ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 1194 - (show annotations)
Wed Oct 31 17:42:29 2012 UTC (6 years, 9 months ago) by ph10
File MIME type: text/html
File size: 127586 byte(s)
More documentation updates
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">EBCDIC CHARACTER CODES</a>
18 <li><a name="TOC3" href="#SEC3">NEWLINE CONVENTIONS</a>
19 <li><a name="TOC4" href="#SEC4">CHARACTERS AND METACHARACTERS</a>
20 <li><a name="TOC5" href="#SEC5">BACKSLASH</a>
21 <li><a name="TOC6" href="#SEC6">CIRCUMFLEX AND DOLLAR</a>
22 <li><a name="TOC7" href="#SEC7">FULL STOP (PERIOD, DOT) AND \N</a>
23 <li><a name="TOC8" href="#SEC8">MATCHING A SINGLE DATA UNIT</a>
25 <li><a name="TOC10" href="#SEC10">POSIX CHARACTER CLASSES</a>
26 <li><a name="TOC11" href="#SEC11">VERTICAL BAR</a>
27 <li><a name="TOC12" href="#SEC12">INTERNAL OPTION SETTING</a>
28 <li><a name="TOC13" href="#SEC13">SUBPATTERNS</a>
29 <li><a name="TOC14" href="#SEC14">DUPLICATE SUBPATTERN NUMBERS</a>
30 <li><a name="TOC15" href="#SEC15">NAMED SUBPATTERNS</a>
31 <li><a name="TOC16" href="#SEC16">REPETITION</a>
33 <li><a name="TOC18" href="#SEC18">BACK REFERENCES</a>
34 <li><a name="TOC19" href="#SEC19">ASSERTIONS</a>
35 <li><a name="TOC20" href="#SEC20">CONDITIONAL SUBPATTERNS</a>
36 <li><a name="TOC21" href="#SEC21">COMMENTS</a>
37 <li><a name="TOC22" href="#SEC22">RECURSIVE PATTERNS</a>
38 <li><a name="TOC23" href="#SEC23">SUBPATTERNS AS SUBROUTINES</a>
39 <li><a name="TOC24" href="#SEC24">ONIGURUMA SUBROUTINE SYNTAX</a>
40 <li><a name="TOC25" href="#SEC25">CALLOUTS</a>
41 <li><a name="TOC26" href="#SEC26">BACKTRACKING CONTROL</a>
42 <li><a name="TOC27" href="#SEC27">SEE ALSO</a>
43 <li><a name="TOC28" href="#SEC28">AUTHOR</a>
44 <li><a name="TOC29" href="#SEC29">REVISION</a>
45 </ul>
46 <br><a name="SEC1" href="#TOC1">PCRE REGULAR EXPRESSION DETAILS</a><br>
47 <P>
48 The syntax and semantics of the regular expressions that are supported by PCRE
49 are described in detail below. There is a quick-reference syntax summary in the
50 <a href="pcresyntax.html"><b>pcresyntax</b></a>
51 page. PCRE tries to match Perl syntax and semantics as closely as it can. PCRE
52 also supports some alternative regular expression syntax (which does not
53 conflict with the Perl syntax) in order to provide some compatibility with
54 regular expressions in Python, .NET, and Oniguruma.
55 </P>
56 <P>
57 Perl's regular expressions are described in its own documentation, and
58 regular expressions in general are covered in a number of books, some of which
59 have copious examples. Jeffrey Friedl's "Mastering Regular Expressions",
60 published by O'Reilly, covers regular expressions in great detail. This
61 description of PCRE's regular expressions is intended as reference material.
62 </P>
63 <P>
64 The original operation of PCRE was on strings of one-byte characters. However,
65 there is now also support for UTF-8 strings in the original library, an
66 extra library that supports 16-bit and UTF-16 character strings, and an
67 extra library that supports 32-bit and UTF-32 character strings. To use these
68 features, PCRE must be built to include appropriate support. When using UTF
69 strings you must either call the compiling function with the PCRE_UTF8,
70 PCRE_UTF16 or PCRE_UTF32 option, or the pattern must start with one of
71 these special sequences:
72 <pre>
73 (*UTF8)
74 (*UTF16)
75 (*UTF32)
76 </pre>
77 Starting a pattern with such a sequence is equivalent to setting the relevant
78 option. This feature is not Perl-compatible. How setting a UTF mode affects
79 pattern matching is mentioned in several places below. There is also a summary
80 of features in the
81 <a href="pcreunicode.html"><b>pcreunicode</b></a>
82 page.
83 </P>
84 <P>
85 Another special sequence that may appear at the start of a pattern or in
86 combination with (*UTF8) or (*UTF16) or (*UTF32) is:
87 <pre>
88 (*UCP)
89 </pre>
90 This has the same effect as setting the PCRE_UCP option: it causes sequences
91 such as \d and \w to use Unicode properties to determine character types,
92 instead of recognizing only characters with codes less than 128 via a lookup
93 table.
94 </P>
95 <P>
96 If a pattern starts with (*NO_START_OPT), it has the same effect as setting the
97 PCRE_NO_START_OPTIMIZE option either at compile or matching time. There are
98 also some more of these special sequences that are concerned with the handling
99 of newlines; they are described below.
100 </P>
101 <P>
102 The remainder of this document discusses the patterns that are supported by
103 PCRE when one its main matching functions, <b>pcre_exec()</b> (8-bit) or
104 <b>pcre[16|32]_exec()</b> (16- or 32-bit), is used. PCRE also has alternative
105 matching functions, <b>pcre_dfa_exec()</b> and <b>pcre[16|32_dfa_exec()</b>,
106 which match using a different algorithm that is not Perl-compatible. Some of
107 the features discussed below are not available when DFA matching is used. The
108 advantages and disadvantages of the alternative functions, and how they differ
109 from the normal functions, are discussed in the
110 <a href="pcrematching.html"><b>pcrematching</b></a>
111 page.
112 </P>
113 <br><a name="SEC2" href="#TOC1">EBCDIC CHARACTER CODES</a><br>
114 <P>
115 PCRE can be compiled to run in an environment that uses EBCDIC as its character
116 code rather than ASCII or Unicode (typically a mainframe system). In the
117 sections below, character code values are ASCII or Unicode; in an EBCDIC
118 environment these characters may have different code values, and there are no
119 code points greater than 255.
120 <a name="newlines"></a></P>
121 <br><a name="SEC3" href="#TOC1">NEWLINE CONVENTIONS</a><br>
122 <P>
123 PCRE supports five different conventions for indicating line breaks in
124 strings: a single CR (carriage return) character, a single LF (linefeed)
125 character, the two-character sequence CRLF, any of the three preceding, or any
126 Unicode newline sequence. The
127 <a href="pcreapi.html"><b>pcreapi</b></a>
128 page has
129 <a href="pcreapi.html#newlines">further discussion</a>
130 about newlines, and shows how to set the newline convention in the
131 <i>options</i> arguments for the compiling and matching functions.
132 </P>
133 <P>
134 It is also possible to specify a newline convention by starting a pattern
135 string with one of the following five sequences:
136 <pre>
137 (*CR) carriage return
138 (*LF) linefeed
139 (*CRLF) carriage return, followed by linefeed
140 (*ANYCRLF) any of the three above
141 (*ANY) all Unicode newline sequences
142 </pre>
143 These override the default and the options given to the compiling function. For
144 example, on a Unix system where LF is the default newline sequence, the pattern
145 <pre>
146 (*CR)a.b
147 </pre>
148 changes the convention to CR. That pattern matches "a\nb" because LF is no
149 longer a newline. Note that these special settings, which are not
150 Perl-compatible, are recognized only at the very start of a pattern, and that
151 they must be in upper case. If more than one of them is present, the last one
152 is used.
153 </P>
154 <P>
155 The newline convention affects the interpretation of the dot metacharacter when
156 PCRE_DOTALL is not set, and also the behaviour of \N. However, it does not
157 affect what the \R escape sequence matches. By default, this is any Unicode
158 newline sequence, for Perl compatibility. However, this can be changed; see the
159 description of \R in the section entitled
160 <a href="#newlineseq">"Newline sequences"</a>
161 below. A change of \R setting can be combined with a change of newline
162 convention.
163 </P>
164 <br><a name="SEC4" href="#TOC1">CHARACTERS AND METACHARACTERS</a><br>
165 <P>
166 A regular expression is a pattern that is matched against a subject string from
167 left to right. Most characters stand for themselves in a pattern, and match the
168 corresponding characters in the subject. As a trivial example, the pattern
169 <pre>
170 The quick brown fox
171 </pre>
172 matches a portion of a subject string that is identical to itself. When
173 caseless matching is specified (the PCRE_CASELESS option), letters are matched
174 independently of case. In a UTF mode, PCRE always understands the concept of
175 case for characters whose values are less than 128, so caseless matching is
176 always possible. For characters with higher values, the concept of case is
177 supported if PCRE is compiled with Unicode property support, but not otherwise.
178 If you want to use caseless matching for characters 128 and above, you must
179 ensure that PCRE is compiled with Unicode property support as well as with
180 UTF support.
181 </P>
182 <P>
183 The power of regular expressions comes from the ability to include alternatives
184 and repetitions in the pattern. These are encoded in the pattern by the use of
185 <i>metacharacters</i>, which do not stand for themselves but instead are
186 interpreted in some special way.
187 </P>
188 <P>
189 There are two different sets of metacharacters: those that are recognized
190 anywhere in the pattern except within square brackets, and those that are
191 recognized within square brackets. Outside square brackets, the metacharacters
192 are as follows:
193 <pre>
194 \ general escape character with several uses
195 ^ assert start of string (or line, in multiline mode)
196 $ assert end of string (or line, in multiline mode)
197 . match any character except newline (by default)
198 [ start character class definition
199 | start of alternative branch
200 ( start subpattern
201 ) end subpattern
202 ? extends the meaning of (
203 also 0 or 1 quantifier
204 also quantifier minimizer
205 * 0 or more quantifier
206 + 1 or more quantifier
207 also "possessive quantifier"
208 { start min/max quantifier
209 </pre>
210 Part of a pattern that is in square brackets is called a "character class". In
211 a character class the only metacharacters are:
212 <pre>
213 \ general escape character
214 ^ negate the class, but only if the first character
215 - indicates character range
216 [ POSIX character class (only if followed by POSIX syntax)
217 ] terminates the character class
218 </pre>
219 The following sections describe the use of each of the metacharacters.
220 </P>
221 <br><a name="SEC5" href="#TOC1">BACKSLASH</a><br>
222 <P>
223 The backslash character has several uses. Firstly, if it is followed by a
224 character that is not a number or a letter, it takes away any special meaning
225 that character may have. This use of backslash as an escape character applies
226 both inside and outside character classes.
227 </P>
228 <P>
229 For example, if you want to match a * character, you write \* in the pattern.
230 This escaping action applies whether or not the following character would
231 otherwise be interpreted as a metacharacter, so it is always safe to precede a
232 non-alphanumeric with backslash to specify that it stands for itself. In
233 particular, if you want to match a backslash, you write \\.
234 </P>
235 <P>
236 In a UTF mode, only ASCII numbers and letters have any special meaning after a
237 backslash. All other characters (in particular, those whose codepoints are
238 greater than 127) are treated as literals.
239 </P>
240 <P>
241 If a pattern is compiled with the PCRE_EXTENDED option, white space in the
242 pattern (other than in a character class) and characters between a # outside
243 a character class and the next newline are ignored. An escaping backslash can
244 be used to include a white space or # character as part of the pattern.
245 </P>
246 <P>
247 If you want to remove the special meaning from a sequence of characters, you
248 can do so by putting them between \Q and \E. This is different from Perl in
249 that $ and @ are handled as literals in \Q...\E sequences in PCRE, whereas in
250 Perl, $ and @ cause variable interpolation. Note the following examples:
251 <pre>
252 Pattern PCRE matches Perl matches
254 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
255 \Qabc\$xyz\E abc\$xyz abc\$xyz
256 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
257 </pre>
258 The \Q...\E sequence is recognized both inside and outside character classes.
259 An isolated \E that is not preceded by \Q is ignored. If \Q is not followed
260 by \E later in the pattern, the literal interpretation continues to the end of
261 the pattern (that is, \E is assumed at the end). If the isolated \Q is inside
262 a character class, this causes an error, because the character class is not
263 terminated.
264 <a name="digitsafterbackslash"></a></P>
265 <br><b>
266 Non-printing characters
267 </b><br>
268 <P>
269 A second use of backslash provides a way of encoding non-printing characters
270 in patterns in a visible manner. There is no restriction on the appearance of
271 non-printing characters, apart from the binary zero that terminates a pattern,
272 but when a pattern is being prepared by text editing, it is often easier to use
273 one of the following escape sequences than the binary character it represents:
274 <pre>
275 \a alarm, that is, the BEL character (hex 07)
276 \cx "control-x", where x is any ASCII character
277 \e escape (hex 1B)
278 \f form feed (hex 0C)
279 \n linefeed (hex 0A)
280 \r carriage return (hex 0D)
281 \t tab (hex 09)
282 \ddd character with octal code ddd, or back reference
283 \xhh character with hex code hh
284 \x{hhh..} character with hex code hhh.. (non-JavaScript mode)
285 \uhhhh character with hex code hhhh (JavaScript mode only)
286 </pre>
287 The precise effect of \cx on ASCII characters is as follows: if x is a lower
288 case letter, it is converted to upper case. Then bit 6 of the character (hex
289 40) is inverted. Thus \cA to \cZ become hex 01 to hex 1A (A is 41, Z is 5A),
290 but \c{ becomes hex 3B ({ is 7B), and \c; becomes hex 7B (; is 3B). If the
291 data item (byte or 16-bit value) following \c has a value greater than 127, a
292 compile-time error occurs. This locks out non-ASCII characters in all modes.
293 </P>
294 <P>
295 The \c facility was designed for use with ASCII characters, but with the
296 extension to Unicode it is even less useful than it once was. It is, however,
297 recognized when PCRE is compiled in EBCDIC mode, where data items are always
298 bytes. In this mode, all values are valid after \c. If the next character is a
299 lower case letter, it is converted to upper case. Then the 0xc0 bits of the
300 byte are inverted. Thus \cA becomes hex 01, as in ASCII (A is C1), but because
301 the EBCDIC letters are disjoint, \cZ becomes hex 29 (Z is E9), and other
302 characters also generate different values.
303 </P>
304 <P>
305 By default, after \x, from zero to two hexadecimal digits are read (letters
306 can be in upper or lower case). Any number of hexadecimal digits may appear
307 between \x{ and }, but the character code is constrained as follows:
308 <pre>
309 8-bit non-UTF mode less than 0x100
310 8-bit UTF-8 mode less than 0x10ffff and a valid codepoint
311 16-bit non-UTF mode less than 0x10000
312 16-bit UTF-16 mode less than 0x10ffff and a valid codepoint
313 32-bit non-UTF mode less than 0x80000000
314 32-bit UTF-32 mode less than 0x10ffff and a valid codepoint
315 </pre>
316 Invalid Unicode codepoints are the range 0xd800 to 0xdfff (the so-called
317 "surrogate" codepoints), and 0xffef.
318 </P>
319 <P>
320 If characters other than hexadecimal digits appear between \x{ and }, or if
321 there is no terminating }, this form of escape is not recognized. Instead, the
322 initial \x will be interpreted as a basic hexadecimal escape, with no
323 following digits, giving a character whose value is zero.
324 </P>
325 <P>
326 If the PCRE_JAVASCRIPT_COMPAT option is set, the interpretation of \x is
327 as just described only when it is followed by two hexadecimal digits.
328 Otherwise, it matches a literal "x" character. In JavaScript mode, support for
329 code points greater than 256 is provided by \u, which must be followed by
330 four hexadecimal digits; otherwise it matches a literal "u" character.
