ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 227 - (show annotations)
Tue Aug 21 15:00:15 2007 UTC (13 years, 6 months ago) by ph10
File MIME type: text/html
File size: 91236 byte(s)
Error occurred while calculating annotation data.
Add (*CR) etc.
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">NEWLINE CONVENTIONS</a>
18 <li><a name="TOC3" href="#SEC3">CHARACTERS AND METACHARACTERS</a>
19 <li><a name="TOC4" href="#SEC4">BACKSLASH</a>
20 <li><a name="TOC5" href="#SEC5">CIRCUMFLEX AND DOLLAR</a>
21 <li><a name="TOC6" href="#SEC6">FULL STOP (PERIOD, DOT)</a>
22 <li><a name="TOC7" href="#SEC7">MATCHING A SINGLE BYTE</a>
24 <li><a name="TOC9" href="#SEC9">POSIX CHARACTER CLASSES</a>
25 <li><a name="TOC10" href="#SEC10">VERTICAL BAR</a>
26 <li><a name="TOC11" href="#SEC11">INTERNAL OPTION SETTING</a>
27 <li><a name="TOC12" href="#SEC12">SUBPATTERNS</a>
28 <li><a name="TOC13" href="#SEC13">DUPLICATE SUBPATTERN NUMBERS</a>
29 <li><a name="TOC14" href="#SEC14">NAMED SUBPATTERNS</a>
30 <li><a name="TOC15" href="#SEC15">REPETITION</a>
32 <li><a name="TOC17" href="#SEC17">BACK REFERENCES</a>
33 <li><a name="TOC18" href="#SEC18">ASSERTIONS</a>
34 <li><a name="TOC19" href="#SEC19">CONDITIONAL SUBPATTERNS</a>
35 <li><a name="TOC20" href="#SEC20">COMMENTS</a>
36 <li><a name="TOC21" href="#SEC21">RECURSIVE PATTERNS</a>
37 <li><a name="TOC22" href="#SEC22">SUBPATTERNS AS SUBROUTINES</a>
38 <li><a name="TOC23" href="#SEC23">CALLOUTS</a>
39 <li><a name="TOC24" href="#SEC24">BACTRACKING CONTROL</a>
40 <li><a name="TOC25" href="#SEC25">SEE ALSO</a>
41 <li><a name="TOC26" href="#SEC26">AUTHOR</a>
42 <li><a name="TOC27" href="#SEC27">REVISION</a>
43 </ul>
44 <br><a name="SEC1" href="#TOC1">PCRE REGULAR EXPRESSION DETAILS</a><br>
45 <P>
46 The syntax and semantics of the regular expressions that are supported by PCRE
47 are described in detail below. There is a quick-reference syntax summary in the
48 <a href="pcresyntax.html"><b>pcresyntax</b></a>
49 page. Perl's regular expressions are described in its own documentation, and
50 regular expressions in general are covered in a number of books, some of which
51 have copious examples. Jeffrey Friedl's "Mastering Regular Expressions",
52 published by O'Reilly, covers regular expressions in great detail. This
53 description of PCRE's regular expressions is intended as reference material.
54 </P>
55 <P>
56 The original operation of PCRE was on strings of one-byte characters. However,
57 there is now also support for UTF-8 character strings. To use this, you must
58 build PCRE to include UTF-8 support, and then call <b>pcre_compile()</b> with
59 the PCRE_UTF8 option. How this affects pattern matching is mentioned in several
60 places below. There is also a summary of UTF-8 features in the
61 <a href="pcre.html#utf8support">section on UTF-8 support</a>
62 in the main
63 <a href="pcre.html"><b>pcre</b></a>
64 page.
65 </P>
66 <P>
67 The remainder of this document discusses the patterns that are supported by
68 PCRE when its main matching function, <b>pcre_exec()</b>, is used.
69 From release 6.0, PCRE offers a second matching function,
70 <b>pcre_dfa_exec()</b>, which matches using a different algorithm that is not
71 Perl-compatible. Some of the features discussed below are not available when
72 <b>pcre_dfa_exec()</b> is used. The advantages and disadvantages of the
73 alternative function, and how it differs from the normal function, are
74 discussed in the
75 <a href="pcrematching.html"><b>pcrematching</b></a>
76 page.
77 </P>
78 <br><a name="SEC2" href="#TOC1">NEWLINE CONVENTIONS</a><br>
79 <P>
80 PCRE supports five different conventions for indicating line breaks in
81 strings: a single CR (carriage return) character, a single LF (linefeed)
82 character, the two-character sequence CRLF, any of the three preceding, or any
83 Unicode newline sequence. The
84 <a href="pcreapi.html"><b>pcreapi</b></a>
85 page has
86 <a href="pcreapi.html#newlines">further discussion</a>
87 about newlines, and shows how to set the newline convention in the
88 <i>options</i> arguments for the compiling and matching functions.
89 </P>
90 <P>
91 It is also possible to specify a newline convention by starting a pattern
92 string with one of the following five sequences:
93 <pre>
94 (*CR) carriage return
95 (*LF) linefeed
96 (*CRLF) carriage return, followed by linefeed
97 (*ANYCRLF) any of the three above
98 (*ANY) all Unicode newline sequences
99 </pre>
100 These override the default and the options given to <b>pcre_compile()</b>. For
101 example, on a Unix system where LF is the default newline sequence, the pattern
102 <pre>
103 (*CR)a.b
104 </pre>
105 changes the convention to CR. That pattern matches "a\nb" because LF is no
106 longer a newline. Note that these special settings, which are not
107 Perl-compatible, are recognized only at the very start of a pattern, and that
108 they must be in upper case.
109 </P>
110 <br><a name="SEC3" href="#TOC1">CHARACTERS AND METACHARACTERS</a><br>
111 <P>
112 A regular expression is a pattern that is matched against a subject string from
113 left to right. Most characters stand for themselves in a pattern, and match the
114 corresponding characters in the subject. As a trivial example, the pattern
115 <pre>
116 The quick brown fox
117 </pre>
118 matches a portion of a subject string that is identical to itself. When
119 caseless matching is specified (the PCRE_CASELESS option), letters are matched
120 independently of case. In UTF-8 mode, PCRE always understands the concept of
121 case for characters whose values are less than 128, so caseless matching is
122 always possible. For characters with higher values, the concept of case is
123 supported if PCRE is compiled with Unicode property support, but not otherwise.
124 If you want to use caseless matching for characters 128 and above, you must
125 ensure that PCRE is compiled with Unicode property support as well as with
126 UTF-8 support.
127 </P>
128 <P>
129 The power of regular expressions comes from the ability to include alternatives
130 and repetitions in the pattern. These are encoded in the pattern by the use of
131 <i>metacharacters</i>, which do not stand for themselves but instead are
132 interpreted in some special way.
133 </P>
134 <P>
135 There are two different sets of metacharacters: those that are recognized
136 anywhere in the pattern except within square brackets, and those that are
137 recognized within square brackets. Outside square brackets, the metacharacters
138 are as follows:
139 <pre>
140 \ general escape character with several uses
141 ^ assert start of string (or line, in multiline mode)
142 $ assert end of string (or line, in multiline mode)
143 . match any character except newline (by default)
144 [ start character class definition
145 | start of alternative branch
146 ( start subpattern
147 ) end subpattern
148 ? extends the meaning of (
149 also 0 or 1 quantifier
150 also quantifier minimizer
151 * 0 or more quantifier
152 + 1 or more quantifier
153 also "possessive quantifier"
154 { start min/max quantifier
155 </pre>
156 Part of a pattern that is in square brackets is called a "character class". In
157 a character class the only metacharacters are:
158 <pre>
159 \ general escape character
160 ^ negate the class, but only if the first character
161 - indicates character range
162 [ POSIX character class (only if followed by POSIX syntax)
163 ] terminates the character class
164 </pre>
165 The following sections describe the use of each of the metacharacters.
166 </P>
167 <br><a name="SEC4" href="#TOC1">BACKSLASH</a><br>
168 <P>
169 The backslash character has several uses. Firstly, if it is followed by a
170 non-alphanumeric character, it takes away any special meaning that character
171 may have. This use of backslash as an escape character applies both inside and
172 outside character classes.
173 </P>
174 <P>
175 For example, if you want to match a * character, you write \* in the pattern.
176 This escaping action applies whether or not the following character would
177 otherwise be interpreted as a metacharacter, so it is always safe to precede a
178 non-alphanumeric with backslash to specify that it stands for itself. In
179 particular, if you want to match a backslash, you write \\.
180 </P>
181 <P>
182 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
183 pattern (other than in a character class) and characters between a # outside
184 a character class and the next newline are ignored. An escaping backslash can
185 be used to include a whitespace or # character as part of the pattern.
186 </P>
187 <P>
188 If you want to remove the special meaning from a sequence of characters, you
189 can do so by putting them between \Q and \E. This is different from Perl in
190 that $ and @ are handled as literals in \Q...\E sequences in PCRE, whereas in
191 Perl, $ and @ cause variable interpolation. Note the following examples:
192 <pre>
193 Pattern PCRE matches Perl matches
195 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
196 \Qabc\$xyz\E abc\$xyz abc\$xyz
197 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
198 </pre>
199 The \Q...\E sequence is recognized both inside and outside character classes.
200 <a name="digitsafterbackslash"></a></P>
201 <br><b>
202 Non-printing characters
203 </b><br>
204 <P>
205 A second use of backslash provides a way of encoding non-printing characters
206 in patterns in a visible manner. There is no restriction on the appearance of
207 non-printing characters, apart from the binary zero that terminates a pattern,
208 but when a pattern is being prepared by text editing, it is usually easier to
209 use one of the following escape sequences than the binary character it
210 represents:
211 <pre>
212 \a alarm, that is, the BEL character (hex 07)
213 \cx "control-x", where x is any character
214 \e escape (hex 1B)
215 \f formfeed (hex 0C)
216 \n linefeed (hex 0A)
217 \r carriage return (hex 0D)
218 \t tab (hex 09)
219 \ddd character with octal code ddd, or backreference
220 \xhh character with hex code hh
221 \x{hhh..} character with hex code hhh..
222 </pre>
223 The precise effect of \cx is as follows: if x is a lower case letter, it
224 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
225 Thus \cz becomes hex 1A, but \c{ becomes hex 3B, while \c; becomes hex
226 7B.
227 </P>
228 <P>
229 After \x, from zero to two hexadecimal digits are read (letters can be in
230 upper or lower case). Any number of hexadecimal digits may appear between \x{
231 and }, but the value of the character code must be less than 256 in non-UTF-8
232 mode, and less than 2**31 in UTF-8 mode. That is, the maximum value in
233 hexadecimal is 7FFFFFFF. Note that this is bigger than the largest Unicode code
234 point, which is 10FFFF.
235 </P>
236 <P>
237 If characters other than hexadecimal digits appear between \x{ and }, or if
238 there is no terminating }, this form of escape is not recognized. Instead, the
239 initial \x will be interpreted as a basic hexadecimal escape, with no
240 following digits, giving a character whose value is zero.
241 </P>
242 <P>
243 Characters whose value is less than 256 can be defined by either of the two
244 syntaxes for \x. There is no difference in the way they are handled. For
245 example, \xdc is exactly the same as \x{dc}.
