/[pcre]/code/trunk/doc/html/pcrepattern.html
ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 954 - (show annotations)
Sat Mar 31 18:09:26 2012 UTC (7 years, 4 months ago) by ph10
File MIME type: text/html
File size: 124458 byte(s)
Add date and PCRE version to .TH macros of all man pages.
1 <html>
2 <head>
3 <title>pcrepattern specification</title>
4 </head>
5 <body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB">
6 <h1>pcrepattern man page</h1>
7 <p>
8 Return to the <a href="index.html">PCRE index page</a>.
9 </p>
10 <p>
11 This page is part of the PCRE HTML documentation. It was generated automatically
12 from the original man page. If there is any nonsense in it, please consult the
13 man page, in case the conversion went wrong.
14 <br>
15 <ul>
16 <li><a name="TOC1" href="#SEC1">PCRE REGULAR EXPRESSION DETAILS</a>
17 <li><a name="TOC2" href="#SEC2">NEWLINE CONVENTIONS</a>
18 <li><a name="TOC3" href="#SEC3">CHARACTERS AND METACHARACTERS</a>
19 <li><a name="TOC4" href="#SEC4">BACKSLASH</a>
20 <li><a name="TOC5" href="#SEC5">CIRCUMFLEX AND DOLLAR</a>
21 <li><a name="TOC6" href="#SEC6">FULL STOP (PERIOD, DOT) AND \N</a>
22 <li><a name="TOC7" href="#SEC7">MATCHING A SINGLE DATA UNIT</a>
23 <li><a name="TOC8" href="#SEC8">SQUARE BRACKETS AND CHARACTER CLASSES</a>
24 <li><a name="TOC9" href="#SEC9">POSIX CHARACTER CLASSES</a>
25 <li><a name="TOC10" href="#SEC10">VERTICAL BAR</a>
26 <li><a name="TOC11" href="#SEC11">INTERNAL OPTION SETTING</a>
27 <li><a name="TOC12" href="#SEC12">SUBPATTERNS</a>
28 <li><a name="TOC13" href="#SEC13">DUPLICATE SUBPATTERN NUMBERS</a>
29 <li><a name="TOC14" href="#SEC14">NAMED SUBPATTERNS</a>
30 <li><a name="TOC15" href="#SEC15">REPETITION</a>
31 <li><a name="TOC16" href="#SEC16">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a>
32 <li><a name="TOC17" href="#SEC17">BACK REFERENCES</a>
33 <li><a name="TOC18" href="#SEC18">ASSERTIONS</a>
34 <li><a name="TOC19" href="#SEC19">CONDITIONAL SUBPATTERNS</a>
35 <li><a name="TOC20" href="#SEC20">COMMENTS</a>
36 <li><a name="TOC21" href="#SEC21">RECURSIVE PATTERNS</a>
37 <li><a name="TOC22" href="#SEC22">SUBPATTERNS AS SUBROUTINES</a>
38 <li><a name="TOC23" href="#SEC23">ONIGURUMA SUBROUTINE SYNTAX</a>
39 <li><a name="TOC24" href="#SEC24">CALLOUTS</a>
40 <li><a name="TOC25" href="#SEC25">BACKTRACKING CONTROL</a>
41 <li><a name="TOC26" href="#SEC26">SEE ALSO</a>
42 <li><a name="TOC27" href="#SEC27">AUTHOR</a>
43 <li><a name="TOC28" href="#SEC28">REVISION</a>
44 </ul>
45 <br><a name="SEC1" href="#TOC1">PCRE REGULAR EXPRESSION DETAILS</a><br>
46 <P>
47 The syntax and semantics of the regular expressions that are supported by PCRE
48 are described in detail below. There is a quick-reference syntax summary in the
49 <a href="pcresyntax.html"><b>pcresyntax</b></a>
50 page. PCRE tries to match Perl syntax and semantics as closely as it can. PCRE
51 also supports some alternative regular expression syntax (which does not
52 conflict with the Perl syntax) in order to provide some compatibility with
53 regular expressions in Python, .NET, and Oniguruma.
54 </P>
55 <P>
56 Perl's regular expressions are described in its own documentation, and
57 regular expressions in general are covered in a number of books, some of which
58 have copious examples. Jeffrey Friedl's "Mastering Regular Expressions",
59 published by O'Reilly, covers regular expressions in great detail. This
60 description of PCRE's regular expressions is intended as reference material.
61 </P>
62 <P>
63 The original operation of PCRE was on strings of one-byte characters. However,
64 there is now also support for UTF-8 strings in the original library, and a
65 second library that supports 16-bit and UTF-16 character strings. To use these
66 features, PCRE must be built to include appropriate support. When using UTF
67 strings you must either call the compiling function with the PCRE_UTF8 or
68 PCRE_UTF16 option, or the pattern must start with one of these special
69 sequences:
70 <pre>
71 (*UTF8)
72 (*UTF16)
73 </pre>
74 Starting a pattern with such a sequence is equivalent to setting the relevant
75 option. This feature is not Perl-compatible. How setting a UTF mode affects
76 pattern matching is mentioned in several places below. There is also a summary
77 of features in the
78 <a href="pcreunicode.html"><b>pcreunicode</b></a>
79 page.
80 </P>
81 <P>
82 Another special sequence that may appear at the start of a pattern or in
83 combination with (*UTF8) or (*UTF16) is:
84 <pre>
85 (*UCP)
86 </pre>
87 This has the same effect as setting the PCRE_UCP option: it causes sequences
88 such as \d and \w to use Unicode properties to determine character types,
89 instead of recognizing only characters with codes less than 128 via a lookup
90 table.