331 Character codes specified by \u in JavaScript mode are constrained in the same
332 was as those specified by \x in non-JavaScript mode.
333 </P>
334 <P>
335 Characters whose value is less than 256 can be defined by either of the two
336 syntaxes for \x (or by \u in JavaScript mode). There is no difference in the
337 way they are handled. For example, \xdc is exactly the same as \x{dc} (or
338 \u00dc in JavaScript mode).
339 </P>
340 <P>
341 After \0 up to two further octal digits are read. If there are fewer than two
342 digits, just those that are present are used. Thus the sequence \0\x\07
343 specifies two binary zeros followed by a BEL character (code value 7). Make
344 sure you supply two digits after the initial zero if the pattern character that
345 follows is itself an octal digit.
346 </P>
347 <P>
348 The handling of a backslash followed by a digit other than 0 is complicated.
349 Outside a character class, PCRE reads it and any following digits as a decimal
350 number. If the number is less than 10, or if there have been at least that many
351 previous capturing left parentheses in the expression, the entire sequence is
352 taken as a <i>back reference</i>. A description of how this works is given
353 <a href="#backreferences">later,</a>
354 following the discussion of
355 <a href="#subpattern">parenthesized subpatterns.</a>
356 </P>
357 <P>
358 Inside a character class, or if the decimal number is greater than 9 and there
359 have not been that many capturing subpatterns, PCRE re-reads up to three octal
360 digits following the backslash, and uses them to generate a data character. Any
361 subsequent digits stand for themselves. The value of the character is
362 constrained in the same way as characters specified in hexadecimal.
363 For example:
364 <pre>
365 \040 is another way of writing an ASCII space
366 \40 is the same, provided there are fewer than 40 previous capturing subpatterns
367 \7 is always a back reference
368 \11 might be a back reference, or another way of writing a tab
369 \011 is always a tab
370 \0113 is a tab followed by the character "3"
371 \113 might be a back reference, otherwise the character with octal code 113
372 \377 might be a back reference, otherwise the value 255 (decimal)
373 \81 is either a back reference, or a binary zero followed by the two characters "8" and "1"
374 </pre>
375 Note that octal values of 100 or greater must not be introduced by a leading
376 zero, because no more than three octal digits are ever read.
377 </P>
378 <P>
379 All the sequences that define a single character value can be used both inside
380 and outside character classes. In addition, inside a character class, \b is
381 interpreted as the backspace character (hex 08).
382 </P>
383 <P>
384 \N is not allowed in a character class. \B, \R, and \X are not special
385 inside a character class. Like other unrecognized escape sequences, they are
386 treated as the literal characters "B", "R", and "X" by default, but cause an
387 error if the PCRE_EXTRA option is set. Outside a character class, these
388 sequences have different meanings.
389 </P>
390 <br><b>
391 Unsupported escape sequences
392 </b><br>
393 <P>
394 In Perl, the sequences \l, \L, \u, and \U are recognized by its string
395 handler and used to modify the case of following characters. By default, PCRE
396 does not support these escape sequences. However, if the PCRE_JAVASCRIPT_COMPAT
397 option is set, \U matches a "U" character, and \u can be used to define a
398 character by code point, as described in the previous section.
399 </P>
400 <br><b>
401 Absolute and relative back references
402 </b><br>
403 <P>
404 The sequence \g followed by an unsigned or a negative number, optionally
405 enclosed in braces, is an absolute or relative back reference. A named back
406 reference can be coded as \g{name}. Back references are discussed
407 <a href="#backreferences">later,</a>
408 following the discussion of
409 <a href="#subpattern">parenthesized subpatterns.</a>
410 </P>
411 <br><b>
412 Absolute and relative subroutine calls
413 </b><br>
414 <P>
415 For compatibility with Oniguruma, the non-Perl syntax \g followed by a name or
416 a number enclosed either in angle brackets or single quotes, is an alternative
417 syntax for referencing a subpattern as a "subroutine". Details are discussed
418 <a href="#onigurumasubroutines">later.</a>
419 Note that \g{...} (Perl syntax) and \g&#60;...&#62; (Oniguruma syntax) are <i>not</i>
420 synonymous. The former is a back reference; the latter is a
421 <a href="#subpatternsassubroutines">subroutine</a>
422 call.
423 <a name="genericchartypes"></a></P>
424 <br><b>
425 Generic character types
426 </b><br>
427 <P>
428 Another use of backslash is for specifying generic character types:
429 <pre>
430 \d any decimal digit
431 \D any character that is not a decimal digit
432 \h any horizontal white space character
433 \H any character that is not a horizontal white space character
434 \s any white space character
435 \S any character that is not a white space character
436 \v any vertical white space character
437 \V any character that is not a vertical white space character
438 \w any "word" character
439 \W any "non-word" character
440 </pre>
441 There is also the single sequence \N, which matches a non-newline character.
442 This is the same as
443 <a href="#fullstopdot">the "." metacharacter</a>
444 when PCRE_DOTALL is not set. Perl also uses \N to match characters by name;
445 PCRE does not support this.
446 </P>
447 <P>
448 Each pair of lower and upper case escape sequences partitions the complete set
449 of characters into two disjoint sets. Any given character matches one, and only
450 one, of each pair. The sequences can appear both inside and outside character
451 classes. They each match one character of the appropriate type. If the current
452 matching point is at the end of the subject string, all of them fail, because
453 there is no character to match.
454 </P>
455 <P>
456 For compatibility with Perl, \s does not match the VT character (code 11).
457 This makes it different from the the POSIX "space" class. The \s characters
458 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
459 included in a Perl script, \s may match the VT character. In PCRE, it never
460 does.
461 </P>
462 <P>
463 A "word" character is an underscore or any character that is a letter or digit.
464 By default, the definition of letters and digits is controlled by PCRE's
465 low-valued character tables, and may vary if locale-specific matching is taking
466 place (see
467 <a href="pcreapi.html#localesupport">"Locale support"</a>
468 in the
469 <a href="pcreapi.html"><b>pcreapi</b></a>
470 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
471 or "french" in Windows, some character codes greater than 128 are used for
472 accented letters, and these are then matched by \w. The use of locales with
473 Unicode is discouraged.
474 </P>
475 <P>
476 By default, in a UTF mode, characters with values greater than 128 never match
477 \d, \s, or \w, and always match \D, \S, and \W. These sequences retain
478 their original meanings from before UTF support was available, mainly for
479 efficiency reasons. However, if PCRE is compiled with Unicode property support,
480 and the PCRE_UCP option is set, the behaviour is changed so that Unicode
481 properties are used to determine character types, as follows:
482 <pre>
483 \d any character that \p{Nd} matches (decimal digit)
484 \s any character that \p{Z} matches, plus HT, LF, FF, CR
485 \w any character that \p{L} or \p{N} matches, plus underscore
486 </pre>
487 The upper case escapes match the inverse sets of characters. Note that \d
488 matches only decimal digits, whereas \w matches any Unicode digit, as well as
489 any Unicode letter, and underscore. Note also that PCRE_UCP affects \b, and
490 \B because they are defined in terms of \w and \W. Matching these sequences
491 is noticeably slower when PCRE_UCP is set.
492 </P>
493 <P>
494 The sequences \h, \H, \v, and \V are features that were added to Perl at
495 release 5.10. In contrast to the other sequences, which match only ASCII
496 characters by default, these always match certain high-valued codepoints,
497 whether or not PCRE_UCP is set. The horizontal space characters are:
498 <pre>
499 U+0009 Horizontal tab (HT)
500 U+0020 Space
501 U+00A0 Non-break space
502 U+1680 Ogham space mark
503 U+180E Mongolian vowel separator
504 U+2000 En quad
505 U+2001 Em quad
506 U+2002 En space
507 U+2003 Em space
508 U+2004 Three-per-em space
509 U+2005 Four-per-em space
510 U+2006 Six-per-em space
511 U+2007 Figure space
512 U+2008 Punctuation space
513 U+2009 Thin space
514 U+200A Hair space
515 U+202F Narrow no-break space
516 U+205F Medium mathematical space
517 U+3000 Ideographic space
518 </pre>
519 The vertical space characters are:
520 <pre>
521 U+000A Linefeed (LF)
522 U+000B Vertical tab (VT)
523 U+000C Form feed (FF)
524 U+000D Carriage return (CR)
525 U+0085 Next line (NEL)
526 U+2028 Line separator
527 U+2029 Paragraph separator
528 </pre>
529 In 8-bit, non-UTF-8 mode, only the characters with codepoints less than 256 are
530 relevant.
531 <a name="newlineseq"></a></P>
532 <br><b>
533 Newline sequences
534 </b><br>
535 <P>
536 Outside a character class, by default, the escape sequence \R matches any
537 Unicode newline sequence. In 8-bit non-UTF-8 mode \R is equivalent to the
538 following:
539 <pre>
540 (?&#62;\r\n|\n|\x0b|\f|\r|\x85)
541 </pre>
542 This is an example of an "atomic group", details of which are given
543 <a href="#atomicgroup">below.</a>
544 This particular group matches either the two-character sequence CR followed by
545 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
546 U+000B), FF (form feed, U+000C), CR (carriage return, U+000D), or NEL (next
547 line, U+0085). The two-character sequence is treated as a single unit that
548 cannot be split.
549 </P>
550 <P>
551 In other modes, two additional characters whose codepoints are greater than 255
552 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
553 Unicode character property support is not needed for these characters to be
554 recognized.
555 </P>
556 <P>
557 It is possible to restrict \R to match only CR, LF, or CRLF (instead of the
558 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
559 either at compile time or when the pattern is matched. (BSR is an abbrevation
560 for "backslash R".) This can be made the default when PCRE is built; if this is
561 the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
562 It is also possible to specify these settings by starting a pattern string with
563 one of the following sequences:
564 <pre>
565 (*BSR_ANYCRLF) CR, LF, or CRLF only
566 (*BSR_UNICODE) any Unicode newline sequence
567 </pre>
568 These override the default and the options given to the compiling function, but
569 they can themselves be overridden by options given to a matching function. Note
570 that these special settings, which are not Perl-compatible, are recognized only
571 at the very start of a pattern, and that they must be in upper case. If more
572 than one of them is present, the last one is used. They can be combined with a
573 change of newline convention; for example, a pattern can start with:
574 <pre>
576 </pre>
577 They can also be combined with the (*UTF8), (*UTF16), (*UTF32) or (*UCP) special
578 sequences. Inside a character class, \R is treated as an unrecognized escape
579 sequence, and so matches the letter "R" by default, but causes an error if
580 PCRE_EXTRA is set.
581 <a name="uniextseq"></a></P>
582 <br><b>
583 Unicode character properties
584 </b><br>
585 <P>
586 When PCRE is built with Unicode character property support, three additional
587 escape sequences that match characters with specific properties are available.
588 When in 8-bit non-UTF-8 mode, these sequences are of course limited to testing
589 characters whose codepoints are less than 256, but they do work in this mode.
590 The extra escape sequences are:
591 <pre>
592 \p{<i>xx</i>} a character with the <i>xx</i> property
593 \P{<i>xx</i>} a character without the <i>xx</i> property
594 \X a Unicode extended grapheme cluster
595 </pre>
596 The property names represented by <i>xx</i> above are limited to the Unicode
597 script names, the general category properties, "Any", which matches any
598 character (including newline), and some special PCRE properties (described
599 in the
600 <a href="#extraprops">next section).</a>
601 Other Perl properties such as "InMusicalSymbols" are not currently supported by
602 PCRE. Note that \P{Any} does not match any characters, so always causes a
603 match failure.
604 </P>
605 <P>
606 Sets of Unicode characters are defined as belonging to certain scripts. A
607 character from one of these sets can be matched using a script name. For
608 example:
609 <pre>
610 \p{Greek}
611 \P{Han}
612 </pre>
613 Those that are not part of an identified script are lumped together as
614 "Common". The current list of scripts is:
615 </P>
616 <P>
617 Arabic,
618 Armenian,
619 Avestan,
620 Balinese,
621 Bamum,
622 Batak,
623 Bengali,
624 Bopomofo,
625 Brahmi,
626 Braille,
627 Buginese,
628 Buhid,
629 Canadian_Aboriginal,
630 Carian,
631 Chakma,
632 Cham,
633 Cherokee,
634 Common,
635 Coptic,
636 Cuneiform,
637 Cypriot,
638 Cyrillic,
639 Deseret,
640 Devanagari,
641 Egyptian_Hieroglyphs,
642 Ethiopic,
643 Georgian,
644 Glagolitic,
645 Gothic,
646 Greek,
647 Gujarati,
648 Gurmukhi,
649 Han,
650 Hangul,
651 Hanunoo,
652 Hebrew,
653 Hiragana,
654 Imperial_Aramaic,
655 Inherited,
656 Inscriptional_Pahlavi,
657 Inscriptional_Parthian,
658 Javanese,
659 Kaithi,
660 Kannada,
661 Katakana,
662 Kayah_Li,
663 Kharoshthi,
664 Khmer,
665 Lao,
666 Latin,
667 Lepcha,
668 Limbu,
669 Linear_B,
670 Lisu,
671 Lycian,
672 Lydian,
673 Malayalam,
674 Mandaic,
675 Meetei_Mayek,
676 Meroitic_Cursive,
677 Meroitic_Hieroglyphs,
678 Miao,
679 Mongolian,
680 Myanmar,
681 New_Tai_Lue,
682 Nko,
683 Ogham,
684 Old_Italic,
685 Old_Persian,
686 Old_South_Arabian,
687 Old_Turkic,
688 Ol_Chiki,
689 Oriya,
690 Osmanya,
691 Phags_Pa,
692 Phoenician,
693 Rejang,
694 Runic,
695 Samaritan,
696 Saurashtra,
697 Sharada,
698 Shavian,
699 Sinhala,
700 Sora_Sompeng,
701 Sundanese,
702 Syloti_Nagri,
703 Syriac,
704 Tagalog,
705 Tagbanwa,
706 Tai_Le,
707 Tai_Tham,
708 Tai_Viet,
709 Takri,
710 Tamil,
711 Telugu,
712 Thaana,
713 Thai,
714 Tibetan,
715 Tifinagh,
716 Ugaritic,
717 Vai,
718 Yi.
719 </P>
720 <P>
721 Each character has exactly one Unicode general category property, specified by
722 a two-letter abbreviation. For compatibility with Perl, negation can be
723 specified by including a circumflex between the opening brace and the property
724 name. For example, \p{^Lu} is the same as \P{Lu}.
725 </P>
726 <P>
727 If only one letter is specified with \p or \P, it includes all the general
728 category properties that start with that letter. In this case, in the absence
729 of negation, the curly brackets in the escape sequence are optional; these two
730 examples have the same effect:
731 <pre>
732 \p{L}
733 \pL
734 </pre>
735 The following general category property codes are supported:
736 <pre>
737 C Other
738 Cc Control
739 Cf Format
740 Cn Unassigned
741 Co Private use
742 Cs Surrogate
744 L Letter
745 Ll Lower case letter
746 Lm Modifier letter
747 Lo Other letter
748 Lt Title case letter
749 Lu Upper case letter
751 M Mark
752 Mc Spacing mark
753 Me Enclosing mark
754 Mn Non-spacing mark
756 N Number
757 Nd Decimal number
758 Nl Letter number
759 No Other number
761 P Punctuation
762 Pc Connector punctuation
763 Pd Dash punctuation
764 Pe Close punctuation
765 Pf Final punctuation
766 Pi Initial punctuation
767 Po Other punctuation
768 Ps Open punctuation
770 S Symbol
771 Sc Currency symbol
772 Sk Modifier symbol
773 Sm Mathematical symbol
774 So Other symbol
776 Z Separator
777 Zl Line separator
778 Zp Paragraph separator
779 Zs Space separator
780 </pre>
781 The special property L& is also supported: it matches a character that has
782 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
783 a modifier or "other".