246 </P>
247 <P>
248 After \0 up to two further octal digits are read. If there are fewer than two
249 digits, just those that are present are used. Thus the sequence \0\x\07
250 specifies two binary zeros followed by a BEL character (code value 7). Make
251 sure you supply two digits after the initial zero if the pattern character that
252 follows is itself an octal digit.
253 </P>
254 <P>
255 The handling of a backslash followed by a digit other than 0 is complicated.
256 Outside a character class, PCRE reads it and any following digits as a decimal
257 number. If the number is less than 10, or if there have been at least that many
258 previous capturing left parentheses in the expression, the entire sequence is
259 taken as a <i>back reference</i>. A description of how this works is given
260 <a href="#backreferences">later,</a>
261 following the discussion of
262 <a href="#subpattern">parenthesized subpatterns.</a>
263 </P>
264 <P>
265 Inside a character class, or if the decimal number is greater than 9 and there
266 have not been that many capturing subpatterns, PCRE re-reads up to three octal
267 digits following the backslash, and uses them to generate a data character. Any
268 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
269 character specified in octal must be less than \400. In UTF-8 mode, values up
270 to \777 are permitted. For example:
271 <pre>
272 \040 is another way of writing a space
273 \40 is the same, provided there are fewer than 40 previous capturing subpatterns
274 \7 is always a back reference
275 \11 might be a back reference, or another way of writing a tab
276 \011 is always a tab
277 \0113 is a tab followed by the character "3"
278 \113 might be a back reference, otherwise the character with octal code 113
279 \377 might be a back reference, otherwise the byte consisting entirely of 1 bits
280 \81 is either a back reference, or a binary zero followed by the two characters "8" and "1"
281 </pre>
282 Note that octal values of 100 or greater must not be introduced by a leading
283 zero, because no more than three octal digits are ever read.
284 </P>
285 <P>
286 All the sequences that define a single character value can be used both inside
287 and outside character classes. In addition, inside a character class, the
288 sequence \b is interpreted as the backspace character (hex 08), and the
289 sequences \R and \X are interpreted as the characters "R" and "X",
290 respectively. Outside a character class, these sequences have different
291 meanings
292 <a href="#uniextseq">(see below).</a>
293 </P>
294 <br><b>
295 Absolute and relative back references
296 </b><br>
297 <P>
298 The sequence \g followed by an unsigned or a negative number, optionally
299 enclosed in braces, is an absolute or relative back reference. A named back
300 reference can be coded as \g{name}. Back references are discussed
301 <a href="#backreferences">later,</a>
302 following the discussion of
303 <a href="#subpattern">parenthesized subpatterns.</a>
304 </P>
305 <br><b>
306 Generic character types
307 </b><br>
308 <P>
309 Another use of backslash is for specifying generic character types. The
310 following are always recognized:
311 <pre>
312 \d any decimal digit
313 \D any character that is not a decimal digit
314 \h any horizontal whitespace character
315 \H any character that is not a horizontal whitespace character
316 \s any whitespace character
317 \S any character that is not a whitespace character
318 \v any vertical whitespace character
319 \V any character that is not a vertical whitespace character
320 \w any "word" character
321 \W any "non-word" character
322 </pre>
323 Each pair of escape sequences partitions the complete set of characters into
324 two disjoint sets. Any given character matches one, and only one, of each pair.
325 </P>
326 <P>
327 These character type sequences can appear both inside and outside character
328 classes. They each match one character of the appropriate type. If the current
329 matching point is at the end of the subject string, all of them fail, since
330 there is no character to match.
331 </P>
332 <P>
333 For compatibility with Perl, \s does not match the VT character (code 11).
334 This makes it different from the the POSIX "space" class. The \s characters
335 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
336 included in a Perl script, \s may match the VT character. In PCRE, it never
337 does.
338 </P>
339 <P>
340 In UTF-8 mode, characters with values greater than 128 never match \d, \s, or
341 \w, and always match \D, \S, and \W. This is true even when Unicode
342 character property support is available. These sequences retain their original
343 meanings from before UTF-8 support was available, mainly for efficiency
344 reasons.
345 </P>
346 <P>
347 The sequences \h, \H, \v, and \V are Perl 5.10 features. In contrast to the
348 other sequences, these do match certain high-valued codepoints in UTF-8 mode.
349 The horizontal space characters are:
350 <pre>
351 U+0009 Horizontal tab
352 U+0020 Space
353 U+00A0 Non-break space
354 U+1680 Ogham space mark
355 U+180E Mongolian vowel separator
356 U+2000 En quad
357 U+2001 Em quad
358 U+2002 En space
359 U+2003 Em space
360 U+2004 Three-per-em space
361 U+2005 Four-per-em space
362 U+2006 Six-per-em space
363 U+2007 Figure space
364 U+2008 Punctuation space
365 U+2009 Thin space
366 U+200A Hair space
367 U+202F Narrow no-break space
368 U+205F Medium mathematical space
369 U+3000 Ideographic space
370 </pre>
371 The vertical space characters are:
372 <pre>
373 U+000A Linefeed
374 U+000B Vertical tab
375 U+000C Formfeed
376 U+000D Carriage return
377 U+0085 Next line
378 U+2028 Line separator
379 U+2029 Paragraph separator
380 </PRE>
381 </P>
382 <P>
383 A "word" character is an underscore or any character less than 256 that is a
384 letter or digit. The definition of letters and digits is controlled by PCRE's
385 low-valued character tables, and may vary if locale-specific matching is taking
386 place (see
387 <a href="pcreapi.html#localesupport">"Locale support"</a>
388 in the
389 <a href="pcreapi.html"><b>pcreapi</b></a>
390 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
391 or "french" in Windows, some character codes greater than 128 are used for
392 accented letters, and these are matched by \w. The use of locales with Unicode
393 is discouraged.
394 </P>
395 <br><b>
396 Newline sequences
397 </b><br>
398 <P>
399 Outside a character class, the escape sequence \R matches any Unicode newline
400 sequence. This is a Perl 5.10 feature. In non-UTF-8 mode \R is equivalent to
401 the following:
402 <pre>
403 (?&#62;\r\n|\n|\x0b|\f|\r|\x85)
404 </pre>
405 This is an example of an "atomic group", details of which are given
406 <a href="#atomicgroup">below.</a>
407 This particular group matches either the two-character sequence CR followed by
408 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
409 U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
410 line, U+0085). The two-character sequence is treated as a single unit that
411 cannot be split.
412 </P>
413 <P>
414 In UTF-8 mode, two additional characters whose codepoints are greater than 255
415 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
416 Unicode character property support is not needed for these characters to be
417 recognized.
418 </P>
419 <P>
420 Inside a character class, \R matches the letter "R".
421 <a name="uniextseq"></a></P>
422 <br><b>
423 Unicode character properties
424 </b><br>
425 <P>
426 When PCRE is built with Unicode character property support, three additional
427 escape sequences that match characters with specific properties are available.
428 When not in UTF-8 mode, these sequences are of course limited to testing
429 characters whose codepoints are less than 256, but they do work in this mode.
430 The extra escape sequences are:
431 <pre>
432 \p{<i>xx</i>} a character with the <i>xx</i> property
433 \P{<i>xx</i>} a character without the <i>xx</i> property
434 \X an extended Unicode sequence
435 </pre>
436 The property names represented by <i>xx</i> above are limited to the Unicode
437 script names, the general category properties, and "Any", which matches any
438 character (including newline). Other properties such as "InMusicalSymbols" are
439 not currently supported by PCRE. Note that \P{Any} does not match any
440 characters, so always causes a match failure.
441 </P>
442 <P>
443 Sets of Unicode characters are defined as belonging to certain scripts. A
444 character from one of these sets can be matched using a script name. For
445 example:
446 <pre>
447 \p{Greek}
448 \P{Han}
449 </pre>
450 Those that are not part of an identified script are lumped together as
451 "Common". The current list of scripts is:
452 </P>
453 <P>
454 Arabic,
455 Armenian,
456 Balinese,
457 Bengali,
458 Bopomofo,
459 Braille,
460 Buginese,
461 Buhid,
462 Canadian_Aboriginal,
463 Cherokee,
464 Common,
465 Coptic,
466 Cuneiform,
467 Cypriot,
468 Cyrillic,
469 Deseret,
470 Devanagari,
471 Ethiopic,
472 Georgian,
473 Glagolitic,
474 Gothic,
475 Greek,
476 Gujarati,
477 Gurmukhi,
478 Han,
479 Hangul,
480 Hanunoo,
481 Hebrew,
482 Hiragana,
483 Inherited,
484 Kannada,
485 Katakana,
486 Kharoshthi,
487 Khmer,
488 Lao,
489 Latin,
490 Limbu,
491 Linear_B,
492 Malayalam,
493 Mongolian,
494 Myanmar,
495 New_Tai_Lue,
496 Nko,
497 Ogham,
498 Old_Italic,
499 Old_Persian,
500 Oriya,
501 Osmanya,
502 Phags_Pa,
503 Phoenician,
504 Runic,
505 Shavian,
506 Sinhala,
507 Syloti_Nagri,
508 Syriac,
509 Tagalog,
510 Tagbanwa,
511 Tai_Le,
512 Tamil,
513 Telugu,
514 Thaana,
515 Thai,
516 Tibetan,
517 Tifinagh,
518 Ugaritic,
519 Yi.
520 </P>
521 <P>
522 Each character has exactly one general category property, specified by a
523 two-letter abbreviation. For compatibility with Perl, negation can be specified
524 by including a circumflex between the opening brace and the property name. For
525 example, \p{^Lu} is the same as \P{Lu}.
526 </P>
527 <P>
528 If only one letter is specified with \p or \P, it includes all the general
529 category properties that start with that letter. In this case, in the absence
530 of negation, the curly brackets in the escape sequence are optional; these two
531 examples have the same effect:
532 <pre>
533 \p{L}
534 \pL
535 </pre>
536 The following general category property codes are supported:
537 <pre>
538 C Other
539 Cc Control
540 Cf Format
541 Cn Unassigned
542 Co Private use
543 Cs Surrogate
545 L Letter
546 Ll Lower case letter
547 Lm Modifier letter
548 Lo Other letter
549 Lt Title case letter
550 Lu Upper case letter
552 M Mark
553 Mc Spacing mark
554 Me Enclosing mark
555 Mn Non-spacing mark
557 N Number
558 Nd Decimal number
559 Nl Letter number
560 No Other number
562 P Punctuation
563 Pc Connector punctuation
564 Pd Dash punctuation
565 Pe Close punctuation
566 Pf Final punctuation
567 Pi Initial punctuation
568 Po Other punctuation
569 Ps Open punctuation
571 S Symbol
572 Sc Currency symbol
573 Sk Modifier symbol
574 Sm Mathematical symbol
575 So Other symbol
577 Z Separator
578 Zl Line separator
579 Zp Paragraph separator
580 Zs Space separator
581 </pre>
582 The special property L& is also supported: it matches a character that has
583 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
584 a modifier or "other".
585 </P>
586 <P>
587 The Cs (Surrogate) property applies only to characters in the range U+D800 to
588 U+DFFF. Such characters are not valid in UTF-8 strings (see RFC 3629) and so
589 cannot be tested by PCRE, unless UTF-8 validity checking has been turned off
590 (see the discussion of PCRE_NO_UTF8_CHECK in the
591 <a href="pcreapi.html"><b>pcreapi</b></a>
592 page).
