (?(R3)...) or (?(R&name)...)
1730 </pre>
1731 the condition is true if the most recent recursion is into the subpattern whose
1732 number or name is given. This condition does not check the entire recursion
1733 stack.
1734 </P>
1735 <P>
1736 At "top level", all these recursion test conditions are false. Recursive
1737 patterns are described below.
1738 </P>
1739 <br><b>
1740 Defining subpatterns for use by reference only
1741 </b><br>
1742 <P>
1743 If the condition is the string (DEFINE), and there is no subpattern with the
1744 name DEFINE, the condition is always false. In this case, there may be only one
1745 alternative in the subpattern. It is always skipped if control reaches this
1746 point in the pattern; the idea of DEFINE is that it can be used to define
1747 "subroutines" that can be referenced from elsewhere. (The use of "subroutines"
1748 is described below.) For example, a pattern to match an IPv4 address could be
1749 written like this (ignore whitespace and line breaks):
1750 <pre>
1751 (?(DEFINE) (?&#60;byte&#62; 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
1752 \b (?&byte) (\.(?&byte)){3} \b
1753 </pre>
1754 The first part of the pattern is a DEFINE group inside which a another group
1755 named "byte" is defined. This matches an individual component of an IPv4
1756 address (a number less than 256). When matching takes place, this part of the
1757 pattern is skipped because DEFINE acts like a false condition.
1758 </P>
1759 <P>
1760 The rest of the pattern uses references to the named group to match the four
1761 dot-separated components of an IPv4 address, insisting on a word boundary at
1762 each end.
1763 </P>
1764 <br><b>
1765 Assertion conditions
1766 </b><br>
1767 <P>
1768 If the condition is not in any of the above formats, it must be an assertion.
1769 This may be a positive or negative lookahead or lookbehind assertion. Consider
1770 this pattern, again containing non-significant white space, and with the two
1771 alternatives on the second line:
1772 <pre>
1773 (?(?=[^a-z]*[a-z])
1774 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
1775 </pre>
1776 The condition is a positive lookahead assertion that matches an optional
1777 sequence of non-letters followed by a letter. In other words, it tests for the
1778 presence of at least one letter in the subject. If a letter is found, the
1779 subject is matched against the first alternative; otherwise it is matched
1780 against the second. This pattern matches strings in one of the two forms
1781 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
1782 <a name="comments"></a></P>
1783 <br><a name="SEC19" href="#TOC1">COMMENTS</a><br>
1784 <P>
1785 The sequence (?# marks the start of a comment that continues up to the next
1786 closing parenthesis. Nested parentheses are not permitted. The characters
1787 that make up a comment play no part in the pattern matching at all.
1788 </P>
1789 <P>
1790 If the PCRE_EXTENDED option is set, an unescaped # character outside a
1791 character class introduces a comment that continues to immediately after the
1792 next newline in the pattern.
1793 <a name="recursion"></a></P>
1794 <br><a name="SEC20" href="#TOC1">RECURSIVE PATTERNS</a><br>
1795 <P>
1796 Consider the problem of matching a string in parentheses, allowing for
1797 unlimited nested parentheses. Without the use of recursion, the best that can
1798 be done is to use a pattern that matches up to some fixed depth of nesting. It
1799 is not possible to handle an arbitrary nesting depth.
1800 </P>
1801 <P>
1802 For some time, Perl has provided a facility that allows regular expressions to
1803 recurse (amongst other things). It does this by interpolating Perl code in the
1804 expression at run time, and the code can refer to the expression itself. A Perl
1805 pattern using code interpolation to solve the parentheses problem can be
1806 created like this:
1807 <pre>
1808 $re = qr{\( (?: (?&#62;[^()]+) | (?p{$re}) )* \)}x;
1809 </pre>
1810 The (?p{...}) item interpolates Perl code at run time, and in this case refers
1811 recursively to the pattern in which it appears.
1812 </P>
1813 <P>
1814 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
1815 supports special syntax for recursion of the entire pattern, and also for
1816 individual subpattern recursion. After its introduction in PCRE and Python,
1817 this kind of recursion was introduced into Perl at release 5.10.
1818 </P>
1819 <P>
1820 A special item that consists of (? followed by a number greater than zero and a
1821 closing parenthesis is a recursive call of the subpattern of the given number,
1822 provided that it occurs inside that subpattern. (If not, it is a "subroutine"
1823 call, which is described in the next section.) The special item (?R) or (?0) is
1824 a recursive call of the entire regular expression.
1825 </P>
1826 <P>
1827 In PCRE (like Python, but unlike Perl), a recursive subpattern call is always
1828 treated as an atomic group. That is, once it has matched some of the subject
1829 string, it is never re-entered, even if it contains untried alternatives and
1830 there is a subsequent matching failure.