784 </P>
785 <P>
786 The Cs (Surrogate) property applies only to characters in the range U+D800 to
787 U+DFFF. Such characters are not valid in Unicode strings and so
788 cannot be tested by PCRE, unless UTF validity checking has been turned off
789 (see the discussion of PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK and
790 PCRE_NO_UTF32_CHECK in the
791 <a href="pcreapi.html"><b>pcreapi</b></a>
792 page). Perl does not support the Cs property.
793 </P>
794 <P>
795 The long synonyms for property names that Perl supports (such as \p{Letter})
796 are not supported by PCRE, nor is it permitted to prefix any of these
797 properties with "Is".
798 </P>
799 <P>
800 No character that is in the Unicode table has the Cn (unassigned) property.
801 Instead, this property is assumed for any code point that is not in the
802 Unicode table.
803 </P>
804 <P>
805 Specifying caseless matching does not affect these escape sequences. For
806 example, \p{Lu} always matches only upper case letters.
807 </P>
808 <P>
809 Matching characters by Unicode property is not fast, because PCRE has to do a
810 multistage table lookup in order to find a character's property. That is why
811 the traditional escape sequences such as \d and \w do not use Unicode
812 properties in PCRE by default, though you can make them do so by setting the
813 PCRE_UCP option or by starting the pattern with (*UCP).
814 </P>
815 <br><b>
816 Extended grapheme clusters
817 </b><br>
818 <P>
819 The \X escape matches any number of Unicode characters that form an "extended
820 grapheme cluster", and treats the sequence as an atomic group
821 <a href="#atomicgroup">(see below).</a>
822 Up to and including release 8.31, PCRE matched an earlier, simpler definition
823 that was equivalent to
824 <pre>
825 (?&#62;\PM\pM*)
826 </pre>
827 That is, it matched a character without the "mark" property, followed by zero
828 or more characters with the "mark" property. Characters with the "mark"
829 property are typically non-spacing accents that affect the preceding character.
830 </P>
831 <P>
832 This simple definition was extended in Unicode to include more complicated
833 kinds of composite character by giving each character a grapheme breaking
834 property, and creating rules that use these properties to define the boundaries
835 of extended grapheme clusters. In releases of PCRE later than 8.31, \X matches
836 one of these clusters.
837 </P>
838 <P>
839 \X always matches at least one character. Then it decides whether to add
840 additional characters according to the following rules for ending a cluster:
841 </P>
842 <P>
843 1. End at the end of the subject string.
844 </P>
845 <P>
846 2. Do not end between CR and LF; otherwise end after any control character.
847 </P>
848 <P>
849 3. Do not break Hangul (a Korean script) syllable sequences. Hangul characters
850 are of five types: L, V, T, LV, and LVT. An L character may be followed by an
851 L, V, LV, or LVT character; an LV or V character may be followed by a V or T
852 character; an LVT or T character may be follwed only by a T character.
853 </P>
854 <P>
855 4. Do not end before extending characters or spacing marks. Characters with
856 the "mark" property always have the "extend" grapheme breaking property.
857 </P>
858 <P>
859 5. Do not end after prepend characters.
860 </P>
861 <P>
862 6. Otherwise, end the cluster.
863 <a name="extraprops"></a></P>
864 <br><b>
865 PCRE's additional properties
866 </b><br>
867 <P>
868 As well as the standard Unicode properties described above, PCRE supports four
869 more that make it possible to convert traditional escape sequences such as \w
870 and \s and POSIX character classes to use Unicode properties. PCRE uses these
871 non-standard, non-Perl properties internally when PCRE_UCP is set. They are:
872 <pre>
873 Xan Any alphanumeric character
874 Xps Any POSIX space character
875 Xsp Any Perl space character
876 Xwd Any Perl "word" character
877 </pre>
878 Xan matches characters that have either the L (letter) or the N (number)
879 property. Xps matches the characters tab, linefeed, vertical tab, form feed, or
880 carriage return, and any other character that has the Z (separator) property.
881 Xsp is the same as Xps, except that vertical tab is excluded. Xwd matches the
882 same characters as Xan, plus underscore.
883 <a name="resetmatchstart"></a></P>
884 <br><b>
885 Resetting the match start
886 </b><br>
887 <P>
888 The escape sequence \K causes any previously matched characters not to be
889 included in the final matched sequence. For example, the pattern:
890 <pre>
891 foo\Kbar
892 </pre>
893 matches "foobar", but reports that it has matched "bar". This feature is
894 similar to a lookbehind assertion
895 <a href="#lookbehind">(described below).</a>
896 However, in this case, the part of the subject before the real match does not
897 have to be of fixed length, as lookbehind assertions do. The use of \K does
898 not interfere with the setting of
899 <a href="#subpattern">captured substrings.</a>
900 For example, when the pattern
901 <pre>
902 (foo)\Kbar
903 </pre>
904 matches "foobar", the first substring is still set to "foo".
905 </P>
906 <P>
907 Perl documents that the use of \K within assertions is "not well defined". In
908 PCRE, \K is acted upon when it occurs inside positive assertions, but is
909 ignored in negative assertions.
910 <a name="smallassertions"></a></P>
911 <br><b>
912 Simple assertions
913 </b><br>
914 <P>
915 The final use of backslash is for certain simple assertions. An assertion
916 specifies a condition that has to be met at a particular point in a match,
917 without consuming any characters from the subject string. The use of
918 subpatterns for more complicated assertions is described
919 <a href="#bigassertions">below.</a>
920 The backslashed assertions are:
921 <pre>
922 \b matches at a word boundary
923 \B matches when not at a word boundary
924 \A matches at the start of the subject
925 \Z matches at the end of the subject
926 also matches before a newline at the end of the subject
927 \z matches only at the end of the subject
928 \G matches at the first matching position in the subject
929 </pre>
930 Inside a character class, \b has a different meaning; it matches the backspace
931 character. If any other of these assertions appears in a character class, by
932 default it matches the corresponding literal character (for example, \B
933 matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
934 escape sequence" error is generated instead.
935 </P>
936 <P>
937 A word boundary is a position in the subject string where the current character
938 and the previous character do not both match \w or \W (i.e. one matches
939 \w and the other matches \W), or the start or end of the string if the
940 first or last character matches \w, respectively. In a UTF mode, the meanings
941 of \w and \W can be changed by setting the PCRE_UCP option. When this is
942 done, it also affects \b and \B. Neither PCRE nor Perl has a separate "start
943 of word" or "end of word" metasequence. However, whatever follows \b normally
944 determines which it is. For example, the fragment \ba matches "a" at the start
945 of a word.
946 </P>
947 <P>
948 The \A, \Z, and \z assertions differ from the traditional circumflex and
949 dollar (described in the next section) in that they only ever match at the very
950 start and end of the subject string, whatever options are set. Thus, they are
951 independent of multiline mode. These three assertions are not affected by the
952 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
953 circumflex and dollar metacharacters. However, if the <i>startoffset</i>
954 argument of <b>pcre_exec()</b> is non-zero, indicating that matching is to start
955 at a point other than the beginning of the subject, \A can never match. The
956 difference between \Z and \z is that \Z matches before a newline at the end
957 of the string as well as at the very end, whereas \z matches only at the end.
958 </P>
959 <P>
960 The \G assertion is true only when the current matching position is at the
961 start point of the match, as specified by the <i>startoffset</i> argument of
962 <b>pcre_exec()</b>. It differs from \A when the value of <i>startoffset</i> is
963 non-zero. By calling <b>pcre_exec()</b> multiple times with appropriate
964 arguments, you can mimic Perl's /g option, and it is in this kind of
965 implementation where \G can be useful.
966 </P>
967 <P>
968 Note, however, that PCRE's interpretation of \G, as the start of the current
969 match, is subtly different from Perl's, which defines it as the end of the
970 previous match. In Perl, these can be different when the previously matched
971 string was empty. Because PCRE does just one match at a time, it cannot
972 reproduce this behaviour.
973 </P>
974 <P>
975 If all the alternatives of a pattern begin with \G, the expression is anchored
976 to the starting match position, and the "anchored" flag is set in the compiled
977 regular expression.
978 </P>
979 <br><a name="SEC6" href="#TOC1">CIRCUMFLEX AND DOLLAR</a><br>
980 <P>
981 Outside a character class, in the default matching mode, the circumflex
982 character is an assertion that is true only if the current matching point is
983 at the start of the subject string. If the <i>startoffset</i> argument of
984 <b>pcre_exec()</b> is non-zero, circumflex can never match if the PCRE_MULTILINE
985 option is unset. Inside a character class, circumflex has an entirely different
986 meaning
987 <a href="#characterclass">(see below).</a>
988 </P>
989 <P>
990 Circumflex need not be the first character of the pattern if a number of
991 alternatives are involved, but it should be the first thing in each alternative
992 in which it appears if the pattern is ever to match that branch. If all
993 possible alternatives start with a circumflex, that is, if the pattern is
994 constrained to match only at the start of the subject, it is said to be an
995 "anchored" pattern. (There are also other constructs that can cause a pattern
996 to be anchored.)
997 </P>
998 <P>
999 A dollar character is an assertion that is true only if the current matching
1000 point is at the end of the subject string, or immediately before a newline
1001 at the end of the string (by default). Dollar need not be the last character of
1002 the pattern if a number of alternatives are involved, but it should be the last
1003 item in any branch in which it appears. Dollar has no special meaning in a
1004 character class.
1005 </P>
1006 <P>
1007 The meaning of dollar can be changed so that it matches only at the very end of
1008 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
1009 does not affect the \Z assertion.
1010 </P>
1011 <P>
1012 The meanings of the circumflex and dollar characters are changed if the
1013 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
1014 immediately after internal newlines as well as at the start of the subject
1015 string. It does not match after a newline that ends the string. A dollar
1016 matches before any newlines in the string, as well as at the very end, when
1017 PCRE_MULTILINE is set. When newline is specified as the two-character
1018 sequence CRLF, isolated CR and LF characters do not indicate newlines.
1019 </P>
1020 <P>
1021 For example, the pattern /^abc$/ matches the subject string "def\nabc" (where
1022 \n represents a newline) in multiline mode, but not otherwise. Consequently,
1023 patterns that are anchored in single line mode because all branches start with
1024 ^ are not anchored in multiline mode, and a match for circumflex is possible
1025 when the <i>startoffset</i> argument of <b>pcre_exec()</b> is non-zero. The
1026 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
1027 </P>
1028 <P>
1029 Note that the sequences \A, \Z, and \z can be used to match the start and
1030 end of the subject in both modes, and if all branches of a pattern start with
1031 \A it is always anchored, whether or not PCRE_MULTILINE is set.
1032 <a name="fullstopdot"></a></P>
1033 <br><a name="SEC7" href="#TOC1">FULL STOP (PERIOD, DOT) AND \N</a><br>
1034 <P>
1035 Outside a character class, a dot in the pattern matches any one character in
1036 the subject string except (by default) a character that signifies the end of a
1037 line.
1038 </P>
1039 <P>
1040 When a line ending is defined as a single character, dot never matches that
1041 character; when the two-character sequence CRLF is used, dot does not match CR
1042 if it is immediately followed by LF, but otherwise it matches all characters
1043 (including isolated CRs and LFs). When any Unicode line endings are being
1044 recognized, dot does not match CR or LF or any of the other line ending
1045 characters.
1046 </P>
1047 <P>
1048 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
1049 option is set, a dot matches any one character, without exception. If the
1050 two-character sequence CRLF is present in the subject string, it takes two dots
1051 to match it.
1052 </P>
1053 <P>
1054 The handling of dot is entirely independent of the handling of circumflex and
1055 dollar, the only relationship being that they both involve newlines. Dot has no
1056 special meaning in a character class.
1057 </P>
1058 <P>
1059 The escape sequence \N behaves like a dot, except that it is not affected by
1060 the PCRE_DOTALL option. In other words, it matches any character except one
1061 that signifies the end of a line. Perl also uses \N to match characters by
1062 name; PCRE does not support this.
1063 </P>
1064 <br><a name="SEC8" href="#TOC1">MATCHING A SINGLE DATA UNIT</a><br>
1065 <P>
1066 Outside a character class, the escape sequence \C matches any one data unit,
1067 whether or not a UTF mode is set. In the 8-bit library, one data unit is one
1068 byte; in the 16-bit library it is a 16-bit unit; in the 32-bit library it is
1069 a 32-bit unit. Unlike a dot, \C always
1070 matches line-ending characters. The feature is provided in Perl in order to
1071 match individual bytes in UTF-8 mode, but it is unclear how it can usefully be
1072 used. Because \C breaks up characters into individual data units, matching one
1073 unit with \C in a UTF mode means that the rest of the string may start with a
1074 malformed UTF character. This has undefined results, because PCRE assumes that
1075 it is dealing with valid UTF strings (and by default it checks this at the
1076 start of processing unless the PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK or
1077 PCRE_NO_UTF32_CHECK option is used).
1078 </P>
1079 <P>
1080 PCRE does not allow \C to appear in lookbehind assertions
1081 <a href="#lookbehind">(described below)</a>
1082 in a UTF mode, because this would make it impossible to calculate the length of
1083 the lookbehind.
1084 </P>
1085 <P>
1086 In general, the \C escape sequence is best avoided. However, one
1087 way of using it that avoids the problem of malformed UTF characters is to use a
1088 lookahead to check the length of the next character, as in this pattern, which
1089 could be used with a UTF-8 string (ignore white space and line breaks):
1090 <pre>
1091 (?| (?=[\x00-\x7f])(\C) |
1092 (?=[\x80-\x{7ff}])(\C)(\C) |
1093 (?=[\x{800}-\x{ffff}])(\C)(\C)(\C) |
1094 (?=[\x{10000}-\x{1fffff}])(\C)(\C)(\C)(\C))
1095 </pre>
1096 A group that starts with (?| resets the capturing parentheses numbers in each
1097 alternative (see
1098 <a href="#dupsubpatternnumber">"Duplicate Subpattern Numbers"</a>
1099 below). The assertions at the start of each branch check the next UTF-8
1100 character for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The
1101 character's individual bytes are then captured by the appropriate number of
1102 groups.
1103 <a name="characterclass"></a></P>
1104 <br><a name="SEC9" href="#TOC1">SQUARE BRACKETS AND CHARACTER CLASSES</a><br>
1105 <P>
1106 An opening square bracket introduces a character class, terminated by a closing
1107 square bracket. A closing square bracket on its own is not special by default.
1108 However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
1109 bracket causes a compile-time error. If a closing square bracket is required as
1110 a member of the class, it should be the first data character in the class
1111 (after an initial circumflex, if present) or escaped with a backslash.
1112 </P>
1113 <P>
1114 A character class matches a single character in the subject. In a UTF mode, the
1115 character may be more than one data unit long. A matched character must be in
1116 the set of characters defined by the class, unless the first character in the
1117 class definition is a circumflex, in which case the subject character must not
1118 be in the set defined by the class. If a circumflex is actually required as a
1119 member of the class, ensure it is not the first character, or escape it with a
1120 backslash.
1121 </P>
1122 <P>
1123 For example, the character class [aeiou] matches any lower case vowel, while
1124 [^aeiou] matches any character that is not a lower case vowel. Note that a
1125 circumflex is just a convenient notation for specifying the characters that
1126 are in the class by enumerating those that are not. A class that starts with a
1127 circumflex is not an assertion; it still consumes a character from the subject
1128 string, and therefore it fails if the current pointer is at the end of the
1129 string.
1130 </P>
1131 <P>
1132 In UTF-8 (UTF-16, UTF-32) mode, characters with values greater than 255 (0xffff)
1133 can be included in a class as a literal string of data units, or by using the
1134 \x{ escaping mechanism.