593 </P>
594 <P>
595 The long synonyms for these properties that Perl supports (such as \p{Letter})
596 are not supported by PCRE, nor is it permitted to prefix any of these
597 properties with "Is".
598 </P>
599 <P>
600 No character that is in the Unicode table has the Cn (unassigned) property.
601 Instead, this property is assumed for any code point that is not in the
602 Unicode table.
603 </P>
604 <P>
605 Specifying caseless matching does not affect these escape sequences. For
606 example, \p{Lu} always matches only upper case letters.
607 </P>
608 <P>
609 The \X escape matches any number of Unicode characters that form an extended
610 Unicode sequence. \X is equivalent to
611 <pre>
612 (?&#62;\PM\pM*)
613 </pre>
614 That is, it matches a character without the "mark" property, followed by zero
615 or more characters with the "mark" property, and treats the sequence as an
616 atomic group
617 <a href="#atomicgroup">(see below).</a>
618 Characters with the "mark" property are typically accents that affect the
619 preceding character. None of them have codepoints less than 256, so in
620 non-UTF-8 mode \X matches any one character.
621 </P>
622 <P>
623 Matching characters by Unicode property is not fast, because PCRE has to search
624 a structure that contains data for over fifteen thousand characters. That is
625 why the traditional escape sequences such as \d and \w do not use Unicode
626 properties in PCRE.
627 <a name="resetmatchstart"></a></P>
628 <br><b>
629 Resetting the match start
630 </b><br>
631 <P>
632 The escape sequence \K, which is a Perl 5.10 feature, causes any previously
633 matched characters not to be included in the final matched sequence. For
634 example, the pattern:
635 <pre>
636 foo\Kbar
637 </pre>
638 matches "foobar", but reports that it has matched "bar". This feature is
639 similar to a lookbehind assertion
640 <a href="#lookbehind">(described below).</a>
641 However, in this case, the part of the subject before the real match does not
642 have to be of fixed length, as lookbehind assertions do. The use of \K does
643 not interfere with the setting of
644 <a href="#subpattern">captured substrings.</a>
645 For example, when the pattern
646 <pre>
647 (foo)\Kbar
648 </pre>
649 matches "foobar", the first substring is still set to "foo".
650 <a name="smallassertions"></a></P>
651 <br><b>
652 Simple assertions
653 </b><br>
654 <P>
655 The final use of backslash is for certain simple assertions. An assertion
656 specifies a condition that has to be met at a particular point in a match,
657 without consuming any characters from the subject string. The use of
658 subpatterns for more complicated assertions is described
659 <a href="#bigassertions">below.</a>
660 The backslashed assertions are:
661 <pre>
662 \b matches at a word boundary
663 \B matches when not at a word boundary
664 \A matches at the start of the subject
665 \Z matches at the end of the subject
666 also matches before a newline at the end of the subject
667 \z matches only at the end of the subject
668 \G matches at the first matching position in the subject
669 </pre>
670 These assertions may not appear in character classes (but note that \b has a
671 different meaning, namely the backspace character, inside a character class).
672 </P>
673 <P>
674 A word boundary is a position in the subject string where the current character
675 and the previous character do not both match \w or \W (i.e. one matches
676 \w and the other matches \W), or the start or end of the string if the
677 first or last character matches \w, respectively.
678 </P>
679 <P>
680 The \A, \Z, and \z assertions differ from the traditional circumflex and
681 dollar (described in the next section) in that they only ever match at the very
682 start and end of the subject string, whatever options are set. Thus, they are
683 independent of multiline mode. These three assertions are not affected by the
684 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
685 circumflex and dollar metacharacters. However, if the <i>startoffset</i>
686 argument of <b>pcre_exec()</b> is non-zero, indicating that matching is to start
687 at a point other than the beginning of the subject, \A can never match. The
688 difference between \Z and \z is that \Z matches before a newline at the end
689 of the string as well as at the very end, whereas \z matches only at the end.
690 </P>
691 <P>
692 The \G assertion is true only when the current matching position is at the
693 start point of the match, as specified by the <i>startoffset</i> argument of
694 <b>pcre_exec()</b>. It differs from \A when the value of <i>startoffset</i> is
695 non-zero. By calling <b>pcre_exec()</b> multiple times with appropriate
696 arguments, you can mimic Perl's /g option, and it is in this kind of
697 implementation where \G can be useful.
698 </P>
699 <P>
700 Note, however, that PCRE's interpretation of \G, as the start of the current
701 match, is subtly different from Perl's, which defines it as the end of the
702 previous match. In Perl, these can be different when the previously matched
703 string was empty. Because PCRE does just one match at a time, it cannot
704 reproduce this behaviour.
705 </P>
706 <P>
707 If all the alternatives of a pattern begin with \G, the expression is anchored
708 to the starting match position, and the "anchored" flag is set in the compiled
709 regular expression.
710 </P>
711 <br><a name="SEC5" href="#TOC1">CIRCUMFLEX AND DOLLAR</a><br>
712 <P>
713 Outside a character class, in the default matching mode, the circumflex
714 character is an assertion that is true only if the current matching point is
715 at the start of the subject string. If the <i>startoffset</i> argument of
716 <b>pcre_exec()</b> is non-zero, circumflex can never match if the PCRE_MULTILINE
717 option is unset. Inside a character class, circumflex has an entirely different
718 meaning
719 <a href="#characterclass">(see below).</a>
720 </P>
721 <P>
722 Circumflex need not be the first character of the pattern if a number of
723 alternatives are involved, but it should be the first thing in each alternative
724 in which it appears if the pattern is ever to match that branch. If all
725 possible alternatives start with a circumflex, that is, if the pattern is
726 constrained to match only at the start of the subject, it is said to be an
727 "anchored" pattern. (There are also other constructs that can cause a pattern
728 to be anchored.)
729 </P>
730 <P>
731 A dollar character is an assertion that is true only if the current matching
732 point is at the end of the subject string, or immediately before a newline
733 at the end of the string (by default). Dollar need not be the last character of
734 the pattern if a number of alternatives are involved, but it should be the last
735 item in any branch in which it appears. Dollar has no special meaning in a
736 character class.
737 </P>
738 <P>
739 The meaning of dollar can be changed so that it matches only at the very end of
740 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
741 does not affect the \Z assertion.
742 </P>
743 <P>
744 The meanings of the circumflex and dollar characters are changed if the
745 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
746 immediately after internal newlines as well as at the start of the subject
747 string. It does not match after a newline that ends the string. A dollar
748 matches before any newlines in the string, as well as at the very end, when
749 PCRE_MULTILINE is set. When newline is specified as the two-character
750 sequence CRLF, isolated CR and LF characters do not indicate newlines.
751 </P>
752 <P>
753 For example, the pattern /^abc$/ matches the subject string "def\nabc" (where
754 \n represents a newline) in multiline mode, but not otherwise. Consequently,
755 patterns that are anchored in single line mode because all branches start with
756 ^ are not anchored in multiline mode, and a match for circumflex is possible
757 when the <i>startoffset</i> argument of <b>pcre_exec()</b> is non-zero. The
758 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
759 </P>
760 <P>
761 Note that the sequences \A, \Z, and \z can be used to match the start and
762 end of the subject in both modes, and if all branches of a pattern start with
763 \A it is always anchored, whether or not PCRE_MULTILINE is set.
764 </P>
765 <br><a name="SEC6" href="#TOC1">FULL STOP (PERIOD, DOT)</a><br>
766 <P>
767 Outside a character class, a dot in the pattern matches any one character in
768 the subject string except (by default) a character that signifies the end of a
769 line. In UTF-8 mode, the matched character may be more than one byte long.
770 </P>
771 <P>
772 When a line ending is defined as a single character, dot never matches that
773 character; when the two-character sequence CRLF is used, dot does not match CR
774 if it is immediately followed by LF, but otherwise it matches all characters
775 (including isolated CRs and LFs). When any Unicode line endings are being
776 recognized, dot does not match CR or LF or any of the other line ending
777 characters.
778 </P>
779 <P>
780 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
781 option is set, a dot matches any one character, without exception. If the
782 two-character sequence CRLF is present in the subject string, it takes two dots
783 to match it.
784 </P>
785 <P>
786 The handling of dot is entirely independent of the handling of circumflex and
787 dollar, the only relationship being that they both involve newlines. Dot has no
788 special meaning in a character class.
789 </P>
790 <br><a name="SEC7" href="#TOC1">MATCHING A SINGLE BYTE</a><br>
791 <P>
792 Outside a character class, the escape sequence \C matches any one byte, both
793 in and out of UTF-8 mode. Unlike a dot, it always matches any line-ending
794 characters. The feature is provided in Perl in order to match individual bytes
795 in UTF-8 mode. Because it breaks up UTF-8 characters into individual bytes,
796 what remains in the string may be a malformed UTF-8 string. For this reason,
797 the \C escape sequence is best avoided.
798 </P>
799 <P>
800 PCRE does not allow \C to appear in lookbehind assertions
801 <a href="#lookbehind">(described below),</a>
802 because in UTF-8 mode this would make it impossible to calculate the length of
803 the lookbehind.
804 <a name="characterclass"></a></P>
805 <br><a name="SEC8" href="#TOC1">SQUARE BRACKETS AND CHARACTER CLASSES</a><br>
806 <P>
807 An opening square bracket introduces a character class, terminated by a closing
808 square bracket. A closing square bracket on its own is not special. If a
809 closing square bracket is required as a member of the class, it should be the
810 first data character in the class (after an initial circumflex, if present) or
811 escaped with a backslash.
812 </P>
813 <P>
814 A character class matches a single character in the subject. In UTF-8 mode, the
815 character may occupy more than one byte. A matched character must be in the set
816 of characters defined by the class, unless the first character in the class
817 definition is a circumflex, in which case the subject character must not be in
818 the set defined by the class. If a circumflex is actually required as a member
819 of the class, ensure it is not the first character, or escape it with a
820 backslash.
821 </P>
822 <P>
823 For example, the character class [aeiou] matches any lower case vowel, while
824 [^aeiou] matches any character that is not a lower case vowel. Note that a
825 circumflex is just a convenient notation for specifying the characters that
826 are in the class by enumerating those that are not. A class that starts with a
827 circumflex is not an assertion: it still consumes a character from the subject
828 string, and therefore it fails if the current pointer is at the end of the
829 string.
830 </P>
831 <P>
832 In UTF-8 mode, characters with values greater than 255 can be included in a
833 class as a literal string of bytes, or by using the \x{ escaping mechanism.
834 </P>
835 <P>
836 When caseless matching is set, any letters in a class represent both their
837 upper case and lower case versions, so for example, a caseless [aeiou] matches
838 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
839 caseful version would. In UTF-8 mode, PCRE always understands the concept of
840 case for characters whose values are less than 128, so caseless matching is
841 always possible. For characters with higher values, the concept of case is
842 supported if PCRE is compiled with Unicode property support, but not otherwise.
843 If you want to use caseless matching for characters 128 and above, you must
844 ensure that PCRE is compiled with Unicode property support as well as with
845 UTF-8 support.
846 </P>
847 <P>
848 Characters that might indicate line breaks are never treated in any special way
849 when matching character classes, whatever line-ending sequence is in use, and
850 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
851 such as [^a] always matches one of these characters.