1831 </P>
1832 <P>
1833 This PCRE pattern solves the nested parentheses problem (assume the
1834 PCRE_EXTENDED option is set so that white space is ignored):
1835 <pre>
1836 \( ( (?&#62;[^()]+) | (?R) )* \)
1837 </pre>
1838 First it matches an opening parenthesis. Then it matches any number of
1839 substrings which can either be a sequence of non-parentheses, or a recursive
1840 match of the pattern itself (that is, a correctly parenthesized substring).
1841 Finally there is a closing parenthesis.
1842 </P>
1843 <P>
1844 If this were part of a larger pattern, you would not want to recurse the entire
1845 pattern, so instead you could use this:
1846 <pre>
1847 ( \( ( (?&#62;[^()]+) | (?1) )* \) )
1848 </pre>
1849 We have put the pattern into parentheses, and caused the recursion to refer to
1850 them instead of the whole pattern.
1851 </P>
1852 <P>
1853 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
1854 is made easier by the use of relative references. (A Perl 5.10 feature.)
1855 Instead of (?1) in the pattern above you can write (?-2) to refer to the second
1856 most recently opened parentheses preceding the recursion. In other words, a
1857 negative number counts capturing parentheses leftwards from the point at which
1858 it is encountered.
1859 </P>
1860 <P>
1861 It is also possible to refer to subsequently opened parentheses, by writing
1862 references such as (?+2). However, these cannot be recursive because the
1863 reference is not inside the parentheses that are referenced. They are always
1864 "subroutine" calls, as described in the next section.
1865 </P>
1866 <P>
1867 An alternative approach is to use named parentheses instead. The Perl syntax
1868 for this is (?&name); PCRE's earlier syntax (?P&#62;name) is also supported. We
1869 could rewrite the above example as follows:
1870 <pre>
1871 (?&#60;pn&#62; \( ( (?&#62;[^()]+) | (?&pn) )* \) )
1872 </pre>
1873 If there is more than one subpattern with the same name, the earliest one is
1874 used.
1875 </P>
1876 <P>
1877 This particular example pattern that we have been looking at contains nested
1878 unlimited repeats, and so the use of atomic grouping for matching strings of
1879 non-parentheses is important when applying the pattern to strings that do not
1880 match. For example, when this pattern is applied to
1881 <pre>
1882 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
1883 </pre>
1884 it yields "no match" quickly. However, if atomic grouping is not used,
1885 the match runs for a very long time indeed because there are so many different
1886 ways the + and * repeats can carve up the subject, and all have to be tested
1887 before failure can be reported.
1888 </P>
1889 <P>
1890 At the end of a match, the values set for any capturing subpatterns are those
1891 from the outermost level of the recursion at which the subpattern value is set.
1892 If you want to obtain intermediate values, a callout function can be used (see
1893 below and the
1894 <a href="pcrecallout.html"><b>pcrecallout</b></a>
1895 documentation). If the pattern above is matched against
1896 <pre>
1897 (ab(cd)ef)
1898 </pre>
1899 the value for the capturing parentheses is "ef", which is the last value taken
1900 on at the top level. If additional parentheses are added, giving
1901 <pre>
1902 \( ( ( (?&#62;[^()]+) | (?R) )* ) \)
1903 ^ ^
1904 ^ ^
1905 </pre>
1906 the string they capture is "ab(cd)ef", the contents of the top level
1907 parentheses. If there are more than 15 capturing parentheses in a pattern, PCRE
1908 has to obtain extra memory to store data during a recursion, which it does by
1909 using <b>pcre_malloc</b>, freeing it via <b>pcre_free</b> afterwards. If no
1910 memory can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
1911 </P>
1912 <P>
1913 Do not confuse the (?R) item with the condition (R), which tests for recursion.
1914 Consider this pattern, which matches text in angle brackets, allowing for
1915 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
1916 recursing), whereas any characters are permitted at the outer level.
1917 <pre>
1918 &#60; (?: (?(R) \d++ | [^&#60;&#62;]*+) | (?R)) * &#62;
1919 </pre>
1920 In this pattern, (?(R) is the start of a conditional subpattern, with two
1921 different alternatives for the recursive and non-recursive cases. The (?R) item
1922 is the actual recursive call.
1923 <a name="subpatternsassubroutines"></a></P>
1924 <br><a name="SEC21" href="#TOC1">SUBPATTERNS AS SUBROUTINES</a><br>
1925 <P>
1926 If the syntax for a recursive subpattern reference (either by number or by
1927 name) is used outside the parentheses to which it refers, it operates like a
1928 subroutine in a programming language. The "called" subpattern may be defined
1929 before or after the reference. A numbered reference can be absolute or
1930 relative, as in these examples:
1931 <pre>
1932 (...(absolute)...)...(?2)...