1135 </P>
1136 <P>
1137 When caseless matching is set, any letters in a class represent both their
1138 upper case and lower case versions, so for example, a caseless [aeiou] matches
1139 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1140 caseful version would. In a UTF mode, PCRE always understands the concept of
1141 case for characters whose values are less than 128, so caseless matching is
1142 always possible. For characters with higher values, the concept of case is
1143 supported if PCRE is compiled with Unicode property support, but not otherwise.
1144 If you want to use caseless matching in a UTF mode for characters 128 and
1145 above, you must ensure that PCRE is compiled with Unicode property support as
1146 well as with UTF support.
1147 </P>
1148 <P>
1149 Characters that might indicate line breaks are never treated in any special way
1150 when matching character classes, whatever line-ending sequence is in use, and
1151 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
1152 such as [^a] always matches one of these characters.
1153 </P>
1154 <P>
1155 The minus (hyphen) character can be used to specify a range of characters in a
1156 character class. For example, [d-m] matches any letter between d and m,
1157 inclusive. If a minus character is required in a class, it must be escaped with
1158 a backslash or appear in a position where it cannot be interpreted as
1159 indicating a range, typically as the first or last character in the class.
1160 </P>
1161 <P>
1162 It is not possible to have the literal character "]" as the end character of a
1163 range. A pattern such as [W-]46] is interpreted as a class of two characters
1164 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1165 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1166 the end of range, so [W-\]46] is interpreted as a class containing a range
1167 followed by two other characters. The octal or hexadecimal representation of
1168 "]" can also be used to end a range.
1169 </P>
1170 <P>
1171 Ranges operate in the collating sequence of character values. They can also be
1172 used for characters specified numerically, for example [\000-\037]. Ranges
1173 can include any characters that are valid for the current mode.
1174 </P>
1175 <P>
1176 If a range that includes letters is used when caseless matching is set, it
1177 matches the letters in either case. For example, [W-c] is equivalent to
1178 [][\\^_`wxyzabc], matched caselessly, and in a non-UTF mode, if character
1179 tables for a French locale are in use, [\xc8-\xcb] matches accented E
1180 characters in both cases. In UTF modes, PCRE supports the concept of case for
1181 characters with values greater than 128 only when it is compiled with Unicode
1182 property support.
1183 </P>
1184 <P>
1185 The character escape sequences \d, \D, \h, \H, \p, \P, \s, \S, \v,
1186 \V, \w, and \W may appear in a character class, and add the characters that
1187 they match to the class. For example, [\dABCDEF] matches any hexadecimal
1188 digit. In UTF modes, the PCRE_UCP option affects the meanings of \d, \s, \w
1189 and their upper case partners, just as it does when they appear outside a
1190 character class, as described in the section entitled
1191 <a href="#genericchartypes">"Generic character types"</a>
1192 above. The escape sequence \b has a different meaning inside a character
1193 class; it matches the backspace character. The sequences \B, \N, \R, and \X
1194 are not special inside a character class. Like any other unrecognized escape
1195 sequences, they are treated as the literal characters "B", "N", "R", and "X" by
1196 default, but cause an error if the PCRE_EXTRA option is set.
1197 </P>
1198 <P>
1199 A circumflex can conveniently be used with the upper case character types to
1200 specify a more restricted set of characters than the matching lower case type.
1201 For example, the class [^\W_] matches any letter or digit, but not underscore,
1202 whereas [\w] includes underscore. A positive character class should be read as
1203 "something OR something OR ..." and a negative class as "NOT something AND NOT
1204 something AND NOT ...".
1205 </P>
1206 <P>
1207 The only metacharacters that are recognized in character classes are backslash,
1208 hyphen (only where it can be interpreted as specifying a range), circumflex
1209 (only at the start), opening square bracket (only when it can be interpreted as
1210 introducing a POSIX class name - see the next section), and the terminating
1211 closing square bracket. However, escaping other non-alphanumeric characters
1212 does no harm.
1213 </P>
1214 <br><a name="SEC10" href="#TOC1">POSIX CHARACTER CLASSES</a><br>
1215 <P>
1216 Perl supports the POSIX notation for character classes. This uses names
1217 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1218 this notation. For example,
1219 <pre>
1220 [01[:alpha:]%]
1221 </pre>
1222 matches "0", "1", any alphabetic character, or "%". The supported class names
1223 are:
1224 <pre>
1225 alnum letters and digits
1226 alpha letters
1227 ascii character codes 0 - 127
1228 blank space or tab only
1229 cntrl control characters
1230 digit decimal digits (same as \d)
1231 graph printing characters, excluding space
1232 lower lower case letters
1233 print printing characters, including space
1234 punct printing characters, excluding letters and digits and space
1235 space white space (not quite the same as \s)
1236 upper upper case letters
1237 word "word" characters (same as \w)
1238 xdigit hexadecimal digits
1239 </pre>
1240 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1241 space (32). Notice that this list includes the VT character (code 11). This
1242 makes "space" different to \s, which does not include VT (for Perl
1243 compatibility).
1244 </P>
1245 <P>
1246 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1247 5.8. Another Perl extension is negation, which is indicated by a ^ character
1248 after the colon. For example,
1249 <pre>
1250 [12[:^digit:]]
1251 </pre>
1252 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1253 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1254 supported, and an error is given if they are encountered.
1255 </P>
1256 <P>
1257 By default, in UTF modes, characters with values greater than 128 do not match
1258 any of the POSIX character classes. However, if the PCRE_UCP option is passed
1259 to <b>pcre_compile()</b>, some of the classes are changed so that Unicode
1260 character properties are used. This is achieved by replacing the POSIX classes
1261 by other sequences, as follows:
1262 <pre>
1263 [:alnum:] becomes \p{Xan}
1264 [:alpha:] becomes \p{L}
1265 [:blank:] becomes \h
1266 [:digit:] becomes \p{Nd}
1267 [:lower:] becomes \p{Ll}
1268 [:space:] becomes \p{Xps}
1269 [:upper:] becomes \p{Lu}
1270 [:word:] becomes \p{Xwd}
1271 </pre>
1272 Negated versions, such as [:^alpha:] use \P instead of \p. The other POSIX
1273 classes are unchanged, and match only characters with code points less than
1274 128.
1275 </P>
1276 <br><a name="SEC11" href="#TOC1">VERTICAL BAR</a><br>
1277 <P>
1278 Vertical bar characters are used to separate alternative patterns. For example,
1279 the pattern
1280 <pre>
1281 gilbert|sullivan
1282 </pre>
1283 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1284 and an empty alternative is permitted (matching the empty string). The matching
1285 process tries each alternative in turn, from left to right, and the first one
1286 that succeeds is used. If the alternatives are within a subpattern
1287 <a href="#subpattern">(defined below),</a>
1288 "succeeds" means matching the rest of the main pattern as well as the
1289 alternative in the subpattern.
1290 </P>
1291 <br><a name="SEC12" href="#TOC1">INTERNAL OPTION SETTING</a><br>
1292 <P>
1293 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1294 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1295 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1296 The option letters are
1297 <pre>
1298 i for PCRE_CASELESS
1300 s for PCRE_DOTALL
1301 x for PCRE_EXTENDED
1302 </pre>
1303 For example, (?im) sets caseless, multiline matching. It is also possible to
1304 unset these options by preceding the letter with a hyphen, and a combined
1305 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1306 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1307 permitted. If a letter appears both before and after the hyphen, the option is
1308 unset.
1309 </P>
1310 <P>
1311 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1312 changed in the same way as the Perl-compatible options by using the characters
1313 J, U and X respectively.
1314 </P>
1315 <P>
1316 When one of these option changes occurs at top level (that is, not inside
1317 subpattern parentheses), the change applies to the remainder of the pattern
1318 that follows. If the change is placed right at the start of a pattern, PCRE
1319 extracts it into the global options (and it will therefore show up in data
1320 extracted by the <b>pcre_fullinfo()</b> function).
1321 </P>
1322 <P>
1323 An option change within a subpattern (see below for a description of
1324 subpatterns) affects only that part of the subpattern that follows it, so
1325 <pre>
1326 (a(?i)b)c
1327 </pre>
1328 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1329 By this means, options can be made to have different settings in different
1330 parts of the pattern. Any changes made in one alternative do carry on
1331 into subsequent branches within the same subpattern. For example,
1332 <pre>
1333 (a(?i)b|c)
1334 </pre>
1335 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1336 branch is abandoned before the option setting. This is because the effects of
1337 option settings happen at compile time. There would be some very weird
1338 behaviour otherwise.
1339 </P>
1340 <P>
1341 <b>Note:</b> There are other PCRE-specific options that can be set by the
1342 application when the compiling or matching functions are called. In some cases
1343 the pattern can contain special leading sequences such as (*CRLF) to override
1344 what the application has set or what has been defaulted. Details are given in
1345 the section entitled
1346 <a href="#newlineseq">"Newline sequences"</a>
1347 above. There are also the (*UTF8), (*UTF16),(*UTF32) and (*UCP) leading
1348 sequences that can be used to set UTF and Unicode property modes; they are
1349 equivalent to setting the PCRE_UTF8, PCRE_UTF16, PCRE_UTF32 and the PCRE_UCP
1350 options, respectively.
1351 <a name="subpattern"></a></P>
1352 <br><a name="SEC13" href="#TOC1">SUBPATTERNS</a><br>
1353 <P>
1354 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1355 Turning part of a pattern into a subpattern does two things:
1356 <br>
1357 <br>
1358 1. It localizes a set of alternatives. For example, the pattern
1359 <pre>
1360 cat(aract|erpillar|)
1361 </pre>
1362 matches "cataract", "caterpillar", or "cat". Without the parentheses, it would
1363 match "cataract", "erpillar" or an empty string.
1364 <br>
1365 <br>
1366 2. It sets up the subpattern as a capturing subpattern. This means that, when
1367 the whole pattern matches, that portion of the subject string that matched the
1368 subpattern is passed back to the caller via the <i>ovector</i> argument of the
1369 matching function. (This applies only to the traditional matching functions;
1370 the DFA matching functions do not support capturing.)
1371 </P>
1372 <P>
1373 Opening parentheses are counted from left to right (starting from 1) to obtain
1374 numbers for the capturing subpatterns. For example, if the string "the red
1375 king" is matched against the pattern
1376 <pre>
1377 the ((red|white) (king|queen))
1378 </pre>
1379 the captured substrings are "red king", "red", and "king", and are numbered 1,
1380 2, and 3, respectively.
1381 </P>
1382 <P>
1383 The fact that plain parentheses fulfil two functions is not always helpful.
1384 There are often times when a grouping subpattern is required without a
1385 capturing requirement. If an opening parenthesis is followed by a question mark
1386 and a colon, the subpattern does not do any capturing, and is not counted when
1387 computing the number of any subsequent capturing subpatterns. For example, if
1388 the string "the white queen" is matched against the pattern
1389 <pre>
1390 the ((?:red|white) (king|queen))
1391 </pre>
1392 the captured substrings are "white queen" and "queen", and are numbered 1 and
1393 2. The maximum number of capturing subpatterns is 65535.
1394 </P>
1395 <P>
1396 As a convenient shorthand, if any option settings are required at the start of
1397 a non-capturing subpattern, the option letters may appear between the "?" and
1398 the ":". Thus the two patterns
1399 <pre>
1400 (?i:saturday|sunday)
1401 (?:(?i)saturday|sunday)
1402 </pre>
1403 match exactly the same set of strings. Because alternative branches are tried
1404 from left to right, and options are not reset until the end of the subpattern
1405 is reached, an option setting in one branch does affect subsequent branches, so
1406 the above patterns match "SUNDAY" as well as "Saturday".
1407 <a name="dupsubpatternnumber"></a></P>
1408 <br><a name="SEC14" href="#TOC1">DUPLICATE SUBPATTERN NUMBERS</a><br>
1409 <P>
1410 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1411 the same numbers for its capturing parentheses. Such a subpattern starts with
1412 (?| and is itself a non-capturing subpattern. For example, consider this
1413 pattern:
1414 <pre>
1415 (?|(Sat)ur|(Sun))day
1416 </pre>
1417 Because the two alternatives are inside a (?| group, both sets of capturing
1418 parentheses are numbered one. Thus, when the pattern matches, you can look
1419 at captured substring number one, whichever alternative matched. This construct
1420 is useful when you want to capture part, but not all, of one of a number of
1421 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1422 number is reset at the start of each branch. The numbers of any capturing
1423 parentheses that follow the subpattern start after the highest number used in
1424 any branch. The following example is taken from the Perl documentation. The
1425 numbers underneath show in which buffer the captured content will be stored.
1426 <pre>
1427 # before ---------------branch-reset----------- after
1428 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1429 # 1 2 2 3 2 3 4
1430 </pre>
1431 A back reference to a numbered subpattern uses the most recent value that is
1432 set for that number by any subpattern. The following pattern matches "abcabc"
1433 or "defdef":
1434 <pre>
1435 /(?|(abc)|(def))\1/
1436 </pre>
1437 In contrast, a subroutine call to a numbered subpattern always refers to the
1438 first one in the pattern with the given number. The following pattern matches
1439 "abcabc" or "defabc":
1440 <pre>
1441 /(?|(abc)|(def))(?1)/
1442 </pre>
1443 If a
1444 <a href="#conditions">condition test</a>
1445 for a subpattern's having matched refers to a non-unique number, the test is
1446 true if any of the subpatterns of that number have matched.
1447 </P>
1448 <P>
1449 An alternative approach to using this "branch reset" feature is to use
1450 duplicate named subpatterns, as described in the next section.
1451 </P>
1452 <br><a name="SEC15" href="#TOC1">NAMED SUBPATTERNS</a><br>
1453 <P>
1454 Identifying capturing parentheses by number is simple, but it can be very hard
1455 to keep track of the numbers in complicated regular expressions. Furthermore,
1456 if an expression is modified, the numbers may change. To help with this
1457 difficulty, PCRE supports the naming of subpatterns. This feature was not
1458 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1459 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1460 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1461 have different names, but PCRE does not.
1462 </P>
1463 <P>
1464 In PCRE, a subpattern can be named in one of three ways: (?&#60;name&#62;...) or
1465 (?'name'...) as in Perl, or (?P&#60;name&#62;...) as in Python. References to capturing
1466 parentheses from other parts of the pattern, such as
1467 <a href="#backreferences">back references,</a>
1468 <a href="#recursion">recursion,</a>
1469 and
1470 <a href="#conditions">conditions,</a>
1471 can be made by name as well as by number.
1472 </P>
1473 <P>
1474 Names consist of up to 32 alphanumeric characters and underscores. Named
1475 capturing parentheses are still allocated numbers as well as names, exactly as
1476 if the names were not present. The PCRE API provides function calls for
1477 extracting the name-to-number translation table from a compiled pattern. There
1478 is also a convenience function for extracting a captured substring by name.
1479 </P>
1480 <P>
1481 By default, a name must be unique within a pattern, but it is possible to relax
1482 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1483 names are also always permitted for subpatterns with the same number, set up as
1484 described in the previous section.) Duplicate names can be useful for patterns
1485 where only one instance of the named parentheses can match. Suppose you want to
1486 match the name of a weekday, either as a 3-letter abbreviation or as the full
1487 name, and in both cases you want to extract the abbreviation. This pattern
1488 (ignoring the line breaks) does the job:
1489 <pre>
1490 (?&#60;DN&#62;Mon|Fri|Sun)(?:day)?|
1491 (?&#60;DN&#62;Tue)(?:sday)?|
1492 (?&#60;DN&#62;Wed)(?:nesday)?|
1493 (?&#60;DN&#62;Thu)(?:rsday)?|
1494 (?&#60;DN&#62;Sat)(?:urday)?
1495 </pre>
1496 There are five capturing substrings, but only one is ever set after a match.
1497 (An alternative way of solving this problem is to use a "branch reset"
1498 subpattern, as described in the previous section.)
1499 </P>
1500 <P>
1501 The convenience function for extracting the data by name returns the substring
1502 for the first (and in this example, the only) subpattern of that name that
1503 matched. This saves searching to find which numbered subpattern it was.