852 </P>
853 <P>
854 The minus (hyphen) character can be used to specify a range of characters in a
855 character class. For example, [d-m] matches any letter between d and m,
856 inclusive. If a minus character is required in a class, it must be escaped with
857 a backslash or appear in a position where it cannot be interpreted as
858 indicating a range, typically as the first or last character in the class.
859 </P>
860 <P>
861 It is not possible to have the literal character "]" as the end character of a
862 range. A pattern such as [W-]46] is interpreted as a class of two characters
863 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
864 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
865 the end of range, so [W-\]46] is interpreted as a class containing a range
866 followed by two other characters. The octal or hexadecimal representation of
867 "]" can also be used to end a range.
868 </P>
869 <P>
870 Ranges operate in the collating sequence of character values. They can also be
871 used for characters specified numerically, for example [\000-\037]. In UTF-8
872 mode, ranges can include characters whose values are greater than 255, for
873 example [\x{100}-\x{2ff}].
874 </P>
875 <P>
876 If a range that includes letters is used when caseless matching is set, it
877 matches the letters in either case. For example, [W-c] is equivalent to
878 [][\\^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
879 tables for a French locale are in use, [\xc8-\xcb] matches accented E
880 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
881 characters with values greater than 128 only when it is compiled with Unicode
882 property support.
883 </P>
884 <P>
885 The character types \d, \D, \p, \P, \s, \S, \w, and \W may also appear
886 in a character class, and add the characters that they match to the class. For
887 example, [\dABCDEF] matches any hexadecimal digit. A circumflex can
888 conveniently be used with the upper case character types to specify a more
889 restricted set of characters than the matching lower case type. For example,
890 the class [^\W_] matches any letter or digit, but not underscore.
891 </P>
892 <P>
893 The only metacharacters that are recognized in character classes are backslash,
894 hyphen (only where it can be interpreted as specifying a range), circumflex
895 (only at the start), opening square bracket (only when it can be interpreted as
896 introducing a POSIX class name - see the next section), and the terminating
897 closing square bracket. However, escaping other non-alphanumeric characters
898 does no harm.
899 </P>
900 <br><a name="SEC9" href="#TOC1">POSIX CHARACTER CLASSES</a><br>
901 <P>
902 Perl supports the POSIX notation for character classes. This uses names
903 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
904 this notation. For example,
905 <pre>
906 [01[:alpha:]%]
907 </pre>
908 matches "0", "1", any alphabetic character, or "%". The supported class names
909 are
910 <pre>
911 alnum letters and digits
912 alpha letters
913 ascii character codes 0 - 127
914 blank space or tab only
915 cntrl control characters
916 digit decimal digits (same as \d)
917 graph printing characters, excluding space
918 lower lower case letters
919 print printing characters, including space
920 punct printing characters, excluding letters and digits
921 space white space (not quite the same as \s)
922 upper upper case letters
923 word "word" characters (same as \w)
924 xdigit hexadecimal digits
925 </pre>
926 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
927 space (32). Notice that this list includes the VT character (code 11). This
928 makes "space" different to \s, which does not include VT (for Perl
929 compatibility).
930 </P>
931 <P>
932 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
933 5.8. Another Perl extension is negation, which is indicated by a ^ character
934 after the colon. For example,
935 <pre>
936 [12[:^digit:]]
937 </pre>
938 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
939 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
940 supported, and an error is given if they are encountered.
941 </P>
942 <P>
943 In UTF-8 mode, characters with values greater than 128 do not match any of
944 the POSIX character classes.
945 </P>
946 <br><a name="SEC10" href="#TOC1">VERTICAL BAR</a><br>
947 <P>
948 Vertical bar characters are used to separate alternative patterns. For example,
949 the pattern
950 <pre>
951 gilbert|sullivan
952 </pre>
953 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
954 and an empty alternative is permitted (matching the empty string). The matching
955 process tries each alternative in turn, from left to right, and the first one
956 that succeeds is used. If the alternatives are within a subpattern
957 <a href="#subpattern">(defined below),</a>
958 "succeeds" means matching the rest of the main pattern as well as the
959 alternative in the subpattern.
960 </P>
961 <br><a name="SEC11" href="#TOC1">INTERNAL OPTION SETTING</a><br>
962 <P>
964 PCRE_EXTENDED options can be changed from within the pattern by a sequence of
965 Perl option letters enclosed between "(?" and ")". The option letters are
966 <pre>
969 s for PCRE_DOTALL
971 </pre>
972 For example, (?im) sets caseless, multiline matching. It is also possible to
973 unset these options by preceding the letter with a hyphen, and a combined
974 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
975 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
976 permitted. If a letter appears both before and after the hyphen, the option is
977 unset.
978 </P>
979 <P>
980 When an option change occurs at top level (that is, not inside subpattern
981 parentheses), the change applies to the remainder of the pattern that follows.
982 If the change is placed right at the start of a pattern, PCRE extracts it into
983 the global options (and it will therefore show up in data extracted by the
984 <b>pcre_fullinfo()</b> function).
985 </P>
986 <P>
987 An option change within a subpattern (see below for a description of
988 subpatterns) affects only that part of the current pattern that follows it, so
989 <pre>
990 (a(?i)b)c
991 </pre>
992 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
993 By this means, options can be made to have different settings in different
994 parts of the pattern. Any changes made in one alternative do carry on
995 into subsequent branches within the same subpattern. For example,
996 <pre>
997 (a(?i)b|c)
998 </pre>
999 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1000 branch is abandoned before the option setting. This is because the effects of
1001 option settings happen at compile time. There would be some very weird
1002 behaviour otherwise.
1003 </P>
1004 <P>
1005 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1006 changed in the same way as the Perl-compatible options by using the characters
1007 J, U and X respectively.
1008 <a name="subpattern"></a></P>
1009 <br><a name="SEC12" href="#TOC1">SUBPATTERNS</a><br>
1010 <P>
1011 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1012 Turning part of a pattern into a subpattern does two things:
1013 <br>
1014 <br>
1015 1. It localizes a set of alternatives. For example, the pattern
1016 <pre>
1017 cat(aract|erpillar|)
1018 </pre>
1019 matches one of the words "cat", "cataract", or "caterpillar". Without the
1020 parentheses, it would match "cataract", "erpillar" or an empty string.
1021 <br>
1022 <br>
1023 2. It sets up the subpattern as a capturing subpattern. This means that, when
1024 the whole pattern matches, that portion of the subject string that matched the
1025 subpattern is passed back to the caller via the <i>ovector</i> argument of
1026 <b>pcre_exec()</b>. Opening parentheses are counted from left to right (starting
1027 from 1) to obtain numbers for the capturing subpatterns.
1028 </P>
1029 <P>
1030 For example, if the string "the red king" is matched against the pattern
1031 <pre>
1032 the ((red|white) (king|queen))
1033 </pre>
1034 the captured substrings are "red king", "red", and "king", and are numbered 1,
1035 2, and 3, respectively.
1036 </P>
1037 <P>
1038 The fact that plain parentheses fulfil two functions is not always helpful.
1039 There are often times when a grouping subpattern is required without a
1040 capturing requirement. If an opening parenthesis is followed by a question mark
1041 and a colon, the subpattern does not do any capturing, and is not counted when
1042 computing the number of any subsequent capturing subpatterns. For example, if
1043 the string "the white queen" is matched against the pattern
1044 <pre>
1045 the ((?:red|white) (king|queen))
1046 </pre>
1047 the captured substrings are "white queen" and "queen", and are numbered 1 and
1048 2. The maximum number of capturing subpatterns is 65535.
1049 </P>
1050 <P>
1051 As a convenient shorthand, if any option settings are required at the start of
1052 a non-capturing subpattern, the option letters may appear between the "?" and
1053 the ":". Thus the two patterns
1054 <pre>
1055 (?i:saturday|sunday)
1056 (?:(?i)saturday|sunday)
1057 </pre>
1058 match exactly the same set of strings. Because alternative branches are tried
1059 from left to right, and options are not reset until the end of the subpattern
1060 is reached, an option setting in one branch does affect subsequent branches, so
1061 the above patterns match "SUNDAY" as well as "Saturday".
1062 </P>
1063 <br><a name="SEC13" href="#TOC1">DUPLICATE SUBPATTERN NUMBERS</a><br>
1064 <P>
1065 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1066 the same numbers for its capturing parentheses. Such a subpattern starts with
1067 (?| and is itself a non-capturing subpattern. For example, consider this
1068 pattern:
1069 <pre>
1070 (?|(Sat)ur|(Sun))day
1071 </pre>
1072 Because the two alternatives are inside a (?| group, both sets of capturing
1073 parentheses are numbered one. Thus, when the pattern matches, you can look
1074 at captured substring number one, whichever alternative matched. This construct
1075 is useful when you want to capture part, but not all, of one of a number of
1076 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1077 number is reset at the start of each branch. The numbers of any capturing
1078 buffers that follow the subpattern start after the highest number used in any
1079 branch. The following example is taken from the Perl documentation.
1080 The numbers underneath show in which buffer the captured content will be
1081 stored.
1082 <pre>
1083 # before ---------------branch-reset----------- after
1084 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1085 # 1 2 2 3 2 3 4
1086 </pre>
1087 A backreference or a recursive call to a numbered subpattern always refers to
1088 the first one in the pattern with the given number.
1089 </P>
1090 <P>
1091 An alternative approach to using this "branch reset" feature is to use
1092 duplicate named subpatterns, as described in the next section.
1093 </P>
1094 <br><a name="SEC14" href="#TOC1">NAMED SUBPATTERNS</a><br>
1095 <P>
1096 Identifying capturing parentheses by number is simple, but it can be very hard
1097 to keep track of the numbers in complicated regular expressions. Furthermore,
1098 if an expression is modified, the numbers may change. To help with this
1099 difficulty, PCRE supports the naming of subpatterns. This feature was not
1100 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1101 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1102 the Perl and the Python syntax.
1103 </P>
1104 <P>
1105 In PCRE, a subpattern can be named in one of three ways: (?&#60;name&#62;...) or
1106 (?'name'...) as in Perl, or (?P&#60;name&#62;...) as in Python. References to capturing
1107 parentheses from other parts of the pattern, such as
1108 <a href="#backreferences">backreferences,</a>
1109 <a href="#recursion">recursion,</a>
1110 and
1111 <a href="#conditions">conditions,</a>
1112 can be made by name as well as by number.
1113 </P>
1114 <P>
1115 Names consist of up to 32 alphanumeric characters and underscores. Named
1116 capturing parentheses are still allocated numbers as well as names, exactly as
1117 if the names were not present. The PCRE API provides function calls for
1118 extracting the name-to-number translation table from a compiled pattern. There
1119 is also a convenience function for extracting a captured substring by name.
1120 </P>
1121 <P>
1122 By default, a name must be unique within a pattern, but it is possible to relax
1123 this constraint by setting the PCRE_DUPNAMES option at compile time. This can
1124 be useful for patterns where only one instance of the named parentheses can
1125 match. Suppose you want to match the name of a weekday, either as a 3-letter
1126 abbreviation or as the full name, and in both cases you want to extract the
1127 abbreviation. This pattern (ignoring the line breaks) does the job:
1128 <pre>
1129 (?&#60;DN&#62;Mon|Fri|Sun)(?:day)?|
1130 (?&#60;DN&#62;Tue)(?:sday)?|
1131 (?&#60;DN&#62;Wed)(?:nesday)?|
1132 (?&#60;DN&#62;Thu)(?:rsday)?|
1133 (?&#60;DN&#62;Sat)(?:urday)?