1933 (...(relative)...)...(?-1)...
1934 (...(?+1)...(relative)...
1935 </pre>
1936 An earlier example pointed out that the pattern
1937 <pre>
1938 (sens|respons)e and \1ibility
1939 </pre>
1940 matches "sense and sensibility" and "response and responsibility", but not
1941 "sense and responsibility". If instead the pattern
1942 <pre>
1943 (sens|respons)e and (?1)ibility
1944 </pre>
1945 is used, it does match "sense and responsibility" as well as the other two
1946 strings. Another example is given in the discussion of DEFINE above.
1947 </P>
1948 <P>
1949 Like recursive subpatterns, a "subroutine" call is always treated as an atomic
1950 group. That is, once it has matched some of the subject string, it is never
1951 re-entered, even if it contains untried alternatives and there is a subsequent
1952 matching failure.
1953 </P>
1954 <P>
1955 When a subpattern is used as a subroutine, processing options such as
1956 case-independence are fixed when the subpattern is defined. They cannot be
1957 changed for different calls. For example, consider this pattern:
1958 <pre>
1959 (abc)(?i:(?-1))
1960 </pre>
1961 It matches "abcabc". It does not match "abcABC" because the change of
1962 processing option does not affect the called subpattern.
1963 </P>
1964 <br><a name="SEC22" href="#TOC1">CALLOUTS</a><br>
1965 <P>
1966 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
1967 code to be obeyed in the middle of matching a regular expression. This makes it
1968 possible, amongst other things, to extract different substrings that match the
1969 same pair of parentheses when there is a repetition.
1970 </P>
1971 <P>
1972 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
1973 code. The feature is called "callout". The caller of PCRE provides an external
1974 function by putting its entry point in the global variable <i>pcre_callout</i>.
1975 By default, this variable contains NULL, which disables all calling out.
1976 </P>
1977 <P>
1978 Within a regular expression, (?C) indicates the points at which the external
1979 function is to be called. If you want to identify different callout points, you
1980 can put a number less than 256 after the letter C. The default value is zero.
1981 For example, this pattern has two callout points:
1982 <pre>
1983 (?C1)abc(?C2)def
1984 </pre>
1985 If the PCRE_AUTO_CALLOUT flag is passed to <b>pcre_compile()</b>, callouts are
1986 automatically installed before each item in the pattern. They are all numbered
1987 255.
1988 </P>
1989 <P>
1990 During matching, when PCRE reaches a callout point (and <i>pcre_callout</i> is
1991 set), the external function is called. It is provided with the number of the
1992 callout, the position in the pattern, and, optionally, one item of data
1993 originally supplied by the caller of <b>pcre_exec()</b>. The callout function
1994 may cause matching to proceed, to backtrack, or to fail altogether. A complete
1995 description of the interface to the callout function is given in the
1996 <a href="pcrecallout.html"><b>pcrecallout</b></a>
1997 documentation.
1998 </P>
1999 <br><a name="SEC23" href="#TOC1">BACTRACKING CONTROL</a><br>
2000 <P>
2001 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2002 are described in the Perl documentation as "experimental and subject to change
2003 or removal in a future version of Perl". It goes on to say: "Their usage in
2004 production code should be noted to avoid problems during upgrades." The same
2005 remarks apply to the PCRE features described in this section.
2006 </P>
2007 <P>
2008 Since these verbs are specifically related to backtracking, they can be used
2009 only when the pattern is to be matched using <b>pcre_exec()</b>, which uses a
2010 backtracking algorithm. They cause an error if encountered by
2011 <b>pcre_dfa_exec()</b>.
2012 </P>
2013 <P>
2014 The new verbs make use of what was previously invalid syntax: an opening
2015 parenthesis followed by an asterisk. In Perl, they are generally of the form
2016 (*VERB:ARG) but PCRE does not support the use of arguments, so its general
2017 form is just (*VERB). Any number of these verbs may occur in a pattern. There
2018 are two kinds:
2019 </P>
2020 <br><b>
2021 Verbs that act immediately
2022 </b><br>
2023 <P>
2024 The following verbs act as soon as they are encountered:
2025 <pre>
2026 (*ACCEPT)
2027 </pre>
2028 This verb causes the match to end successfully, skipping the remainder of the
2029 pattern. When inside a recursion, only the innermost pattern is ended
2030 immediately. PCRE differs from Perl in what happens if the (*ACCEPT) is inside
2031 capturing parentheses. In Perl, the data so far is captured: in PCRE no data is
2032 captured. For example:
2033 <pre>
2034 A(A|B(*ACCEPT)|C)D
2035 </pre>
2036 This matches "AB", "AAD", or "ACD", but when it matches "AB", no data is
2037 captured.