1504 </P>
1505 <P>
1506 If you make a back reference to a non-unique named subpattern from elsewhere in
1507 the pattern, the one that corresponds to the first occurrence of the name is
1508 used. In the absence of duplicate numbers (see the previous section) this is
1509 the one with the lowest number. If you use a named reference in a condition
1510 test (see the
1511 <a href="#conditions">section about conditions</a>
1512 below), either to check whether a subpattern has matched, or to check for
1513 recursion, all subpatterns with the same name are tested. If the condition is
1514 true for any one of them, the overall condition is true. This is the same
1515 behaviour as testing by number. For further details of the interfaces for
1516 handling named subpatterns, see the
1517 <a href="pcreapi.html"><b>pcreapi</b></a>
1518 documentation.
1519 </P>
1520 <P>
1521 <b>Warning:</b> You cannot use different names to distinguish between two
1522 subpatterns with the same number because PCRE uses only the numbers when
1523 matching. For this reason, an error is given at compile time if different names
1524 are given to subpatterns with the same number. However, you can give the same
1525 name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
1526 </P>
1527 <br><a name="SEC16" href="#TOC1">REPETITION</a><br>
1528 <P>
1529 Repetition is specified by quantifiers, which can follow any of the following
1530 items:
1531 <pre>
1532 a literal data character
1533 the dot metacharacter
1534 the \C escape sequence
1535 the \X escape sequence
1536 the \R escape sequence
1537 an escape such as \d or \pL that matches a single character
1538 a character class
1539 a back reference (see next section)
1540 a parenthesized subpattern (including assertions)
1541 a subroutine call to a subpattern (recursive or otherwise)
1542 </pre>
1543 The general repetition quantifier specifies a minimum and maximum number of
1544 permitted matches, by giving the two numbers in curly brackets (braces),
1545 separated by a comma. The numbers must be less than 65536, and the first must
1546 be less than or equal to the second. For example:
1547 <pre>
1548 z{2,4}
1549 </pre>
1550 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1551 character. If the second number is omitted, but the comma is present, there is
1552 no upper limit; if the second number and the comma are both omitted, the
1553 quantifier specifies an exact number of required matches. Thus
1554 <pre>
1555 [aeiou]{3,}
1556 </pre>
1557 matches at least 3 successive vowels, but may match many more, while
1558 <pre>
1559 \d{8}
1560 </pre>
1561 matches exactly 8 digits. An opening curly bracket that appears in a position
1562 where a quantifier is not allowed, or one that does not match the syntax of a
1563 quantifier, is taken as a literal character. For example, {,6} is not a
1564 quantifier, but a literal string of four characters.
1565 </P>
1566 <P>
1567 In UTF modes, quantifiers apply to characters rather than to individual data
1568 units. Thus, for example, \x{100}{2} matches two characters, each of
1569 which is represented by a two-byte sequence in a UTF-8 string. Similarly,
1570 \X{3} matches three Unicode extended grapheme clusters, each of which may be
1571 several data units long (and they may be of different lengths).
1572 </P>
1573 <P>
1574 The quantifier {0} is permitted, causing the expression to behave as if the
1575 previous item and the quantifier were not present. This may be useful for
1576 subpatterns that are referenced as
1577 <a href="#subpatternsassubroutines">subroutines</a>
1578 from elsewhere in the pattern (but see also the section entitled
1579 <a href="#subdefine">"Defining subpatterns for use by reference only"</a>
1580 below). Items other than subpatterns that have a {0} quantifier are omitted
1581 from the compiled pattern.
1582 </P>
1583 <P>
1584 For convenience, the three most common quantifiers have single-character
1585 abbreviations:
1586 <pre>
1587 * is equivalent to {0,}
1588 + is equivalent to {1,}
1589 ? is equivalent to {0,1}
1590 </pre>
1591 It is possible to construct infinite loops by following a subpattern that can
1592 match no characters with a quantifier that has no upper limit, for example:
1593 <pre>
1594 (a?)*
1595 </pre>
1596 Earlier versions of Perl and PCRE used to give an error at compile time for
1597 such patterns. However, because there are cases where this can be useful, such
1598 patterns are now accepted, but if any repetition of the subpattern does in fact
1599 match no characters, the loop is forcibly broken.
1600 </P>
1601 <P>
1602 By default, the quantifiers are "greedy", that is, they match as much as
1603 possible (up to the maximum number of permitted times), without causing the
1604 rest of the pattern to fail. The classic example of where this gives problems
1605 is in trying to match comments in C programs. These appear between /* and */
1606 and within the comment, individual * and / characters may appear. An attempt to
1607 match C comments by applying the pattern
1608 <pre>
1609 /\*.*\*/
1610 </pre>
1611 to the string
1612 <pre>
1613 /* first comment */ not comment /* second comment */
1614 </pre>
1615 fails, because it matches the entire string owing to the greediness of the .*
1616 item.
1617 </P>
1618 <P>
1619 However, if a quantifier is followed by a question mark, it ceases to be
1620 greedy, and instead matches the minimum number of times possible, so the
1621 pattern
1622 <pre>
1623 /\*.*?\*/
1624 </pre>
1625 does the right thing with the C comments. The meaning of the various
1626 quantifiers is not otherwise changed, just the preferred number of matches.
1627 Do not confuse this use of question mark with its use as a quantifier in its
1628 own right. Because it has two uses, it can sometimes appear doubled, as in
1629 <pre>
1630 \d??\d
1631 </pre>
1632 which matches one digit by preference, but can match two if that is the only
1633 way the rest of the pattern matches.
1634 </P>
1635 <P>
1636 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1637 the quantifiers are not greedy by default, but individual ones can be made
1638 greedy by following them with a question mark. In other words, it inverts the
1639 default behaviour.
1640 </P>
1641 <P>
1642 When a parenthesized subpattern is quantified with a minimum repeat count that
1643 is greater than 1 or with a limited maximum, more memory is required for the
1644 compiled pattern, in proportion to the size of the minimum or maximum.
1645 </P>
1646 <P>
1647 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1648 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1649 implicitly anchored, because whatever follows will be tried against every
1650 character position in the subject string, so there is no point in retrying the
1651 overall match at any position after the first. PCRE normally treats such a
1652 pattern as though it were preceded by \A.
1653 </P>
1654 <P>
1655 In cases where it is known that the subject string contains no newlines, it is
1656 worth setting PCRE_DOTALL in order to obtain this optimization, or
1657 alternatively using ^ to indicate anchoring explicitly.
1658 </P>
1659 <P>
1660 However, there are some cases where the optimization cannot be used. When .*
1661 is inside capturing parentheses that are the subject of a back reference
1662 elsewhere in the pattern, a match at the start may fail where a later one
1663 succeeds. Consider, for example:
1664 <pre>
1665 (.*)abc\1
1666 </pre>
1667 If the subject is "xyz123abc123" the match point is the fourth character. For
1668 this reason, such a pattern is not implicitly anchored.
1669 </P>
1670 <P>
1671 Another case where implicit anchoring is not applied is when the leading .* is
1672 inside an atomic group. Once again, a match at the start may fail where a later
1673 one succeeds. Consider this pattern:
1674 <pre>
1675 (?&#62;.*?a)b
1676 </pre>
1677 It matches "ab" in the subject "aab". The use of the backtracking control verbs
1678 (*PRUNE) and (*SKIP) also disable this optimization.
1679 </P>
1680 <P>
1681 When a capturing subpattern is repeated, the value captured is the substring
1682 that matched the final iteration. For example, after
1683 <pre>
1684 (tweedle[dume]{3}\s*)+
1685 </pre>
1686 has matched "tweedledum tweedledee" the value of the captured substring is
1687 "tweedledee". However, if there are nested capturing subpatterns, the
1688 corresponding captured values may have been set in previous iterations. For
1689 example, after
1690 <pre>
1691 /(a|(b))+/
1692 </pre>
1693 matches "aba" the value of the second captured substring is "b".
1694 <a name="atomicgroup"></a></P>
1695 <br><a name="SEC17" href="#TOC1">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a><br>
1696 <P>
1697 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1698 repetition, failure of what follows normally causes the repeated item to be
1699 re-evaluated to see if a different number of repeats allows the rest of the
1700 pattern to match. Sometimes it is useful to prevent this, either to change the
1701 nature of the match, or to cause it fail earlier than it otherwise might, when
1702 the author of the pattern knows there is no point in carrying on.
1703 </P>
1704 <P>
1705 Consider, for example, the pattern \d+foo when applied to the subject line
1706 <pre>
1707 123456bar
1708 </pre>
1709 After matching all 6 digits and then failing to match "foo", the normal
1710 action of the matcher is to try again with only 5 digits matching the \d+
1711 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1712 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1713 that once a subpattern has matched, it is not to be re-evaluated in this way.
1714 </P>
1715 <P>
1716 If we use atomic grouping for the previous example, the matcher gives up
1717 immediately on failing to match "foo" the first time. The notation is a kind of
1718 special parenthesis, starting with (?&#62; as in this example:
1719 <pre>
1720 (?&#62;\d+)foo
1721 </pre>
1722 This kind of parenthesis "locks up" the part of the pattern it contains once
1723 it has matched, and a failure further into the pattern is prevented from
1724 backtracking into it. Backtracking past it to previous items, however, works as
1725 normal.
1726 </P>
1727 <P>
1728 An alternative description is that a subpattern of this type matches the string
1729 of characters that an identical standalone pattern would match, if anchored at
1730 the current point in the subject string.
1731 </P>
1732 <P>
1733 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1734 the above example can be thought of as a maximizing repeat that must swallow
1735 everything it can. So, while both \d+ and \d+? are prepared to adjust the
1736 number of digits they match in order to make the rest of the pattern match,
1737 (?&#62;\d+) can only match an entire sequence of digits.
1738 </P>
1739 <P>
1740 Atomic groups in general can of course contain arbitrarily complicated
1741 subpatterns, and can be nested. However, when the subpattern for an atomic
1742 group is just a single repeated item, as in the example above, a simpler
1743 notation, called a "possessive quantifier" can be used. This consists of an
1744 additional + character following a quantifier. Using this notation, the
1745 previous example can be rewritten as
1746 <pre>
1747 \d++foo
1748 </pre>
1749 Note that a possessive quantifier can be used with an entire group, for
1750 example:
1751 <pre>
1752 (abc|xyz){2,3}+
1753 </pre>
1754 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1755 option is ignored. They are a convenient notation for the simpler forms of
1756 atomic group. However, there is no difference in the meaning of a possessive
1757 quantifier and the equivalent atomic group, though there may be a performance
1758 difference; possessive quantifiers should be slightly faster.
1759 </P>
1760 <P>
1761 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1762 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1763 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1764 package, and PCRE copied it from there. It ultimately found its way into Perl
1765 at release 5.10.
1766 </P>
1767 <P>
1768 PCRE has an optimization that automatically "possessifies" certain simple
1769 pattern constructs. For example, the sequence A+B is treated as A++B because
1770 there is no point in backtracking into a sequence of A's when B must follow.
1771 </P>
1772 <P>
1773 When a pattern contains an unlimited repeat inside a subpattern that can itself
1774 be repeated an unlimited number of times, the use of an atomic group is the
1775 only way to avoid some failing matches taking a very long time indeed. The
1776 pattern
1777 <pre>
1778 (\D+|&#60;\d+&#62;)*[!?]
1779 </pre>
1780 matches an unlimited number of substrings that either consist of non-digits, or
1781 digits enclosed in &#60;&#62;, followed by either ! or ?. When it matches, it runs
1782 quickly. However, if it is applied to
1783 <pre>
1784 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1785 </pre>
1786 it takes a long time before reporting failure. This is because the string can
1787 be divided between the internal \D+ repeat and the external * repeat in a
1788 large number of ways, and all have to be tried. (The example uses [!?] rather
1789 than a single character at the end, because both PCRE and Perl have an
1790 optimization that allows for fast failure when a single character is used. They
1791 remember the last single character that is required for a match, and fail early
1792 if it is not present in the string.) If the pattern is changed so that it uses
1793 an atomic group, like this:
1794 <pre>
1795 ((?&#62;\D+)|&#60;\d+&#62;)*[!?]
1796 </pre>
1797 sequences of non-digits cannot be broken, and failure happens quickly.
1798 <a name="backreferences"></a></P>
1799 <br><a name="SEC18" href="#TOC1">BACK REFERENCES</a><br>
1800 <P>
1801 Outside a character class, a backslash followed by a digit greater than 0 (and
1802 possibly further digits) is a back reference to a capturing subpattern earlier
1803 (that is, to its left) in the pattern, provided there have been that many
1804 previous capturing left parentheses.
1805 </P>
1806 <P>
1807 However, if the decimal number following the backslash is less than 10, it is
1808 always taken as a back reference, and causes an error only if there are not
1809 that many capturing left parentheses in the entire pattern. In other words, the
1810 parentheses that are referenced need not be to the left of the reference for
1811 numbers less than 10. A "forward back reference" of this type can make sense
1812 when a repetition is involved and the subpattern to the right has participated
1813 in an earlier iteration.
1814 </P>
1815 <P>
1816 It is not possible to have a numerical "forward back reference" to a subpattern
1817 whose number is 10 or more using this syntax because a sequence such as \50 is
1818 interpreted as a character defined in octal. See the subsection entitled
1819 "Non-printing characters"
1820 <a href="#digitsafterbackslash">above</a>
1821 for further details of the handling of digits following a backslash. There is
1822 no such problem when named parentheses are used. A back reference to any
1823 subpattern is possible using named parentheses (see below).
1824 </P>
1825 <P>
1826 Another way of avoiding the ambiguity inherent in the use of digits following a
1827 backslash is to use the \g escape sequence. This escape must be followed by an
1828 unsigned number or a negative number, optionally enclosed in braces. These
1829 examples are all identical:
1830 <pre>
1831 (ring), \1
1832 (ring), \g1
1833 (ring), \g{1}
1834 </pre>
1835 An unsigned number specifies an absolute reference without the ambiguity that
1836 is present in the older syntax. It is also useful when literal digits follow
1837 the reference. A negative number is a relative reference. Consider this
1838 example:
1839 <pre>
1840 (abc(def)ghi)\g{-1}
1841 </pre>
1842 The sequence \g{-1} is a reference to the most recently started capturing
1843 subpattern before \g, that is, is it equivalent to \2 in this example.
1844 Similarly, \g{-2} would be equivalent to \1. The use of relative references
1845 can be helpful in long patterns, and also in patterns that are created by
1846 joining together fragments that contain references within themselves.
1847 </P>
1848 <P>
1849 A back reference matches whatever actually matched the capturing subpattern in
1850 the current subject string, rather than anything matching the subpattern
1851 itself (see
1852 <a href="#subpatternsassubroutines">"Subpatterns as subroutines"</a>
1853 below for a way of doing that). So the pattern
1854 <pre>
1855 (sens|respons)e and \1ibility
1856 </pre>
1857 matches "sense and sensibility" and "response and responsibility", but not
1858 "sense and responsibility". If caseful matching is in force at the time of the
1859 back reference, the case of letters is relevant. For example,
1860 <pre>
1861 ((?i)rah)\s+\1
1862 </pre>
1863 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1864 capturing subpattern is matched caselessly.
1865 </P>
1866 <P>
1867 There are several different ways of writing back references to named
1868 subpatterns. The .NET syntax \k{name} and the Perl syntax \k&#60;name&#62; or
1869 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1870 back reference syntax, in which \g can be used for both numeric and named
1871 references, is also supported. We could rewrite the above example in any of
1872 the following ways:
1873 <pre>
1874 (?&#60;p1&#62;(?i)rah)\s+\k&#60;p1&#62;
1875 (?'p1'(?i)rah)\s+\k{p1}
1876 (?P&#60;p1&#62;(?i)rah)\s+(?P=p1)
1877 (?&#60;p1&#62;(?i)rah)\s+\g{p1}
1878 </pre>
1879 A subpattern that is referenced by name may appear in the pattern before or
1880 after the reference.