1134 </pre>
1135 There are five capturing substrings, but only one is ever set after a match.
1136 (An alternative way of solving this problem is to use a "branch reset"
1137 subpattern, as described in the previous section.)
1138 </P>
1139 <P>
1140 The convenience function for extracting the data by name returns the substring
1141 for the first (and in this example, the only) subpattern of that name that
1142 matched. This saves searching to find which numbered subpattern it was. If you
1143 make a reference to a non-unique named subpattern from elsewhere in the
1144 pattern, the one that corresponds to the lowest number is used. For further
1145 details of the interfaces for handling named subpatterns, see the
1146 <a href="pcreapi.html"><b>pcreapi</b></a>
1147 documentation.
1148 </P>
1149 <br><a name="SEC15" href="#TOC1">REPETITION</a><br>
1150 <P>
1151 Repetition is specified by quantifiers, which can follow any of the following
1152 items:
1153 <pre>
1154 a literal data character
1155 the dot metacharacter
1156 the \C escape sequence
1157 the \X escape sequence (in UTF-8 mode with Unicode properties)
1158 the \R escape sequence
1159 an escape such as \d that matches a single character
1160 a character class
1161 a back reference (see next section)
1162 a parenthesized subpattern (unless it is an assertion)
1163 </pre>
1164 The general repetition quantifier specifies a minimum and maximum number of
1165 permitted matches, by giving the two numbers in curly brackets (braces),
1166 separated by a comma. The numbers must be less than 65536, and the first must
1167 be less than or equal to the second. For example:
1168 <pre>
1169 z{2,4}
1170 </pre>
1171 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1172 character. If the second number is omitted, but the comma is present, there is
1173 no upper limit; if the second number and the comma are both omitted, the
1174 quantifier specifies an exact number of required matches. Thus
1175 <pre>
1176 [aeiou]{3,}
1177 </pre>
1178 matches at least 3 successive vowels, but may match many more, while
1179 <pre>
1180 \d{8}
1181 </pre>
1182 matches exactly 8 digits. An opening curly bracket that appears in a position
1183 where a quantifier is not allowed, or one that does not match the syntax of a
1184 quantifier, is taken as a literal character. For example, {,6} is not a
1185 quantifier, but a literal string of four characters.
1186 </P>
1187 <P>
1188 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1189 bytes. Thus, for example, \x{100}{2} matches two UTF-8 characters, each of
1190 which is represented by a two-byte sequence. Similarly, when Unicode property
1191 support is available, \X{3} matches three Unicode extended sequences, each of
1192 which may be several bytes long (and they may be of different lengths).
1193 </P>
1194 <P>
1195 The quantifier {0} is permitted, causing the expression to behave as if the
1196 previous item and the quantifier were not present.
1197 </P>
1198 <P>
1199 For convenience, the three most common quantifiers have single-character
1200 abbreviations:
1201 <pre>
1202 * is equivalent to {0,}
1203 + is equivalent to {1,}
1204 ? is equivalent to {0,1}
1205 </pre>
1206 It is possible to construct infinite loops by following a subpattern that can
1207 match no characters with a quantifier that has no upper limit, for example:
1208 <pre>
1209 (a?)*
1210 </pre>
1211 Earlier versions of Perl and PCRE used to give an error at compile time for
1212 such patterns. However, because there are cases where this can be useful, such
1213 patterns are now accepted, but if any repetition of the subpattern does in fact
1214 match no characters, the loop is forcibly broken.
1215 </P>
1216 <P>
1217 By default, the quantifiers are "greedy", that is, they match as much as
1218 possible (up to the maximum number of permitted times), without causing the
1219 rest of the pattern to fail. The classic example of where this gives problems
1220 is in trying to match comments in C programs. These appear between /* and */
1221 and within the comment, individual * and / characters may appear. An attempt to
1222 match C comments by applying the pattern
1223 <pre>
1224 /\*.*\*/
1225 </pre>
1226 to the string
1227 <pre>
1228 /* first comment */ not comment /* second comment */
1229 </pre>
1230 fails, because it matches the entire string owing to the greediness of the .*
1231 item.
1232 </P>
1233 <P>
1234 However, if a quantifier is followed by a question mark, it ceases to be
1235 greedy, and instead matches the minimum number of times possible, so the
1236 pattern
1237 <pre>
1238 /\*.*?\*/
1239 </pre>
1240 does the right thing with the C comments. The meaning of the various
1241 quantifiers is not otherwise changed, just the preferred number of matches.
1242 Do not confuse this use of question mark with its use as a quantifier in its
1243 own right. Because it has two uses, it can sometimes appear doubled, as in
1244 <pre>
1245 \d??\d
1246 </pre>
1247 which matches one digit by preference, but can match two if that is the only
1248 way the rest of the pattern matches.
1249 </P>
1250 <P>
1251 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1252 the quantifiers are not greedy by default, but individual ones can be made
1253 greedy by following them with a question mark. In other words, it inverts the
1254 default behaviour.
1255 </P>
1256 <P>
1257 When a parenthesized subpattern is quantified with a minimum repeat count that
1258 is greater than 1 or with a limited maximum, more memory is required for the
1259 compiled pattern, in proportion to the size of the minimum or maximum.
1260 </P>
1261 <P>
1262 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1263 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1264 implicitly anchored, because whatever follows will be tried against every
1265 character position in the subject string, so there is no point in retrying the
1266 overall match at any position after the first. PCRE normally treats such a
1267 pattern as though it were preceded by \A.
1268 </P>
1269 <P>
1270 In cases where it is known that the subject string contains no newlines, it is
1271 worth setting PCRE_DOTALL in order to obtain this optimization, or
1272 alternatively using ^ to indicate anchoring explicitly.
1273 </P>
1274 <P>
1275 However, there is one situation where the optimization cannot be used. When .*
1276 is inside capturing parentheses that are the subject of a backreference
1277 elsewhere in the pattern, a match at the start may fail where a later one
1278 succeeds. Consider, for example:
1279 <pre>
1280 (.*)abc\1
1281 </pre>
1282 If the subject is "xyz123abc123" the match point is the fourth character. For
1283 this reason, such a pattern is not implicitly anchored.
1284 </P>
1285 <P>
1286 When a capturing subpattern is repeated, the value captured is the substring
1287 that matched the final iteration. For example, after
1288 <pre>
1289 (tweedle[dume]{3}\s*)+
1290 </pre>
1291 has matched "tweedledum tweedledee" the value of the captured substring is
1292 "tweedledee". However, if there are nested capturing subpatterns, the
1293 corresponding captured values may have been set in previous iterations. For
1294 example, after
1295 <pre>
1296 /(a|(b))+/
1297 </pre>
1298 matches "aba" the value of the second captured substring is "b".
1299 <a name="atomicgroup"></a></P>
1300 <br><a name="SEC16" href="#TOC1">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a><br>
1301 <P>
1302 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1303 repetition, failure of what follows normally causes the repeated item to be
1304 re-evaluated to see if a different number of repeats allows the rest of the
1305 pattern to match. Sometimes it is useful to prevent this, either to change the
1306 nature of the match, or to cause it fail earlier than it otherwise might, when
1307 the author of the pattern knows there is no point in carrying on.
1308 </P>
1309 <P>
1310 Consider, for example, the pattern \d+foo when applied to the subject line
1311 <pre>
1312 123456bar
1313 </pre>
1314 After matching all 6 digits and then failing to match "foo", the normal
1315 action of the matcher is to try again with only 5 digits matching the \d+
1316 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1317 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1318 that once a subpattern has matched, it is not to be re-evaluated in this way.
1319 </P>
1320 <P>
1321 If we use atomic grouping for the previous example, the matcher gives up
1322 immediately on failing to match "foo" the first time. The notation is a kind of
1323 special parenthesis, starting with (?&#62; as in this example:
1324 <pre>
1325 (?&#62;\d+)foo
1326 </pre>
1327 This kind of parenthesis "locks up" the part of the pattern it contains once
1328 it has matched, and a failure further into the pattern is prevented from
1329 backtracking into it. Backtracking past it to previous items, however, works as
1330 normal.
1331 </P>
1332 <P>
1333 An alternative description is that a subpattern of this type matches the string
1334 of characters that an identical standalone pattern would match, if anchored at
1335 the current point in the subject string.
1336 </P>
1337 <P>
1338 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1339 the above example can be thought of as a maximizing repeat that must swallow
1340 everything it can. So, while both \d+ and \d+? are prepared to adjust the
1341 number of digits they match in order to make the rest of the pattern match,
1342 (?&#62;\d+) can only match an entire sequence of digits.
1343 </P>
1344 <P>
1345 Atomic groups in general can of course contain arbitrarily complicated
1346 subpatterns, and can be nested. However, when the subpattern for an atomic
1347 group is just a single repeated item, as in the example above, a simpler
1348 notation, called a "possessive quantifier" can be used. This consists of an
1349 additional + character following a quantifier. Using this notation, the
1350 previous example can be rewritten as
1351 <pre>
1352 \d++foo
1353 </pre>
1354 Note that a possessive quantifier can be used with an entire group, for
1355 example:
1356 <pre>
1357 (abc|xyz){2,3}+
1358 </pre>
1359 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1360 option is ignored. They are a convenient notation for the simpler forms of
1361 atomic group. However, there is no difference in the meaning of a possessive
1362 quantifier and the equivalent atomic group, though there may be a performance
1363 difference; possessive quantifiers should be slightly faster.
1364 </P>
1365 <P>
1366 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1367 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1368 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1369 package, and PCRE copied it from there. It ultimately found its way into Perl
1370 at release 5.10.
1371 </P>
1372 <P>
1373 PCRE has an optimization that automatically "possessifies" certain simple
1374 pattern constructs. For example, the sequence A+B is treated as A++B because
1375 there is no point in backtracking into a sequence of A's when B must follow.
1376 </P>
1377 <P>
1378 When a pattern contains an unlimited repeat inside a subpattern that can itself
1379 be repeated an unlimited number of times, the use of an atomic group is the
1380 only way to avoid some failing matches taking a very long time indeed. The
1381 pattern
1382 <pre>
1383 (\D+|&#60;\d+&#62;)*[!?]
1384 </pre>
1385 matches an unlimited number of substrings that either consist of non-digits, or
1386 digits enclosed in &#60;&#62;, followed by either ! or ?. When it matches, it runs
1387 quickly. However, if it is applied to
1388 <pre>
1389 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1390 </pre>
1391 it takes a long time before reporting failure. This is because the string can
1392 be divided between the internal \D+ repeat and the external * repeat in a
1393 large number of ways, and all have to be tried. (The example uses [!?] rather
1394 than a single character at the end, because both PCRE and Perl have an
1395 optimization that allows for fast failure when a single character is used. They
1396 remember the last single character that is required for a match, and fail early
1397 if it is not present in the string.) If the pattern is changed so that it uses
1398 an atomic group, like this:
1399 <pre>
1400 ((?&#62;\D+)|&#60;\d+&#62;)*[!?]