2038 <pre>
2039 (*FAIL) or (*F)
2040 </pre>
2041 This verb causes the match to fail, forcing backtracking to occur. It is
2042 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2043 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2044 Perl features that are not present in PCRE. The nearest equivalent is the
2045 callout feature, as for example in this pattern:
2046 <pre>
2047 a+(?C)(*FAIL)
2048 </pre>
2049 A match with the string "aaaa" always fails, but the callout is taken before
2050 each backtrack happens (in this example, 10 times).
2051 </P>
2052 <br><b>
2053 Verbs that act after backtracking
2054 </b><br>
2055 <P>
2056 The following verbs do nothing when they are encountered. Matching continues
2057 with what follows, but if there is no subsequent match, a failure is forced.
2058 The verbs differ in exactly what kind of failure occurs.
2059 <pre>
2060 (*COMMIT)
2061 </pre>
2062 This verb causes the whole match to fail outright if the rest of the pattern
2063 does not match. Even if the pattern is unanchored, no further attempts to find
2064 a match by advancing the start point take place. Once (*COMMIT) has been
2065 passed, <b>pcre_exec()</b> is committed to finding a match at the current
2066 starting point, or not at all. For example:
2067 <pre>
2068 a+(*COMMIT)b
2069 </pre>
2070 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2071 dynamic anchor, or "I've started, so I must finish."
2072 <pre>
2073 (*PRUNE)
2074 </pre>
2075 This verb causes the match to fail at the current position if the rest of the
2076 pattern does not match. If the pattern is unanchored, the normal "bumpalong"
2077 advance to the next starting character then happens. Backtracking can occur as
2078 usual to the left of (*PRUNE), or when matching to the right of (*PRUNE), but
2079 if there is no match to the right, backtracking cannot cross (*PRUNE).
2080 In simple cases, the use of (*PRUNE) is just an alternative to an atomic
2081 group or possessive quantifier, but there are some uses of (*PRUNE) that cannot
2082 be expressed in any other way.
2083 <pre>
2084 (*SKIP)
2085 </pre>
2086 This verb is like (*PRUNE), except that if the pattern is unanchored, the
2087 "bumpalong" advance is not to the next character, but to the position in the
2088 subject where (*SKIP) was encountered. (*SKIP) signifies that whatever text
2089 was matched leading up to it cannot be part of a successful match. Consider:
2090 <pre>
2091 a+(*SKIP)b
2092 </pre>
2093 If the subject is "aaaac...", after the first match attempt fails (starting at
2094 the first character in the string), the starting point skips on to start the
2095 next attempt at "c". Note that a possessive quantifer does not have the same
2096 effect in this example; although it would suppress backtracking during the
2097 first match attempt, the second attempt would start at the second character
2098 instead of skipping on to "c".
2099 <pre>
2100 (*THEN)
2101 </pre>
2102 This verb causes a skip to the next alternation if the rest of the pattern does
2103 not match. That is, it cancels pending backtracking, but only within the
2104 current alternation. Its name comes from the observation that it can be used
2105 for a pattern-based if-then-else block:
2106 <pre>
2107 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2108 </pre>
2109 If the COND1 pattern matches, FOO is tried (and possibly further items after
2110 the end of the group if FOO succeeds); on failure the matcher skips to the
2111 second alternative and tries COND2, without backtracking into COND1. If (*THEN)
2112 is used outside of any alternation, it acts exactly like (*PRUNE).
2113 </P>
2114 <br><a name="SEC24" href="#TOC1">SEE ALSO</a><br>
2115 <P>
2116 <b>pcreapi</b>(3), <b>pcrecallout</b>(3), <b>pcrematching</b>(3), <b>pcre</b>(3).
2117 </P>
2118 <br><a name="SEC25" href="#TOC1">AUTHOR</a><br>
2119 <P>
2120 Philip Hazel
2121 <br>
2122 University Computing Service
2123 <br>
2124 Cambridge CB2 3QH, England.
2125 <br>
2126 </P>
2127 <br><a name="SEC26" href="#TOC1">REVISION</a><br>
2128 <P>
2129 Last updated: 09 August 2007
2130 <br>
2131 Copyright &copy; 1997-2007 University of Cambridge.
2132 <br>
2133 <p>
2134 Return to the <a href="index.html">PCRE index page</a>.
2135 </p>


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

  ViewVC Help
Powered by ViewVC 1.1.5