1881 </P>
1882 <P>
1883 There may be more than one back reference to the same subpattern. If a
1884 subpattern has not actually been used in a particular match, any back
1885 references to it always fail by default. For example, the pattern
1886 <pre>
1887 (a|(bc))\2
1888 </pre>
1889 always fails if it starts to match "a" rather than "bc". However, if the
1890 PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1891 unset value matches an empty string.
1892 </P>
1893 <P>
1894 Because there may be many capturing parentheses in a pattern, all digits
1895 following a backslash are taken as part of a potential back reference number.
1896 If the pattern continues with a digit character, some delimiter must be used to
1897 terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1898 white space. Otherwise, the \g{ syntax or an empty comment (see
1899 <a href="#comments">"Comments"</a>
1900 below) can be used.
1901 </P>
1902 <br><b>
1903 Recursive back references
1904 </b><br>
1905 <P>
1906 A back reference that occurs inside the parentheses to which it refers fails
1907 when the subpattern is first used, so, for example, (a\1) never matches.
1908 However, such references can be useful inside repeated subpatterns. For
1909 example, the pattern
1910 <pre>
1911 (a|b\1)+
1912 </pre>
1913 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1914 the subpattern, the back reference matches the character string corresponding
1915 to the previous iteration. In order for this to work, the pattern must be such
1916 that the first iteration does not need to match the back reference. This can be
1917 done using alternation, as in the example above, or by a quantifier with a
1918 minimum of zero.
1919 </P>
1920 <P>
1921 Back references of this type cause the group that they reference to be treated
1922 as an
1923 <a href="#atomicgroup">atomic group.</a>
1924 Once the whole group has been matched, a subsequent matching failure cannot
1925 cause backtracking into the middle of the group.
1926 <a name="bigassertions"></a></P>
1927 <br><a name="SEC19" href="#TOC1">ASSERTIONS</a><br>
1928 <P>
1929 An assertion is a test on the characters following or preceding the current
1930 matching point that does not actually consume any characters. The simple
1931 assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are described
1932 <a href="#smallassertions">above.</a>
1933 </P>
1934 <P>
1935 More complicated assertions are coded as subpatterns. There are two kinds:
1936 those that look ahead of the current position in the subject string, and those
1937 that look behind it. An assertion subpattern is matched in the normal way,
1938 except that it does not cause the current matching position to be changed.
1939 </P>
1940 <P>
1941 Assertion subpatterns are not capturing subpatterns. If such an assertion
1942 contains capturing subpatterns within it, these are counted for the purposes of
1943 numbering the capturing subpatterns in the whole pattern. However, substring
1944 capturing is carried out only for positive assertions, because it does not make
1945 sense for negative assertions.
1946 </P>
1947 <P>
1948 For compatibility with Perl, assertion subpatterns may be repeated; though
1949 it makes no sense to assert the same thing several times, the side effect of
1950 capturing parentheses may occasionally be useful. In practice, there only three
1951 cases:
1952 <br>
1953 <br>
1954 (1) If the quantifier is {0}, the assertion is never obeyed during matching.
1955 However, it may contain internal capturing parenthesized groups that are called
1956 from elsewhere via the
1957 <a href="#subpatternsassubroutines">subroutine mechanism.</a>
1958 <br>
1959 <br>
1960 (2) If quantifier is {0,n} where n is greater than zero, it is treated as if it
1961 were {0,1}. At run time, the rest of the pattern match is tried with and
1962 without the assertion, the order depending on the greediness of the quantifier.
1963 <br>
1964 <br>
1965 (3) If the minimum repetition is greater than zero, the quantifier is ignored.
1966 The assertion is obeyed just once when encountered during matching.
1967 </P>
1968 <br><b>
1969 Lookahead assertions
1970 </b><br>
1971 <P>
1972 Lookahead assertions start with (?= for positive assertions and (?! for
1973 negative assertions. For example,
1974 <pre>
1975 \w+(?=;)
1976 </pre>
1977 matches a word followed by a semicolon, but does not include the semicolon in
1978 the match, and
1979 <pre>
1980 foo(?!bar)
1981 </pre>
1982 matches any occurrence of "foo" that is not followed by "bar". Note that the
1983 apparently similar pattern
1984 <pre>
1985 (?!foo)bar
1986 </pre>
1987 does not find an occurrence of "bar" that is preceded by something other than
1988 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1989 (?!foo) is always true when the next three characters are "bar". A
1990 lookbehind assertion is needed to achieve the other effect.
1991 </P>
1992 <P>
1993 If you want to force a matching failure at some point in a pattern, the most
1994 convenient way to do it is with (?!) because an empty string always matches, so
1995 an assertion that requires there not to be an empty string must always fail.
1996 The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).
1997 <a name="lookbehind"></a></P>
1998 <br><b>
1999 Lookbehind assertions
2000 </b><br>
2001 <P>
2002 Lookbehind assertions start with (?&#60;= for positive assertions and (?&#60;! for
2003 negative assertions. For example,
2004 <pre>
2005 (?&#60;!foo)bar
2006 </pre>
2007 does find an occurrence of "bar" that is not preceded by "foo". The contents of
2008 a lookbehind assertion are restricted such that all the strings it matches must
2009 have a fixed length. However, if there are several top-level alternatives, they
2010 do not all have to have the same fixed length. Thus
2011 <pre>
2012 (?&#60;=bullock|donkey)
2013 </pre>
2014 is permitted, but
2015 <pre>
2016 (?&#60;!dogs?|cats?)
2017 </pre>
2018 causes an error at compile time. Branches that match different length strings
2019 are permitted only at the top level of a lookbehind assertion. This is an
2020 extension compared with Perl, which requires all branches to match the same
2021 length of string. An assertion such as
2022 <pre>
2023 (?&#60;=ab(c|de))
2024 </pre>
2025 is not permitted, because its single top-level branch can match two different
2026 lengths, but it is acceptable to PCRE if rewritten to use two top-level
2027 branches:
2028 <pre>
2029 (?&#60;=abc|abde)
2030 </pre>
2031 In some cases, the escape sequence \K
2032 <a href="#resetmatchstart">(see above)</a>
2033 can be used instead of a lookbehind assertion to get round the fixed-length
2034 restriction.
2035 </P>
2036 <P>
2037 The implementation of lookbehind assertions is, for each alternative, to
2038 temporarily move the current position back by the fixed length and then try to
2039 match. If there are insufficient characters before the current position, the
2040 assertion fails.
2041 </P>
2042 <P>
2043 In a UTF mode, PCRE does not allow the \C escape (which matches a single data
2044 unit even in a UTF mode) to appear in lookbehind assertions, because it makes
2045 it impossible to calculate the length of the lookbehind. The \X and \R
2046 escapes, which can match different numbers of data units, are also not
2047 permitted.
2048 </P>
2049 <P>
2050 <a href="#subpatternsassubroutines">"Subroutine"</a>
2051 calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
2052 as the subpattern matches a fixed-length string.
2053 <a href="#recursion">Recursion,</a>
2054 however, is not supported.
2055 </P>
2056 <P>
2057 Possessive quantifiers can be used in conjunction with lookbehind assertions to
2058 specify efficient matching of fixed-length strings at the end of subject
2059 strings. Consider a simple pattern such as
2060 <pre>
2061 abcd$
2062 </pre>
2063 when applied to a long string that does not match. Because matching proceeds
2064 from left to right, PCRE will look for each "a" in the subject and then see if
2065 what follows matches the rest of the pattern. If the pattern is specified as
2066 <pre>
2067 ^.*abcd$
2068 </pre>
2069 the initial .* matches the entire string at first, but when this fails (because
2070 there is no following "a"), it backtracks to match all but the last character,
2071 then all but the last two characters, and so on. Once again the search for "a"
2072 covers the entire string, from right to left, so we are no better off. However,
2073 if the pattern is written as
2074 <pre>
2075 ^.*+(?&#60;=abcd)
2076 </pre>
2077 there can be no backtracking for the .*+ item; it can match only the entire
2078 string. The subsequent lookbehind assertion does a single test on the last four
2079 characters. If it fails, the match fails immediately. For long strings, this
2080 approach makes a significant difference to the processing time.
2081 </P>
2082 <br><b>
2083 Using multiple assertions
2084 </b><br>
2085 <P>
2086 Several assertions (of any sort) may occur in succession. For example,
2087 <pre>
2088 (?&#60;=\d{3})(?&#60;!999)foo
2089 </pre>
2090 matches "foo" preceded by three digits that are not "999". Notice that each of
2091 the assertions is applied independently at the same point in the subject
2092 string. First there is a check that the previous three characters are all
2093 digits, and then there is a check that the same three characters are not "999".
2094 This pattern does <i>not</i> match "foo" preceded by six characters, the first
2095 of which are digits and the last three of which are not "999". For example, it
2096 doesn't match "123abcfoo". A pattern to do that is
2097 <pre>
2098 (?&#60;=\d{3}...)(?&#60;!999)foo
2099 </pre>
2100 This time the first assertion looks at the preceding six characters, checking
2101 that the first three are digits, and then the second assertion checks that the
2102 preceding three characters are not "999".
2103 </P>
2104 <P>
2105 Assertions can be nested in any combination. For example,
2106 <pre>
2107 (?&#60;=(?&#60;!foo)bar)baz
2108 </pre>
2109 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
2110 preceded by "foo", while
2111 <pre>
2112 (?&#60;=\d{3}(?!999)...)foo
2113 </pre>
2114 is another pattern that matches "foo" preceded by three digits and any three
2115 characters that are not "999".
2116 <a name="conditions"></a></P>
2117 <br><a name="SEC20" href="#TOC1">CONDITIONAL SUBPATTERNS</a><br>
2118 <P>
2119 It is possible to cause the matching process to obey a subpattern
2120 conditionally or to choose between two alternative subpatterns, depending on
2121 the result of an assertion, or whether a specific capturing subpattern has
2122 already been matched. The two possible forms of conditional subpattern are:
2123 <pre>
2124 (?(condition)yes-pattern)
2125 (?(condition)yes-pattern|no-pattern)
2126 </pre>
2127 If the condition is satisfied, the yes-pattern is used; otherwise the
2128 no-pattern (if present) is used. If there are more than two alternatives in the
2129 subpattern, a compile-time error occurs. Each of the two alternatives may
2130 itself contain nested subpatterns of any form, including conditional
2131 subpatterns; the restriction to two alternatives applies only at the level of
2132 the condition. This pattern fragment is an example where the alternatives are
2133 complex:
2134 <pre>
2135 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
2137 </PRE>
2138 </P>
2139 <P>
2140 There are four kinds of condition: references to subpatterns, references to
2141 recursion, a pseudo-condition called DEFINE, and assertions.
2142 </P>
2143 <br><b>
2144 Checking for a used subpattern by number
2145 </b><br>
2146 <P>
2147 If the text between the parentheses consists of a sequence of digits, the
2148 condition is true if a capturing subpattern of that number has previously
2149 matched. If there is more than one capturing subpattern with the same number
2150 (see the earlier
2151 <a href="#recursion">section about duplicate subpattern numbers),</a>
2152 the condition is true if any of them have matched. An alternative notation is
2153 to precede the digits with a plus or minus sign. In this case, the subpattern
2154 number is relative rather than absolute. The most recently opened parentheses
2155 can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside
2156 loops it can also make sense to refer to subsequent groups. The next
2157 parentheses to be opened can be referenced as (?(+1), and so on. (The value
2158 zero in any of these forms is not used; it provokes a compile-time error.)
2159 </P>
2160 <P>
2161 Consider the following pattern, which contains non-significant white space to
2162 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
2163 three parts for ease of discussion:
2164 <pre>
2165 ( \( )? [^()]+ (?(1) \) )
2166 </pre>
2167 The first part matches an optional opening parenthesis, and if that
2168 character is present, sets it as the first captured substring. The second part
2169 matches one or more characters that are not parentheses. The third part is a
2170 conditional subpattern that tests whether or not the first set of parentheses
2171 matched. If they did, that is, if subject started with an opening parenthesis,
2172 the condition is true, and so the yes-pattern is executed and a closing
2173 parenthesis is required. Otherwise, since no-pattern is not present, the
2174 subpattern matches nothing. In other words, this pattern matches a sequence of
2175 non-parentheses, optionally enclosed in parentheses.
2176 </P>
2177 <P>
2178 If you were embedding this pattern in a larger one, you could use a relative
2179 reference:
2180 <pre>
2181 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
2182 </pre>
2183 This makes the fragment independent of the parentheses in the larger pattern.
2184 </P>
2185 <br><b>
2186 Checking for a used subpattern by name
2187 </b><br>
2188 <P>
2189 Perl uses the syntax (?(&#60;name&#62;)...) or (?('name')...) to test for a used
2190 subpattern by name. For compatibility with earlier versions of PCRE, which had
2191 this facility before Perl, the syntax (?(name)...) is also recognized. However,
2192 there is a possible ambiguity with this syntax, because subpattern names may
2193 consist entirely of digits. PCRE looks first for a named subpattern; if it
2194 cannot find one and the name consists entirely of digits, PCRE looks for a
2195 subpattern of that number, which must be greater than zero. Using subpattern
2196 names that consist entirely of digits is not recommended.
2197 </P>
2198 <P>
2199 Rewriting the above example to use a named subpattern gives this:
2200 <pre>
2201 (?&#60;OPEN&#62; \( )? [^()]+ (?(&#60;OPEN&#62;) \) )
2202 </pre>
2203 If the name used in a condition of this kind is a duplicate, the test is
2204 applied to all subpatterns of the same name, and is true if any one of them has
2205 matched.
2206 </P>
2207 <br><b>
2208 Checking for pattern recursion
2209 </b><br>
2210 <P>
2211 If the condition is the string (R), and there is no subpattern with the name R,
2212 the condition is true if a recursive call to the whole pattern or any
2213 subpattern has been made. If digits or a name preceded by ampersand follow the
2214 letter R, for example:
2215 <pre>
2216 (?(R3)...) or (?(R&name)...)
2217 </pre>
2218 the condition is true if the most recent recursion is into a subpattern whose
2219 number or name is given. This condition does not check the entire recursion
2220 stack. If the name used in a condition of this kind is a duplicate, the test is
2221 applied to all subpatterns of the same name, and is true if any one of them is
2222 the most recent recursion.
2223 </P>
2224 <P>
2225 At "top level", all these recursion test conditions are false.
2226 <a href="#recursion">The syntax for recursive patterns</a>
2227 is described below.
2228 <a name="subdefine"></a></P>
2229 <br><b>
2230 Defining subpatterns for use by reference only
2231 </b><br>
2232 <P>
2233 If the condition is the string (DEFINE), and there is no subpattern with the
2234 name DEFINE, the condition is always false. In this case, there may be only one
2235 alternative in the subpattern. It is always skipped if control reaches this
2236 point in the pattern; the idea of DEFINE is that it can be used to define
2237 subroutines that can be referenced from elsewhere. (The use of
2238 <a href="#subpatternsassubroutines">subroutines</a>
2239 is described below.) For example, a pattern to match an IPv4 address such as
2240 "" could be written like this (ignore white space and line
2241 breaks):
2242 <pre>
2243 (?(DEFINE) (?&#60;byte&#62; 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
2244 \b (?&byte) (\.(?&byte)){3} \b
2245 </pre>
2246 The first part of the pattern is a DEFINE group inside which a another group
2247 named "byte" is defined. This matches an individual component of an IPv4
2248 address (a number less than 256). When matching takes place, this part of the
2249 pattern is skipped because DEFINE acts like a false condition. The rest of the
2250 pattern uses references to the named group to match the four dot-separated
2251 components of an IPv4 address, insisting on a word boundary at each end.