1401 </pre>
1402 sequences of non-digits cannot be broken, and failure happens quickly.
1403 <a name="backreferences"></a></P>
1404 <br><a name="SEC17" href="#TOC1">BACK REFERENCES</a><br>
1405 <P>
1406 Outside a character class, a backslash followed by a digit greater than 0 (and
1407 possibly further digits) is a back reference to a capturing subpattern earlier
1408 (that is, to its left) in the pattern, provided there have been that many
1409 previous capturing left parentheses.
1410 </P>
1411 <P>
1412 However, if the decimal number following the backslash is less than 10, it is
1413 always taken as a back reference, and causes an error only if there are not
1414 that many capturing left parentheses in the entire pattern. In other words, the
1415 parentheses that are referenced need not be to the left of the reference for
1416 numbers less than 10. A "forward back reference" of this type can make sense
1417 when a repetition is involved and the subpattern to the right has participated
1418 in an earlier iteration.
1419 </P>
1420 <P>
1421 It is not possible to have a numerical "forward back reference" to a subpattern
1422 whose number is 10 or more using this syntax because a sequence such as \50 is
1423 interpreted as a character defined in octal. See the subsection entitled
1424 "Non-printing characters"
1425 <a href="#digitsafterbackslash">above</a>
1426 for further details of the handling of digits following a backslash. There is
1427 no such problem when named parentheses are used. A back reference to any
1428 subpattern is possible using named parentheses (see below).
1429 </P>
1430 <P>
1431 Another way of avoiding the ambiguity inherent in the use of digits following a
1432 backslash is to use the \g escape sequence, which is a feature introduced in
1433 Perl 5.10. This escape must be followed by an unsigned number or a negative
1434 number, optionally enclosed in braces. These examples are all identical:
1435 <pre>
1436 (ring), \1
1437 (ring), \g1
1438 (ring), \g{1}
1439 </pre>
1440 An unsigned number specifies an absolute reference without the ambiguity that
1441 is present in the older syntax. It is also useful when literal digits follow
1442 the reference. A negative number is a relative reference. Consider this
1443 example:
1444 <pre>
1445 (abc(def)ghi)\g{-1}
1446 </pre>
1447 The sequence \g{-1} is a reference to the most recently started capturing
1448 subpattern before \g, that is, is it equivalent to \2. Similarly, \g{-2}
1449 would be equivalent to \1. The use of relative references can be helpful in
1450 long patterns, and also in patterns that are created by joining together
1451 fragments that contain references within themselves.
1452 </P>
1453 <P>
1454 A back reference matches whatever actually matched the capturing subpattern in
1455 the current subject string, rather than anything matching the subpattern
1456 itself (see
1457 <a href="#subpatternsassubroutines">"Subpatterns as subroutines"</a>
1458 below for a way of doing that). So the pattern
1459 <pre>
1460 (sens|respons)e and \1ibility
1461 </pre>
1462 matches "sense and sensibility" and "response and responsibility", but not
1463 "sense and responsibility". If caseful matching is in force at the time of the
1464 back reference, the case of letters is relevant. For example,
1465 <pre>
1466 ((?i)rah)\s+\1
1467 </pre>
1468 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1469 capturing subpattern is matched caselessly.
1470 </P>
1471 <P>
1472 There are several different ways of writing back references to named
1473 subpatterns. The .NET syntax \k{name} and the Perl syntax \k&#60;name&#62; or
1474 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1475 back reference syntax, in which \g can be used for both numeric and named
1476 references, is also supported. We could rewrite the above example in any of
1477 the following ways:
1478 <pre>
1479 (?&#60;p1&#62;(?i)rah)\s+\k&#60;p1&#62;
1480 (?'p1'(?i)rah)\s+\k{p1}
1481 (?P&#60;p1&#62;(?i)rah)\s+(?P=p1)
1482 (?&#60;p1&#62;(?i)rah)\s+\g{p1}
1483 </pre>
1484 A subpattern that is referenced by name may appear in the pattern before or
1485 after the reference.
1486 </P>
1487 <P>
1488 There may be more than one back reference to the same subpattern. If a
1489 subpattern has not actually been used in a particular match, any back
1490 references to it always fail. For example, the pattern
1491 <pre>
1492 (a|(bc))\2
1493 </pre>
1494 always fails if it starts to match "a" rather than "bc". Because there may be
1495 many capturing parentheses in a pattern, all digits following the backslash are
1496 taken as part of a potential back reference number. If the pattern continues
1497 with a digit character, some delimiter must be used to terminate the back
1498 reference. If the PCRE_EXTENDED option is set, this can be whitespace.
1499 Otherwise an empty comment (see
1500 <a href="#comments">"Comments"</a>
1501 below) can be used.
1502 </P>
1503 <P>
1504 A back reference that occurs inside the parentheses to which it refers fails
1505 when the subpattern is first used, so, for example, (a\1) never matches.
1506 However, such references can be useful inside repeated subpatterns. For
1507 example, the pattern
1508 <pre>
1509 (a|b\1)+
1510 </pre>
1511 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1512 the subpattern, the back reference matches the character string corresponding
1513 to the previous iteration. In order for this to work, the pattern must be such
1514 that the first iteration does not need to match the back reference. This can be
1515 done using alternation, as in the example above, or by a quantifier with a
1516 minimum of zero.
1517 <a name="bigassertions"></a></P>
1518 <br><a name="SEC18" href="#TOC1">ASSERTIONS</a><br>
1519 <P>
1520 An assertion is a test on the characters following or preceding the current
1521 matching point that does not actually consume any characters. The simple
1522 assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are described
1523 <a href="#smallassertions">above.</a>
1524 </P>
1525 <P>
1526 More complicated assertions are coded as subpatterns. There are two kinds:
1527 those that look ahead of the current position in the subject string, and those
1528 that look behind it. An assertion subpattern is matched in the normal way,
1529 except that it does not cause the current matching position to be changed.
1530 </P>
1531 <P>
1532 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1533 because it makes no sense to assert the same thing several times. If any kind
1534 of assertion contains capturing subpatterns within it, these are counted for
1535 the purposes of numbering the capturing subpatterns in the whole pattern.
1536 However, substring capturing is carried out only for positive assertions,
1537 because it does not make sense for negative assertions.
1538 </P>
1539 <br><b>
1540 Lookahead assertions
1541 </b><br>
1542 <P>
1543 Lookahead assertions start with (?= for positive assertions and (?! for
1544 negative assertions. For example,
1545 <pre>
1546 \w+(?=;)
1547 </pre>
1548 matches a word followed by a semicolon, but does not include the semicolon in
1549 the match, and
1550 <pre>
1551 foo(?!bar)
1552 </pre>
1553 matches any occurrence of "foo" that is not followed by "bar". Note that the
1554 apparently similar pattern
1555 <pre>
1556 (?!foo)bar
1557 </pre>
1558 does not find an occurrence of "bar" that is preceded by something other than
1559 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1560 (?!foo) is always true when the next three characters are "bar". A
1561 lookbehind assertion is needed to achieve the other effect.
1562 </P>
1563 <P>
1564 If you want to force a matching failure at some point in a pattern, the most
1565 convenient way to do it is with (?!) because an empty string always matches, so
1566 an assertion that requires there not to be an empty string must always fail.
1567 <a name="lookbehind"></a></P>
1568 <br><b>
1569 Lookbehind assertions
1570 </b><br>
1571 <P>
1572 Lookbehind assertions start with (?&#60;= for positive assertions and (?&#60;! for
1573 negative assertions. For example,
1574 <pre>
1575 (?&#60;!foo)bar
1576 </pre>
1577 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1578 a lookbehind assertion are restricted such that all the strings it matches must
1579 have a fixed length. However, if there are several top-level alternatives, they
1580 do not all have to have the same fixed length. Thus
1581 <pre>
1582 (?&#60;=bullock|donkey)
1583 </pre>
1584 is permitted, but
1585 <pre>
1586 (?&#60;!dogs?|cats?)
1587 </pre>
1588 causes an error at compile time. Branches that match different length strings
1589 are permitted only at the top level of a lookbehind assertion. This is an
1590 extension compared with Perl (at least for 5.8), which requires all branches to
1591 match the same length of string. An assertion such as
1592 <pre>
1593 (?&#60;=ab(c|de))
1594 </pre>
1595 is not permitted, because its single top-level branch can match two different
1596 lengths, but it is acceptable if rewritten to use two top-level branches:
1597 <pre>
1598 (?&#60;=abc|abde)
1599 </pre>
1600 In some cases, the Perl 5.10 escape sequence \K
1601 <a href="#resetmatchstart">(see above)</a>
1602 can be used instead of a lookbehind assertion; this is not restricted to a
1603 fixed-length.
1604 </P>
1605 <P>
1606 The implementation of lookbehind assertions is, for each alternative, to
1607 temporarily move the current position back by the fixed length and then try to
1608 match. If there are insufficient characters before the current position, the
1609 assertion fails.
1610 </P>
1611 <P>
1612 PCRE does not allow the \C escape (which matches a single byte in UTF-8 mode)
1613 to appear in lookbehind assertions, because it makes it impossible to calculate
1614 the length of the lookbehind. The \X and \R escapes, which can match
1615 different numbers of bytes, are also not permitted.
1616 </P>
1617 <P>
1618 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1619 specify efficient matching at the end of the subject string. Consider a simple
1620 pattern such as
1621 <pre>
1622 abcd$
1623 </pre>
1624 when applied to a long string that does not match. Because matching proceeds
1625 from left to right, PCRE will look for each "a" in the subject and then see if
1626 what follows matches the rest of the pattern. If the pattern is specified as
1627 <pre>
1628 ^.*abcd$
1629 </pre>
1630 the initial .* matches the entire string at first, but when this fails (because
1631 there is no following "a"), it backtracks to match all but the last character,
1632 then all but the last two characters, and so on. Once again the search for "a"
1633 covers the entire string, from right to left, so we are no better off. However,
1634 if the pattern is written as
1635 <pre>
1636 ^.*+(?&#60;=abcd)
1637 </pre>
1638 there can be no backtracking for the .*+ item; it can match only the entire
1639 string. The subsequent lookbehind assertion does a single test on the last four
1640 characters. If it fails, the match fails immediately. For long strings, this
1641 approach makes a significant difference to the processing time.
1642 </P>
1643 <br><b>
1644 Using multiple assertions
1645 </b><br>
1646 <P>
1647 Several assertions (of any sort) may occur in succession. For example,
1648 <pre>
1649 (?&#60;=\d{3})(?&#60;!999)foo
1650 </pre>
1651 matches "foo" preceded by three digits that are not "999". Notice that each of
1652 the assertions is applied independently at the same point in the subject
1653 string. First there is a check that the previous three characters are all
1654 digits, and then there is a check that the same three characters are not "999".
1655 This pattern does <i>not</i> match "foo" preceded by six characters, the first
1656 of which are digits and the last three of which are not "999". For example, it
1657 doesn't match "123abcfoo". A pattern to do that is
1658 <pre>
1659 (?&#60;=\d{3}...)(?&#60;!999)foo
1660 </pre>
1661 This time the first assertion looks at the preceding six characters, checking
1662 that the first three are digits, and then the second assertion checks that the
1663 preceding three characters are not "999".
1664 </P>
1665 <P>
1666 Assertions can be nested in any combination. For example,
1667 <pre>
1668 (?&#60;=(?&#60;!foo)bar)baz
1669 </pre>
1670 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1671 preceded by "foo", while
1672 <pre>
1673 (?&#60;=\d{3}(?!999)...)foo
1674 </pre>
1675 is another pattern that matches "foo" preceded by three digits and any three
1676 characters that are not "999".
1677 <a name="conditions"></a></P>
1678 <br><a name="SEC19" href="#TOC1">CONDITIONAL SUBPATTERNS</a><br>
1679 <P>
1680 It is possible to cause the matching process to obey a subpattern
1681 conditionally or to choose between two alternative subpatterns, depending on
1682 the result of an assertion, or whether a previous capturing subpattern matched
1683 or not. The two possible forms of conditional subpattern are
1684 <pre>
1685 (?(condition)yes-pattern)
1686 (?(condition)yes-pattern|no-pattern)
1687 </pre>
1688 If the condition is satisfied, the yes-pattern is used; otherwise the
1689 no-pattern (if present) is used. If there are more than two alternatives in the
1690 subpattern, a compile-time error occurs.
1691 </P>
1692 <P>
1693 There are four kinds of condition: references to subpatterns, references to
1694 recursion, a pseudo-condition called DEFINE, and assertions.
1695 </P>
1696 <br><b>
1697 Checking for a used subpattern by number
1698 </b><br>
1699 <P>
1700 If the text between the parentheses consists of a sequence of digits, the
1701 condition is true if the capturing subpattern of that number has previously
1702 matched. An alternative notation is to precede the digits with a plus or minus
1703 sign. In this case, the subpattern number is relative rather than absolute.
1704 The most recently opened parentheses can be referenced by (?(-1), the next most
1705 recent by (?(-2), and so on. In looping constructs it can also make sense to
1706 refer to subsequent groups with constructs such as (?(+2).
1707 </P>
1708 <P>
1709 Consider the following pattern, which contains non-significant white space to
1710 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
1711 three parts for ease of discussion:
1712 <pre>
1713 ( \( )? [^()]+ (?(1) \) )
1714 </pre>
1715 The first part matches an optional opening parenthesis, and if that
1716 character is present, sets it as the first captured substring. The second part
1717 matches one or more characters that are not parentheses. The third part is a
1718 conditional subpattern that tests whether the first set of parentheses matched
1719 or not. If they did, that is, if subject started with an opening parenthesis,
1720 the condition is true, and so the yes-pattern is executed and a closing
1721 parenthesis is required. Otherwise, since no-pattern is not present, the
1722 subpattern matches nothing. In other words, this pattern matches a sequence of
1723 non-parentheses, optionally enclosed in parentheses.
1724 </P>
1725 <P>
1726 If you were embedding this pattern in a larger one, you could use a relative
1727 reference:
1728 <pre>
1729 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
1730 </pre>
1731 This makes the fragment independent of the parentheses in the larger pattern.
1732 </P>
1733 <br><b>
1734 Checking for a used subpattern by name
1735 </b><br>
1736 <P>
1737 Perl uses the syntax (?(&#60;name&#62;)...) or (?('name')...) to test for a used
1738 subpattern by name. For compatibility with earlier versions of PCRE, which had
1739 this facility before Perl, the syntax (?(name)...) is also recognized. However,
1740 there is a possible ambiguity with this syntax, because subpattern names may
1741 consist entirely of digits. PCRE looks first for a named subpattern; if it
1742 cannot find one and the name consists entirely of digits, PCRE looks for a
1743 subpattern of that number, which must be greater than zero. Using subpattern
1744 names that consist entirely of digits is not recommended.
1745 </P>
1746 <P>
1747 Rewriting the above example to use a named subpattern gives this:
1748 <pre>
1749 (?&#60;OPEN&#62; \( )? [^()]+ (?(&#60;OPEN&#62;) \) )
1751 </PRE>
1752 </P>
1753 <br><b>
1754 Checking for pattern recursion
1755 </b><br>
1756 <P>
1757 If the condition is the string (R), and there is no subpattern with the name R,
1758 the condition is true if a recursive call to the whole pattern or any
1759 subpattern has been made. If digits or a name preceded by ampersand follow the
1760 letter R, for example:
1761 <pre>
1762 (?(R3)...) or (?(R&name)...)
1763 </pre>
1764 the condition is true if the most recent recursion is into the subpattern whose
1765 number or name is given. This condition does not check the entire recursion
1766 stack.
1767 </P>
1768 <P>
1769 At "top level", all these recursion test conditions are false. Recursive
1770 patterns are described below.
1771 </P>
1772 <br><b>
1773 Defining subpatterns for use by reference only
1774 </b><br>
1775 <P>
1776 If the condition is the string (DEFINE), and there is no subpattern with the
1777 name DEFINE, the condition is always false. In this case, there may be only one
1778 alternative in the subpattern. It is always skipped if control reaches this
1779 point in the pattern; the idea of DEFINE is that it can be used to define
1780 "subroutines" that can be referenced from elsewhere. (The use of "subroutines"
1781 is described below.) For example, a pattern to match an IPv4 address could be
1782 written like this (ignore whitespace and line breaks):
1783 <pre>
1784 (?(DEFINE) (?&#60;byte&#62; 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
1785 \b (?&byte) (\.(?&byte)){3} \b
1786 </pre>
1787 The first part of the pattern is a DEFINE group inside which a another group
1788 named "byte" is defined. This matches an individual component of an IPv4
1789 address (a number less than 256). When matching takes place, this part of the
1790 pattern is skipped because DEFINE acts like a false condition.
1791 </P>
1792 <P>
1793 The rest of the pattern uses references to the named group to match the four
1794 dot-separated components of an IPv4 address, insisting on a word boundary at
1795 each end.
1796 </P>
1797 <br><b>
1798 Assertion conditions
1799 </b><br>
1800 <P>
1801 If the condition is not in any of the above formats, it must be an assertion.
1802 This may be a positive or negative lookahead or lookbehind assertion. Consider
1803 this pattern, again containing non-significant white space, and with the two
1804 alternatives on the second line:
1805 <pre>
1806 (?(?=[^a-z]*[a-z])
1807 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
1808 </pre>
1809 The condition is a positive lookahead assertion that matches an optional
1810 sequence of non-letters followed by a letter. In other words, it tests for the
1811 presence of at least one letter in the subject. If a letter is found, the
1812 subject is matched against the first alternative; otherwise it is matched
1813 against the second. This pattern matches strings in one of the two forms
1814 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
1815 <a name="comments"></a></P>
1816 <br><a name="SEC20" href="#TOC1">COMMENTS</a><br>
1817 <P>
1818 The sequence (?# marks the start of a comment that continues up to the next
1819 closing parenthesis. Nested parentheses are not permitted. The characters
1820 that make up a comment play no part in the pattern matching at all.
1821 </P>
1822 <P>
1823 If the PCRE_EXTENDED option is set, an unescaped # character outside a
1824 character class introduces a comment that continues to immediately after the
1825 next newline in the pattern.
1826 <a name="recursion"></a></P>
1827 <br><a name="SEC21" href="#TOC1">RECURSIVE PATTERNS</a><br>
1828 <P>
1829 Consider the problem of matching a string in parentheses, allowing for
1830 unlimited nested parentheses. Without the use of recursion, the best that can
1831 be done is to use a pattern that matches up to some fixed depth of nesting. It
1832 is not possible to handle an arbitrary nesting depth.
1833 </P>
1834 <P>
1835 For some time, Perl has provided a facility that allows regular expressions to
1836 recurse (amongst other things). It does this by interpolating Perl code in the
1837 expression at run time, and the code can refer to the expression itself. A Perl
1838 pattern using code interpolation to solve the parentheses problem can be
1839 created like this:
1840 <pre>
1841 $re = qr{\( (?: (?&#62;[^()]+) | (?p{$re}) )* \)}x;
1842 </pre>
1843 The (?p{...}) item interpolates Perl code at run time, and in this case refers
1844 recursively to the pattern in which it appears.
1845 </P>
1846 <P>
1847 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
1848 supports special syntax for recursion of the entire pattern, and also for
1849 individual subpattern recursion. After its introduction in PCRE and Python,
1850 this kind of recursion was introduced into Perl at release 5.10.
1851 </P>
1852 <P>
1853 A special item that consists of (? followed by a number greater than zero and a
1854 closing parenthesis is a recursive call of the subpattern of the given number,
1855 provided that it occurs inside that subpattern. (If not, it is a "subroutine"
1856 call, which is described in the next section.) The special item (?R) or (?0) is
1857 a recursive call of the entire regular expression.
1858 </P>
1859 <P>
1860 In PCRE (like Python, but unlike Perl), a recursive subpattern call is always
1861 treated as an atomic group. That is, once it has matched some of the subject
1862 string, it is never re-entered, even if it contains untried alternatives and
1863 there is a subsequent matching failure.
1864 </P>
1865 <P>
1866 This PCRE pattern solves the nested parentheses problem (assume the
1867 PCRE_EXTENDED option is set so that white space is ignored):
1868 <pre>
1869 \( ( (?&#62;[^()]+) | (?R) )* \)
1870 </pre>
1871 First it matches an opening parenthesis. Then it matches any number of
1872 substrings which can either be a sequence of non-parentheses, or a recursive
1873 match of the pattern itself (that is, a correctly parenthesized substring).
1874 Finally there is a closing parenthesis.
1875 </P>
1876 <P>
1877 If this were part of a larger pattern, you would not want to recurse the entire
1878 pattern, so instead you could use this:
1879 <pre>
1880 ( \( ( (?&#62;[^()]+) | (?1) )* \) )
1881 </pre>
1882 We have put the pattern into parentheses, and caused the recursion to refer to
1883 them instead of the whole pattern.
1884 </P>
1885 <P>
1886 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
1887 is made easier by the use of relative references. (A Perl 5.10 feature.)
1888 Instead of (?1) in the pattern above you can write (?-2) to refer to the second
1889 most recently opened parentheses preceding the recursion. In other words, a
1890 negative number counts capturing parentheses leftwards from the point at which
1891 it is encountered.
1892 </P>
1893 <P>
1894 It is also possible to refer to subsequently opened parentheses, by writing
1895 references such as (?+2). However, these cannot be recursive because the
1896 reference is not inside the parentheses that are referenced. They are always
1897 "subroutine" calls, as described in the next section.
1898 </P>
1899 <P>
1900 An alternative approach is to use named parentheses instead. The Perl syntax
1901 for this is (?&name); PCRE's earlier syntax (?P&#62;name) is also supported. We
1902 could rewrite the above example as follows:
1903 <pre>
1904 (?&#60;pn&#62; \( ( (?&#62;[^()]+) | (?&pn) )* \) )
1905 </pre>
1906 If there is more than one subpattern with the same name, the earliest one is
1907 used.
1908 </P>
1909 <P>
1910 This particular example pattern that we have been looking at contains nested
1911 unlimited repeats, and so the use of atomic grouping for matching strings of
1912 non-parentheses is important when applying the pattern to strings that do not
1913 match. For example, when this pattern is applied to
1914 <pre>
1915 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
1916 </pre>
1917 it yields "no match" quickly. However, if atomic grouping is not used,
1918 the match runs for a very long time indeed because there are so many different
1919 ways the + and * repeats can carve up the subject, and all have to be tested
1920 before failure can be reported.