2252 </P>
2253 <br><b>
2254 Assertion conditions
2255 </b><br>
2256 <P>
2257 If the condition is not in any of the above formats, it must be an assertion.
2258 This may be a positive or negative lookahead or lookbehind assertion. Consider
2259 this pattern, again containing non-significant white space, and with the two
2260 alternatives on the second line:
2261 <pre>
2262 (?(?=[^a-z]*[a-z])
2263 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
2264 </pre>
2265 The condition is a positive lookahead assertion that matches an optional
2266 sequence of non-letters followed by a letter. In other words, it tests for the
2267 presence of at least one letter in the subject. If a letter is found, the
2268 subject is matched against the first alternative; otherwise it is matched
2269 against the second. This pattern matches strings in one of the two forms
2270 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2271 <a name="comments"></a></P>
2272 <br><a name="SEC21" href="#TOC1">COMMENTS</a><br>
2273 <P>
2274 There are two ways of including comments in patterns that are processed by
2275 PCRE. In both cases, the start of the comment must not be in a character class,
2276 nor in the middle of any other sequence of related characters such as (?: or a
2277 subpattern name or number. The characters that make up a comment play no part
2278 in the pattern matching.
2279 </P>
2280 <P>
2281 The sequence (?# marks the start of a comment that continues up to the next
2282 closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED
2283 option is set, an unescaped # character also introduces a comment, which in
2284 this case continues to immediately after the next newline character or
2285 character sequence in the pattern. Which characters are interpreted as newlines
2286 is controlled by the options passed to a compiling function or by a special
2287 sequence at the start of the pattern, as described in the section entitled
2288 <a href="#newlines">"Newline conventions"</a>
2289 above. Note that the end of this type of comment is a literal newline sequence
2290 in the pattern; escape sequences that happen to represent a newline do not
2291 count. For example, consider this pattern when PCRE_EXTENDED is set, and the
2292 default newline convention is in force:
2293 <pre>
2294 abc #comment \n still comment
2295 </pre>
2296 On encountering the # character, <b>pcre_compile()</b> skips along, looking for
2297 a newline in the pattern. The sequence \n is still literal at this stage, so
2298 it does not terminate the comment. Only an actual character with the code value
2299 0x0a (the default newline) does so.
2300 <a name="recursion"></a></P>
2301 <br><a name="SEC22" href="#TOC1">RECURSIVE PATTERNS</a><br>
2302 <P>
2303 Consider the problem of matching a string in parentheses, allowing for
2304 unlimited nested parentheses. Without the use of recursion, the best that can
2305 be done is to use a pattern that matches up to some fixed depth of nesting. It
2306 is not possible to handle an arbitrary nesting depth.
2307 </P>
2308 <P>
2309 For some time, Perl has provided a facility that allows regular expressions to
2310 recurse (amongst other things). It does this by interpolating Perl code in the
2311 expression at run time, and the code can refer to the expression itself. A Perl
2312 pattern using code interpolation to solve the parentheses problem can be
2313 created like this:
2314 <pre>
2315 $re = qr{\( (?: (?&#62;[^()]+) | (?p{$re}) )* \)}x;
2316 </pre>
2317 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2318 recursively to the pattern in which it appears.
2319 </P>
2320 <P>
2321 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2322 supports special syntax for recursion of the entire pattern, and also for
2323 individual subpattern recursion. After its introduction in PCRE and Python,
2324 this kind of recursion was subsequently introduced into Perl at release 5.10.
2325 </P>
2326 <P>
2327 A special item that consists of (? followed by a number greater than zero and a
2328 closing parenthesis is a recursive subroutine call of the subpattern of the
2329 given number, provided that it occurs inside that subpattern. (If not, it is a
2330 <a href="#subpatternsassubroutines">non-recursive subroutine</a>
2331 call, which is described in the next section.) The special item (?R) or (?0) is
2332 a recursive call of the entire regular expression.
2333 </P>
2334 <P>
2335 This PCRE pattern solves the nested parentheses problem (assume the
2336 PCRE_EXTENDED option is set so that white space is ignored):
2337 <pre>
2338 \( ( [^()]++ | (?R) )* \)
2339 </pre>
2340 First it matches an opening parenthesis. Then it matches any number of
2341 substrings which can either be a sequence of non-parentheses, or a recursive
2342 match of the pattern itself (that is, a correctly parenthesized substring).
2343 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2344 to avoid backtracking into sequences of non-parentheses.
2345 </P>
2346 <P>
2347 If this were part of a larger pattern, you would not want to recurse the entire
2348 pattern, so instead you could use this:
2349 <pre>
2350 ( \( ( [^()]++ | (?1) )* \) )
2351 </pre>
2352 We have put the pattern into parentheses, and caused the recursion to refer to
2353 them instead of the whole pattern.
2354 </P>
2355 <P>
2356 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2357 is made easier by the use of relative references. Instead of (?1) in the
2358 pattern above you can write (?-2) to refer to the second most recently opened
2359 parentheses preceding the recursion. In other words, a negative number counts
2360 capturing parentheses leftwards from the point at which it is encountered.
2361 </P>
2362 <P>
2363 It is also possible to refer to subsequently opened parentheses, by writing
2364 references such as (?+2). However, these cannot be recursive because the
2365 reference is not inside the parentheses that are referenced. They are always
2366 <a href="#subpatternsassubroutines">non-recursive subroutine</a>
2367 calls, as described in the next section.
2368 </P>
2369 <P>
2370 An alternative approach is to use named parentheses instead. The Perl syntax
2371 for this is (?&name); PCRE's earlier syntax (?P&#62;name) is also supported. We
2372 could rewrite the above example as follows:
2373 <pre>
2374 (?&#60;pn&#62; \( ( [^()]++ | (?&pn) )* \) )
2375 </pre>
2376 If there is more than one subpattern with the same name, the earliest one is
2377 used.
2378 </P>
2379 <P>
2380 This particular example pattern that we have been looking at contains nested
2381 unlimited repeats, and so the use of a possessive quantifier for matching
2382 strings of non-parentheses is important when applying the pattern to strings
2383 that do not match. For example, when this pattern is applied to
2384 <pre>
2385 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2386 </pre>
2387 it yields "no match" quickly. However, if a possessive quantifier is not used,
2388 the match runs for a very long time indeed because there are so many different
2389 ways the + and * repeats can carve up the subject, and all have to be tested
2390 before failure can be reported.
2391 </P>
2392 <P>
2393 At the end of a match, the values of capturing parentheses are those from
2394 the outermost level. If you want to obtain intermediate values, a callout
2395 function can be used (see below and the
2396 <a href="pcrecallout.html"><b>pcrecallout</b></a>
2397 documentation). If the pattern above is matched against
2398 <pre>
2399 (ab(cd)ef)
2400 </pre>
2401 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2402 the last value taken on at the top level. If a capturing subpattern is not
2403 matched at the top level, its final captured value is unset, even if it was
2404 (temporarily) set at a deeper level during the matching process.
2405 </P>
2406 <P>
2407 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2408 obtain extra memory to store data during a recursion, which it does by using
2409 <b>pcre_malloc</b>, freeing it via <b>pcre_free</b> afterwards. If no memory can
2410 be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2411 </P>
2412 <P>
2413 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2414 Consider this pattern, which matches text in angle brackets, allowing for
2415 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2416 recursing), whereas any characters are permitted at the outer level.
2417 <pre>
2418 &#60; (?: (?(R) \d++ | [^&#60;&#62;]*+) | (?R)) * &#62;
2419 </pre>
2420 In this pattern, (?(R) is the start of a conditional subpattern, with two
2421 different alternatives for the recursive and non-recursive cases. The (?R) item
2422 is the actual recursive call.
2423 <a name="recursiondifference"></a></P>
2424 <br><b>
2425 Differences in recursion processing between PCRE and Perl
2426 </b><br>
2427 <P>
2428 Recursion processing in PCRE differs from Perl in two important ways. In PCRE
2429 (like Python, but unlike Perl), a recursive subpattern call is always treated
2430 as an atomic group. That is, once it has matched some of the subject string, it
2431 is never re-entered, even if it contains untried alternatives and there is a
2432 subsequent matching failure. This can be illustrated by the following pattern,
2433 which purports to match a palindromic string that contains an odd number of
2434 characters (for example, "a", "aba", "abcba", "abcdcba"):
2435 <pre>
2436 ^(.|(.)(?1)\2)$
2437 </pre>
2438 The idea is that it either matches a single character, or two identical
2439 characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2440 it does not if the pattern is longer than three characters. Consider the
2441 subject string "abcba":
2442 </P>
2443 <P>
2444 At the top level, the first character is matched, but as it is not at the end
2445 of the string, the first alternative fails; the second alternative is taken
2446 and the recursion kicks in. The recursive call to subpattern 1 successfully
2447 matches the next character ("b"). (Note that the beginning and end of line
2448 tests are not part of the recursion).
2449 </P>
2450 <P>
2451 Back at the top level, the next character ("c") is compared with what
2452 subpattern 2 matched, which was "a". This fails. Because the recursion is
2453 treated as an atomic group, there are now no backtracking points, and so the
2454 entire match fails. (Perl is able, at this point, to re-enter the recursion and
2455 try the second alternative.) However, if the pattern is written with the
2456 alternatives in the other order, things are different:
2457 <pre>
2458 ^((.)(?1)\2|.)$
2459 </pre>
2460 This time, the recursing alternative is tried first, and continues to recurse
2461 until it runs out of characters, at which point the recursion fails. But this
2462 time we do have another alternative to try at the higher level. That is the big
2463 difference: in the previous case the remaining alternative is at a deeper
2464 recursion level, which PCRE cannot use.
2465 </P>
2466 <P>
2467 To change the pattern so that it matches all palindromic strings, not just
2468 those with an odd number of characters, it is tempting to change the pattern to
2469 this:
2470 <pre>
2471 ^((.)(?1)\2|.?)$
2472 </pre>
2473 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2474 deeper recursion has matched a single character, it cannot be entered again in
2475 order to match an empty string. The solution is to separate the two cases, and
2476 write out the odd and even cases as alternatives at the higher level:
2477 <pre>
2478 ^(?:((.)(?1)\2|)|((.)(?3)\4|.))
2479 </pre>
2480 If you want to match typical palindromic phrases, the pattern has to ignore all
2481 non-word characters, which can be done like this:
2482 <pre>
2483 ^\W*+(?:((.)\W*+(?1)\W*+\2|)|((.)\W*+(?3)\W*+\4|\W*+.\W*+))\W*+$
2484 </pre>
2485 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2486 man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2487 the use of the possessive quantifier *+ to avoid backtracking into sequences of
2488 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2489 more) to match typical phrases, and Perl takes so long that you think it has
2490 gone into a loop.
2491 </P>
2492 <P>
2493 <b>WARNING</b>: The palindrome-matching patterns above work only if the subject
2494 string does not start with a palindrome that is shorter than the entire string.
2495 For example, although "abcba" is correctly matched, if the subject is "ababa",
2496 PCRE finds the palindrome "aba" at the start, then fails at top level because
2497 the end of the string does not follow. Once again, it cannot jump back into the
2498 recursion to try other alternatives, so the entire match fails.
2499 </P>
2500 <P>
2501 The second way in which PCRE and Perl differ in their recursion processing is
2502 in the handling of captured values. In Perl, when a subpattern is called
2503 recursively or as a subpattern (see the next section), it has no access to any
2504 values that were captured outside the recursion, whereas in PCRE these values
2505 can be referenced. Consider this pattern:
2506 <pre>
2507 ^(.)(\1|a(?2))
2508 </pre>
2509 In PCRE, this pattern matches "bab". The first capturing parentheses match "b",
2510 then in the second group, when the back reference \1 fails to match "b", the
2511 second alternative matches "a" and then recurses. In the recursion, \1 does
2512 now match "b" and so the whole match succeeds. In Perl, the pattern fails to
2513 match because inside the recursive call \1 cannot access the externally set
2514 value.
2515 <a name="subpatternsassubroutines"></a></P>
2516 <br><a name="SEC23" href="#TOC1">SUBPATTERNS AS SUBROUTINES</a><br>
2517 <P>
2518 If the syntax for a recursive subpattern call (either by number or by
2519 name) is used outside the parentheses to which it refers, it operates like a
2520 subroutine in a programming language. The called subpattern may be defined
2521 before or after the reference. A numbered reference can be absolute or
2522 relative, as in these examples:
2523 <pre>
2524 (...(absolute)...)...(?2)...
2525 (...(relative)...)...(?-1)...
2526 (...(?+1)...(relative)...
2527 </pre>
2528 An earlier example pointed out that the pattern
2529 <pre>
2530 (sens|respons)e and \1ibility
2531 </pre>
2532 matches "sense and sensibility" and "response and responsibility", but not
2533 "sense and responsibility". If instead the pattern
2534 <pre>
2535 (sens|respons)e and (?1)ibility
2536 </pre>
2537 is used, it does match "sense and responsibility" as well as the other two
2538 strings. Another example is given in the discussion of DEFINE above.
2539 </P>
2540 <P>
2541 All subroutine calls, whether recursive or not, are always treated as atomic
2542 groups. That is, once a subroutine has matched some of the subject string, it
2543 is never re-entered, even if it contains untried alternatives and there is a
2544 subsequent matching failure. Any capturing parentheses that are set during the
2545 subroutine call revert to their previous values afterwards.
2546 </P>
2547 <P>
2548 Processing options such as case-independence are fixed when a subpattern is
2549 defined, so if it is used as a subroutine, such options cannot be changed for
2550 different calls. For example, consider this pattern:
2551 <pre>
2552 (abc)(?i:(?-1))
2553 </pre>
2554 It matches "abcabc". It does not match "abcABC" because the change of
2555 processing option does not affect the called subpattern.
2556 <a name="onigurumasubroutines"></a></P>
2557 <br><a name="SEC24" href="#TOC1">ONIGURUMA SUBROUTINE SYNTAX</a><br>
2558 <P>
2559 For compatibility with Oniguruma, the non-Perl syntax \g followed by a name or
2560 a number enclosed either in angle brackets or single quotes, is an alternative
2561 syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2562 are two of the examples used above, rewritten using this syntax:
2563 <pre>
2564 (?&#60;pn&#62; \( ( (?&#62;[^()]+) | \g&#60;pn&#62; )* \) )
2565 (sens|respons)e and \g'1'ibility
2566 </pre>
2567 PCRE supports an extension to Oniguruma: if a number is preceded by a
2568 plus or a minus sign it is taken as a relative reference. For example:
2569 <pre>
2570 (abc)(?i:\g&#60;-1&#62;)
2571 </pre>
2572 Note that \g{...} (Perl syntax) and \g&#60;...&#62; (Oniguruma syntax) are <i>not</i>
2573 synonymous. The former is a back reference; the latter is a subroutine call.
2574 </P>
2575 <br><a name="SEC25" href="#TOC1">CALLOUTS</a><br>
2576 <P>
2577 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2578 code to be obeyed in the middle of matching a regular expression. This makes it
2579 possible, amongst other things, to extract different substrings that match the
2580 same pair of parentheses when there is a repetition.
2581 </P>
2582 <P>
2583 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2584 code. The feature is called "callout". The caller of PCRE provides an external
2585 function by putting its entry point in the global variable <i>pcre_callout</i>
2586 (8-bit library) or <i>pcre[16|32]_callout</i> (16-bit or 32-bit library).
2587 By default, this variable contains NULL, which disables all calling out.
2588 </P>
2589 <P>
2590 Within a regular expression, (?C) indicates the points at which the external
2591 function is to be called. If you want to identify different callout points, you
2592 can put a number less than 256 after the letter C. The default value is zero.
2593 For example, this pattern has two callout points:
2594 <pre>
2595 (?C1)abc(?C2)def
2596 </pre>
2597 If the PCRE_AUTO_CALLOUT flag is passed to a compiling function, callouts are
2598 automatically installed before each item in the pattern. They are all numbered
2599 255.