1921 </P>
1922 <P>
1923 At the end of a match, the values set for any capturing subpatterns are those
1924 from the outermost level of the recursion at which the subpattern value is set.
1925 If you want to obtain intermediate values, a callout function can be used (see
1926 below and the
1927 <a href="pcrecallout.html"><b>pcrecallout</b></a>
1928 documentation). If the pattern above is matched against
1929 <pre>
1930 (ab(cd)ef)
1931 </pre>
1932 the value for the capturing parentheses is "ef", which is the last value taken
1933 on at the top level. If additional parentheses are added, giving
1934 <pre>
1935 \( ( ( (?&#62;[^()]+) | (?R) )* ) \)
1936 ^ ^
1937 ^ ^
1938 </pre>
1939 the string they capture is "ab(cd)ef", the contents of the top level
1940 parentheses. If there are more than 15 capturing parentheses in a pattern, PCRE
1941 has to obtain extra memory to store data during a recursion, which it does by
1942 using <b>pcre_malloc</b>, freeing it via <b>pcre_free</b> afterwards. If no
1943 memory can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
1944 </P>
1945 <P>
1946 Do not confuse the (?R) item with the condition (R), which tests for recursion.
1947 Consider this pattern, which matches text in angle brackets, allowing for
1948 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
1949 recursing), whereas any characters are permitted at the outer level.
1950 <pre>
1951 &#60; (?: (?(R) \d++ | [^&#60;&#62;]*+) | (?R)) * &#62;
1952 </pre>
1953 In this pattern, (?(R) is the start of a conditional subpattern, with two
1954 different alternatives for the recursive and non-recursive cases. The (?R) item
1955 is the actual recursive call.
1956 <a name="subpatternsassubroutines"></a></P>
1957 <br><a name="SEC22" href="#TOC1">SUBPATTERNS AS SUBROUTINES</a><br>
1958 <P>
1959 If the syntax for a recursive subpattern reference (either by number or by
1960 name) is used outside the parentheses to which it refers, it operates like a
1961 subroutine in a programming language. The "called" subpattern may be defined
1962 before or after the reference. A numbered reference can be absolute or
1963 relative, as in these examples:
1964 <pre>
1965 (...(absolute)...)...(?2)...
1966 (...(relative)...)...(?-1)...
1967 (...(?+1)...(relative)...
1968 </pre>
1969 An earlier example pointed out that the pattern
1970 <pre>
1971 (sens|respons)e and \1ibility
1972 </pre>
1973 matches "sense and sensibility" and "response and responsibility", but not
1974 "sense and responsibility". If instead the pattern
1975 <pre>
1976 (sens|respons)e and (?1)ibility
1977 </pre>
1978 is used, it does match "sense and responsibility" as well as the other two
1979 strings. Another example is given in the discussion of DEFINE above.
1980 </P>
1981 <P>
1982 Like recursive subpatterns, a "subroutine" call is always treated as an atomic
1983 group. That is, once it has matched some of the subject string, it is never
1984 re-entered, even if it contains untried alternatives and there is a subsequent
1985 matching failure.
1986 </P>
1987 <P>
1988 When a subpattern is used as a subroutine, processing options such as
1989 case-independence are fixed when the subpattern is defined. They cannot be
1990 changed for different calls. For example, consider this pattern:
1991 <pre>
1992 (abc)(?i:(?-1))
1993 </pre>
1994 It matches "abcabc". It does not match "abcABC" because the change of
1995 processing option does not affect the called subpattern.
1996 </P>
1997 <br><a name="SEC23" href="#TOC1">CALLOUTS</a><br>
1998 <P>
1999 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2000 code to be obeyed in the middle of matching a regular expression. This makes it
2001 possible, amongst other things, to extract different substrings that match the
2002 same pair of parentheses when there is a repetition.
2003 </P>
2004 <P>
2005 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2006 code. The feature is called "callout". The caller of PCRE provides an external
2007 function by putting its entry point in the global variable <i>pcre_callout</i>.
2008 By default, this variable contains NULL, which disables all calling out.
2009 </P>
2010 <P>
2011 Within a regular expression, (?C) indicates the points at which the external
2012 function is to be called. If you want to identify different callout points, you
2013 can put a number less than 256 after the letter C. The default value is zero.
2014 For example, this pattern has two callout points:
2015 <pre>
2016 (?C1)abc(?C2)def
2017 </pre>
2018 If the PCRE_AUTO_CALLOUT flag is passed to <b>pcre_compile()</b>, callouts are
2019 automatically installed before each item in the pattern. They are all numbered
2020 255.
2021 </P>
2022 <P>
2023 During matching, when PCRE reaches a callout point (and <i>pcre_callout</i> is
2024 set), the external function is called. It is provided with the number of the
2025 callout, the position in the pattern, and, optionally, one item of data
2026 originally supplied by the caller of <b>pcre_exec()</b>. The callout function
2027 may cause matching to proceed, to backtrack, or to fail altogether. A complete
2028 description of the interface to the callout function is given in the
2029 <a href="pcrecallout.html"><b>pcrecallout</b></a>
2030 documentation.
2031 </P>
2032 <br><a name="SEC24" href="#TOC1">BACTRACKING CONTROL</a><br>
2033 <P>
2034 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2035 are described in the Perl documentation as "experimental and subject to change
2036 or removal in a future version of Perl". It goes on to say: "Their usage in
2037 production code should be noted to avoid problems during upgrades." The same
2038 remarks apply to the PCRE features described in this section.
2039 </P>
2040 <P>
2041 Since these verbs are specifically related to backtracking, they can be used
2042 only when the pattern is to be matched using <b>pcre_exec()</b>, which uses a
2043 backtracking algorithm. They cause an error if encountered by
2044 <b>pcre_dfa_exec()</b>.
2045 </P>
2046 <P>
2047 The new verbs make use of what was previously invalid syntax: an opening
2048 parenthesis followed by an asterisk. In Perl, they are generally of the form
2049 (*VERB:ARG) but PCRE does not support the use of arguments, so its general
2050 form is just (*VERB). Any number of these verbs may occur in a pattern. There
2051 are two kinds:
2052 </P>
2053 <br><b>
2054 Verbs that act immediately
2055 </b><br>
2056 <P>
2057 The following verbs act as soon as they are encountered:
2058 <pre>
2059 (*ACCEPT)
2060 </pre>
2061 This verb causes the match to end successfully, skipping the remainder of the
2062 pattern. When inside a recursion, only the innermost pattern is ended
2063 immediately. PCRE differs from Perl in what happens if the (*ACCEPT) is inside
2064 capturing parentheses. In Perl, the data so far is captured: in PCRE no data is
2065 captured. For example:
2066 <pre>
2067 A(A|B(*ACCEPT)|C)D
2068 </pre>
2069 This matches "AB", "AAD", or "ACD", but when it matches "AB", no data is
2070 captured.
2071 <pre>
2072 (*FAIL) or (*F)
2073 </pre>
2074 This verb causes the match to fail, forcing backtracking to occur. It is
2075 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2076 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2077 Perl features that are not present in PCRE. The nearest equivalent is the
2078 callout feature, as for example in this pattern:
2079 <pre>
2080 a+(?C)(*FAIL)
2081 </pre>
2082 A match with the string "aaaa" always fails, but the callout is taken before
2083 each backtrack happens (in this example, 10 times).
2084 </P>
2085 <br><b>
2086 Verbs that act after backtracking
2087 </b><br>
2088 <P>
2089 The following verbs do nothing when they are encountered. Matching continues
2090 with what follows, but if there is no subsequent match, a failure is forced.
2091 The verbs differ in exactly what kind of failure occurs.
2092 <pre>
2093 (*COMMIT)
2094 </pre>
2095 This verb causes the whole match to fail outright if the rest of the pattern
2096 does not match. Even if the pattern is unanchored, no further attempts to find
2097 a match by advancing the start point take place. Once (*COMMIT) has been
2098 passed, <b>pcre_exec()</b> is committed to finding a match at the current
2099 starting point, or not at all. For example:
2100 <pre>
2101 a+(*COMMIT)b
2102 </pre>
2103 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2104 dynamic anchor, or "I've started, so I must finish."
2105 <pre>
2106 (*PRUNE)
2107 </pre>
2108 This verb causes the match to fail at the current position if the rest of the
2109 pattern does not match. If the pattern is unanchored, the normal "bumpalong"
2110 advance to the next starting character then happens. Backtracking can occur as
2111 usual to the left of (*PRUNE), or when matching to the right of (*PRUNE), but
2112 if there is no match to the right, backtracking cannot cross (*PRUNE).
2113 In simple cases, the use of (*PRUNE) is just an alternative to an atomic
2114 group or possessive quantifier, but there are some uses of (*PRUNE) that cannot
2115 be expressed in any other way.
2116 <pre>
2117 (*SKIP)
2118 </pre>
2119 This verb is like (*PRUNE), except that if the pattern is unanchored, the
2120 "bumpalong" advance is not to the next character, but to the position in the
2121 subject where (*SKIP) was encountered. (*SKIP) signifies that whatever text
2122 was matched leading up to it cannot be part of a successful match. Consider:
2123 <pre>
2124 a+(*SKIP)b
2125 </pre>
2126 If the subject is "aaaac...", after the first match attempt fails (starting at
2127 the first character in the string), the starting point skips on to start the
2128 next attempt at "c". Note that a possessive quantifer does not have the same
2129 effect in this example; although it would suppress backtracking during the
2130 first match attempt, the second attempt would start at the second character
2131 instead of skipping on to "c".
2132 <pre>
2133 (*THEN)
2134 </pre>
2135 This verb causes a skip to the next alternation if the rest of the pattern does
2136 not match. That is, it cancels pending backtracking, but only within the
2137 current alternation. Its name comes from the observation that it can be used
2138 for a pattern-based if-then-else block:
2139 <pre>
2140 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2141 </pre>
2142 If the COND1 pattern matches, FOO is tried (and possibly further items after
2143 the end of the group if FOO succeeds); on failure the matcher skips to the
2144 second alternative and tries COND2, without backtracking into COND1. If (*THEN)
2145 is used outside of any alternation, it acts exactly like (*PRUNE).
2146 </P>
2147 <br><a name="SEC25" href="#TOC1">SEE ALSO</a><br>
2148 <P>
2149 <b>pcreapi</b>(3), <b>pcrecallout</b>(3), <b>pcrematching</b>(3), <b>pcre</b>(3).
2150 </P>
2151 <br><a name="SEC26" href="#TOC1">AUTHOR</a><br>
2152 <P>
2153 Philip Hazel
2154 <br>
2155 University Computing Service
2156 <br>
2157 Cambridge CB2 3QH, England.
2158 <br>
2159 </P>
2160 <br><a name="SEC27" href="#TOC1">REVISION</a><br>
2161 <P>
2162 Last updated: 21 August 2007
2163 <br>
2164 Copyright &copy; 1997-2007 University of Cambridge.
2165 <br>
2166 <p>
2167 Return to the <a href="index.html">PCRE index page</a>.
2168 </p>


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

  ViewVC Help
Powered by ViewVC 1.1.5