2600 </P>
2601 <P>
2602 During matching, when PCRE reaches a callout point, the external function is
2603 called. It is provided with the number of the callout, the position in the
2604 pattern, and, optionally, one item of data originally supplied by the caller of
2605 the matching function. The callout function may cause matching to proceed, to
2606 backtrack, or to fail altogether. A complete description of the interface to
2607 the callout function is given in the
2608 <a href="pcrecallout.html"><b>pcrecallout</b></a>
2609 documentation.
2610 <a name="backtrackcontrol"></a></P>
2611 <br><a name="SEC26" href="#TOC1">BACKTRACKING CONTROL</a><br>
2612 <P>
2613 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2614 are described in the Perl documentation as "experimental and subject to change
2615 or removal in a future version of Perl". It goes on to say: "Their usage in
2616 production code should be noted to avoid problems during upgrades." The same
2617 remarks apply to the PCRE features described in this section.
2618 </P>
2619 <P>
2620 Since these verbs are specifically related to backtracking, most of them can be
2621 used only when the pattern is to be matched using one of the traditional
2622 matching functions, which use a backtracking algorithm. With the exception of
2623 (*FAIL), which behaves like a failing negative assertion, they cause an error
2624 if encountered by a DFA matching function.
2625 </P>
2626 <P>
2627 If any of these verbs are used in an assertion or in a subpattern that is
2628 called as a subroutine (whether or not recursively), their effect is confined
2629 to that subpattern; it does not extend to the surrounding pattern, with one
2630 exception: the name from a *(MARK), (*PRUNE), or (*THEN) that is encountered in
2631 a successful positive assertion <i>is</i> passed back when a match succeeds
2632 (compare capturing parentheses in assertions). Note that such subpatterns are
2633 processed as anchored at the point where they are tested. Note also that Perl's
2634 treatment of subroutines and assertions is different in some cases.
2635 </P>
2636 <P>
2637 The new verbs make use of what was previously invalid syntax: an opening
2638 parenthesis followed by an asterisk. They are generally of the form
2639 (*VERB) or (*VERB:NAME). Some may take either form, with differing behaviour,
2640 depending on whether or not an argument is present. A name is any sequence of
2641 characters that does not include a closing parenthesis. The maximum length of
2642 name is 255 in the 8-bit library and 65535 in the 16-bit and 32-bit library.
2643 If the name is empty, that is, if the closing parenthesis immediately follows
2644 the colon, the effect is as if the colon were not there. Any number of these
2645 verbs may occur in a pattern.
2646 <a name="nooptimize"></a></P>
2647 <br><b>
2648 Optimizations that affect backtracking verbs
2649 </b><br>
2650 <P>
2651 PCRE contains some optimizations that are used to speed up matching by running
2652 some checks at the start of each match attempt. For example, it may know the
2653 minimum length of matching subject, or that a particular character must be
2654 present. When one of these optimizations suppresses the running of a match, any
2655 included backtracking verbs will not, of course, be processed. You can suppress
2656 the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2657 when calling <b>pcre_compile()</b> or <b>pcre_exec()</b>, or by starting the
2658 pattern with (*NO_START_OPT). There is more discussion of this option in the
2659 section entitled
2660 <a href="pcreapi.html#execoptions">"Option bits for <b>pcre_exec()</b>"</a>
2661 in the
2662 <a href="pcreapi.html"><b>pcreapi</b></a>
2663 documentation.
2664 </P>
2665 <P>
2666 Experiments with Perl suggest that it too has similar optimizations, sometimes
2667 leading to anomalous results.
2668 </P>
2669 <br><b>
2670 Verbs that act immediately
2671 </b><br>
2672 <P>
2673 The following verbs act as soon as they are encountered. They may not be
2674 followed by a name.
2675 <pre>
2676 (*ACCEPT)
2677 </pre>
2678 This verb causes the match to end successfully, skipping the remainder of the
2679 pattern. However, when it is inside a subpattern that is called as a
2680 subroutine, only that subpattern is ended successfully. Matching then continues
2681 at the outer level. If (*ACCEPT) is inside capturing parentheses, the data so
2682 far is captured. For example:
2683 <pre>
2684 A((?:A|B(*ACCEPT)|C)D)
2685 </pre>
2686 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2687 the outer parentheses.
2688 <pre>
2689 (*FAIL) or (*F)
2690 </pre>
2691 This verb causes a matching failure, forcing backtracking to occur. It is
2692 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2693 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2694 Perl features that are not present in PCRE. The nearest equivalent is the
2695 callout feature, as for example in this pattern:
2696 <pre>
2697 a+(?C)(*FAIL)
2698 </pre>
2699 A match with the string "aaaa" always fails, but the callout is taken before
2700 each backtrack happens (in this example, 10 times).
2701 </P>
2702 <br><b>
2703 Recording which path was taken
2704 </b><br>
2705 <P>
2706 There is one verb whose main purpose is to track how a match was arrived at,
2707 though it also has a secondary use in conjunction with advancing the match
2708 starting point (see (*SKIP) below).
2709 <pre>
2710 (*MARK:NAME) or (*:NAME)
2711 </pre>
2712 A name is always required with this verb. There may be as many instances of
2713 (*MARK) as you like in a pattern, and their names do not have to be unique.
2714 </P>
2715 <P>
2716 When a match succeeds, the name of the last-encountered (*MARK) on the matching
2717 path is passed back to the caller as described in the section entitled
2718 <a href="pcreapi.html#extradata">"Extra data for <b>pcre_exec()</b>"</a>
2719 in the
2720 <a href="pcreapi.html"><b>pcreapi</b></a>
2721 documentation. Here is an example of <b>pcretest</b> output, where the /K
2722 modifier requests the retrieval and outputting of (*MARK) data:
2723 <pre>
2724 re&#62; /X(*MARK:A)Y|X(*MARK:B)Z/K
2725 data&#62; XY
2726 0: XY
2727 MK: A
2728 XZ
2729 0: XZ
2730 MK: B
2731 </pre>
2732 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2733 indicates which of the two alternatives matched. This is a more efficient way
2734 of obtaining this information than putting each alternative in its own
2735 capturing parentheses.
2736 </P>
2737 <P>
2738 If (*MARK) is encountered in a positive assertion, its name is recorded and
2739 passed back if it is the last-encountered. This does not happen for negative
2740 assertions.
2741 </P>
2742 <P>
2743 After a partial match or a failed match, the name of the last encountered
2744 (*MARK) in the entire match process is returned. For example:
2745 <pre>
2746 re&#62; /X(*MARK:A)Y|X(*MARK:B)Z/K
2747 data&#62; XP
2748 No match, mark = B
2749 </pre>
2750 Note that in this unanchored example the mark is retained from the match
2751 attempt that started at the letter "X" in the subject. Subsequent match
2752 attempts starting at "P" and then with an empty string do not get as far as the
2753 (*MARK) item, but nevertheless do not reset it.
2754 </P>
2755 <P>
2756 If you are interested in (*MARK) values after failed matches, you should
2757 probably set the PCRE_NO_START_OPTIMIZE option
2758 <a href="#nooptimize">(see above)</a>
2759 to ensure that the match is always attempted.
2760 </P>
2761 <br><b>
2762 Verbs that act after backtracking
2763 </b><br>
2764 <P>
2765 The following verbs do nothing when they are encountered. Matching continues
2766 with what follows, but if there is no subsequent match, causing a backtrack to
2767 the verb, a failure is forced. That is, backtracking cannot pass to the left of
2768 the verb. However, when one of these verbs appears inside an atomic group, its
2769 effect is confined to that group, because once the group has been matched,
2770 there is never any backtracking into it. In this situation, backtracking can
2771 "jump back" to the left of the entire atomic group. (Remember also, as stated
2772 above, that this localization also applies in subroutine calls and assertions.)
2773 </P>
2774 <P>
2775 These verbs differ in exactly what kind of failure occurs when backtracking
2776 reaches them.
2777 <pre>
2778 (*COMMIT)
2779 </pre>
2780 This verb, which may not be followed by a name, causes the whole match to fail
2781 outright if the rest of the pattern does not match. Even if the pattern is
2782 unanchored, no further attempts to find a match by advancing the starting point
2783 take place. Once (*COMMIT) has been passed, <b>pcre_exec()</b> is committed to
2784 finding a match at the current starting point, or not at all. For example:
2785 <pre>
2786 a+(*COMMIT)b
2787 </pre>
2788 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2789 dynamic anchor, or "I've started, so I must finish." The name of the most
2790 recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2791 match failure.
2792 </P>
2793 <P>
2794 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2795 unless PCRE's start-of-match optimizations are turned off, as shown in this
2796 <b>pcretest</b> example:
2797 <pre>
2798 re&#62; /(*COMMIT)abc/
2799 data&#62; xyzabc
2800 0: abc
2801 xyzabc\Y
2802 No match
2803 </pre>
2804 PCRE knows that any match must start with "a", so the optimization skips along
2805 the subject to "a" before running the first match attempt, which succeeds. When
2806 the optimization is disabled by the \Y escape in the second subject, the match
2807 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2808 starting points.
2809 <pre>
2810 (*PRUNE) or (*PRUNE:NAME)
2811 </pre>
2812 This verb causes the match to fail at the current starting position in the
2813 subject if the rest of the pattern does not match. If the pattern is
2814 unanchored, the normal "bumpalong" advance to the next starting character then
2815 happens. Backtracking can occur as usual to the left of (*PRUNE), before it is
2816 reached, or when matching to the right of (*PRUNE), but if there is no match to
2817 the right, backtracking cannot cross (*PRUNE). In simple cases, the use of
2818 (*PRUNE) is just an alternative to an atomic group or possessive quantifier,
2819 but there are some uses of (*PRUNE) that cannot be expressed in any other way.
2820 The behaviour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE). In an
2821 anchored pattern (*PRUNE) has the same effect as (*COMMIT).
2822 <pre>
2823 (*SKIP)
2824 </pre>
2825 This verb, when given without a name, is like (*PRUNE), except that if the
2826 pattern is unanchored, the "bumpalong" advance is not to the next character,
2827 but to the position in the subject where (*SKIP) was encountered. (*SKIP)
2828 signifies that whatever text was matched leading up to it cannot be part of a
2829 successful match. Consider:
2830 <pre>
2831 a+(*SKIP)b
2832 </pre>
2833 If the subject is "aaaac...", after the first match attempt fails (starting at
2834 the first character in the string), the starting point skips on to start the
2835 next attempt at "c". Note that a possessive quantifer does not have the same
2836 effect as this example; although it would suppress backtracking during the
2837 first match attempt, the second attempt would start at the second character
2838 instead of skipping on to "c".
2839 <pre>
2840 (*SKIP:NAME)
2841 </pre>
2842 When (*SKIP) has an associated name, its behaviour is modified. If the
2843 following pattern fails to match, the previous path through the pattern is
2844 searched for the most recent (*MARK) that has the same name. If one is found,
2845 the "bumpalong" advance is to the subject position that corresponds to that
2846 (*MARK) instead of to where (*SKIP) was encountered. If no (*MARK) with a
2847 matching name is found, the (*SKIP) is ignored.
2848 <pre>
2849 (*THEN) or (*THEN:NAME)
2850 </pre>
2851 This verb causes a skip to the next innermost alternative if the rest of the
2852 pattern does not match. That is, it cancels pending backtracking, but only
2853 within the current alternative. Its name comes from the observation that it can
2854 be used for a pattern-based if-then-else block:
2855 <pre>
2856 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2857 </pre>
2858 If the COND1 pattern matches, FOO is tried (and possibly further items after
2859 the end of the group if FOO succeeds); on failure, the matcher skips to the
2860 second alternative and tries COND2, without backtracking into COND1. The
2861 behaviour of (*THEN:NAME) is exactly the same as (*MARK:NAME)(*THEN).
2862 If (*THEN) is not inside an alternation, it acts like (*PRUNE).
2863 </P>
2864 <P>
2865 Note that a subpattern that does not contain a | character is just a part of
2866 the enclosing alternative; it is not a nested alternation with only one
2867 alternative. The effect of (*THEN) extends beyond such a subpattern to the
2868 enclosing alternative. Consider this pattern, where A, B, etc. are complex
2869 pattern fragments that do not contain any | characters at this level:
2870 <pre>
2871 A (B(*THEN)C) | D
2872 </pre>
2873 If A and B are matched, but there is a failure in C, matching does not
2874 backtrack into A; instead it moves to the next alternative, that is, D.
2875 However, if the subpattern containing (*THEN) is given an alternative, it
2876 behaves differently:
2877 <pre>
2878 A (B(*THEN)C | (*FAIL)) | D
2879 </pre>
2880 The effect of (*THEN) is now confined to the inner subpattern. After a failure
2881 in C, matching moves to (*FAIL), which causes the whole subpattern to fail
2882 because there are no more alternatives to try. In this case, matching does now
2883 backtrack into A.
2884 </P>
2885 <P>
2886 Note also that a conditional subpattern is not considered as having two
2887 alternatives, because only one is ever used. In other words, the | character in
2888 a conditional subpattern has a different meaning. Ignoring white space,
2889 consider:
2890 <pre>
2891 ^.*? (?(?=a) a | b(*THEN)c )
2892 </pre>
2893 If the subject is "ba", this pattern does not match. Because .*? is ungreedy,
2894 it initially matches zero characters. The condition (?=a) then fails, the
2895 character "b" is matched, but "c" is not. At this point, matching does not
2896 backtrack to .*? as might perhaps be expected from the presence of the |
2897 character. The conditional subpattern is part of the single alternative that
2898 comprises the whole pattern, and so the match fails. (If there was a backtrack
2899 into .*?, allowing it to match "b", the match would succeed.)
2900 </P>
2901 <P>
2902 The verbs just described provide four different "strengths" of control when
2903 subsequent matching fails. (*THEN) is the weakest, carrying on the match at the
2904 next alternative. (*PRUNE) comes next, failing the match at the current
2905 starting position, but allowing an advance to the next character (for an
2906 unanchored pattern). (*SKIP) is similar, except that the advance may be more
2907 than one character. (*COMMIT) is the strongest, causing the entire match to
2908 fail.
2909 </P>
2910 <P>
2911 If more than one such verb is present in a pattern, the "strongest" one wins.
2912 For example, consider this pattern, where A, B, etc. are complex pattern
2913 fragments:
2914 <pre>
2915 (A(*COMMIT)B(*THEN)C|D)
2916 </pre>
2917 Once A has matched, PCRE is committed to this match, at the current starting
2918 position. If subsequently B matches, but C does not, the normal (*THEN) action
2919 of trying the next alternative (that is, D) does not happen because (*COMMIT)
2920 overrides.
2921 </P>
2922 <br><a name="SEC27" href="#TOC1">SEE ALSO</a><br>
2923 <P>
2924 <b>pcreapi</b>(3), <b>pcrecallout</b>(3), <b>pcrematching</b>(3),
2925 <b>pcresyntax</b>(3), <b>pcre</b>(3), <b>pcre16(3)</b>, <b>pcre32(3)</b>.
2926 </P>
2927 <br><a name="SEC28" href="#TOC1">AUTHOR</a><br>
2928 <P>
2929 Philip Hazel
2930 <br>
2931 University Computing Service
2932 <br>
2933 Cambridge CB2 3QH, England.
2934 <br>
2935 </P>
2936 <br><a name="SEC29" href="#TOC1">REVISION</a><br>
2937 <P>
2938 Last updated: 10 September 2012
2939 <br>
2940 Copyright &copy; 1997-2012 University of Cambridge.
2941 <br>
2942 <p>
2943 Return to the <a href="index.html">PCRE index page</a>.
2944 </p>


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

  ViewVC Help
Powered by ViewVC 1.1.5