91 </P>
92 <P>
93 If a pattern starts with (*NO_START_OPT), it has the same effect as setting the
94 PCRE_NO_START_OPTIMIZE option either at compile or matching time. There are
95 also some more of these special sequences that are concerned with the handling
96 of newlines; they are described below.
97 </P>
98 <P>
99 The remainder of this document discusses the patterns that are supported by
100 PCRE when one its main matching functions, <b>pcre_exec()</b> (8-bit) or
101 <b>pcre16_exec()</b> (16-bit), is used. PCRE also has alternative matching
102 functions, <b>pcre_dfa_exec()</b> and <b>pcre16_dfa_exec()</b>, which match using
103 a different algorithm that is not Perl-compatible. Some of the features
104 discussed below are not available when DFA matching is used. The advantages and
105 disadvantages of the alternative functions, and how they differ from the normal
106 functions, are discussed in the
107 <a href="pcrematching.html"><b>pcrematching</b></a>
108 page.
109 <a name="newlines"></a></P>
110 <br><a name="SEC2" href="#TOC1">NEWLINE CONVENTIONS</a><br>
111 <P>
112 PCRE supports five different conventions for indicating line breaks in
113 strings: a single CR (carriage return) character, a single LF (linefeed)
114 character, the two-character sequence CRLF, any of the three preceding, or any
115 Unicode newline sequence. The
116 <a href="pcreapi.html"><b>pcreapi</b></a>
117 page has
118 <a href="pcreapi.html#newlines">further discussion</a>
119 about newlines, and shows how to set the newline convention in the
120 <i>options</i> arguments for the compiling and matching functions.
121 </P>
122 <P>
123 It is also possible to specify a newline convention by starting a pattern
124 string with one of the following five sequences:
125 <pre>
126 (*CR) carriage return
127 (*LF) linefeed
128 (*CRLF) carriage return, followed by linefeed
129 (*ANYCRLF) any of the three above
130 (*ANY) all Unicode newline sequences
131 </pre>
132 These override the default and the options given to the compiling function. For
133 example, on a Unix system where LF is the default newline sequence, the pattern
134 <pre>
135 (*CR)a.b
136 </pre>
137 changes the convention to CR. That pattern matches "a\nb" because LF is no
138 longer a newline. Note that these special settings, which are not
139 Perl-compatible, are recognized only at the very start of a pattern, and that
140 they must be in upper case. If more than one of them is present, the last one
141 is used.
142 </P>
143 <P>
144 The newline convention affects the interpretation of the dot metacharacter when
145 PCRE_DOTALL is not set, and also the behaviour of \N. However, it does not
146 affect what the \R escape sequence matches. By default, this is any Unicode
147 newline sequence, for Perl compatibility. However, this can be changed; see the
148 description of \R in the section entitled
149 <a href="#newlineseq">"Newline sequences"</a>
150 below. A change of \R setting can be combined with a change of newline
151 convention.
152 </P>
153 <br><a name="SEC3" href="#TOC1">CHARACTERS AND METACHARACTERS</a><br>
154 <P>
155 A regular expression is a pattern that is matched against a subject string from
156 left to right. Most characters stand for themselves in a pattern, and match the
157 corresponding characters in the subject. As a trivial example, the pattern
158 <pre>
159 The quick brown fox
160 </pre>
161 matches a portion of a subject string that is identical to itself. When
162 caseless matching is specified (the PCRE_CASELESS option), letters are matched
163 independently of case. In a UTF mode, PCRE always understands the concept of
164 case for characters whose values are less than 128, so caseless matching is
165 always possible. For characters with higher values, the concept of case is
166 supported if PCRE is compiled with Unicode property support, but not otherwise.
167 If you want to use caseless matching for characters 128 and above, you must
168 ensure that PCRE is compiled with Unicode property support as well as with
169 UTF support.
170 </P>
171 <P>
172 The power of regular expressions comes from the ability to include alternatives
173 and repetitions in the pattern. These are encoded in the pattern by the use of
174 <i>metacharacters</i>, which do not stand for themselves but instead are
175 interpreted in some special way.
176 </P>
177 <P>
178 There are two different sets of metacharacters: those that are recognized
179 anywhere in the pattern except within square brackets, and those that are
180 recognized within square brackets. Outside square brackets, the metacharacters
181 are as follows:
182 <pre>
183 \ general escape character with several uses
184 ^ assert start of string (or line, in multiline mode)
185 $ assert end of string (or line, in multiline mode)
186 . match any character except newline (by default)
187 [ start character class definition
188 | start of alternative branch
189 ( start subpattern
190 ) end subpattern
191 ? extends the meaning of (
192 also 0 or 1 quantifier
193 also quantifier minimizer
194 * 0 or more quantifier
195 + 1 or more quantifier
196 also "possessive quantifier"
197 { start min/max quantifier
198 </pre>
199 Part of a pattern that is in square brackets is called a "character class". In
200 a character class the only metacharacters are:
201 <pre>
202 \ general escape character
203 ^ negate the class, but only if the first character
204 - indicates character range
205 [ POSIX character class (only if followed by POSIX syntax)
206 ] terminates the character class
207 </pre>
208 The following sections describe the use of each of the metacharacters.
209 </P>
210 <br><a name="SEC4" href="#TOC1">BACKSLASH</a><br>
211 <P>
212 The backslash character has several uses. Firstly, if it is followed by a
213 character that is not a number or a letter, it takes away any special meaning
214 that character may have. This use of backslash as an escape character applies
215 both inside and outside character classes.
216 </P>
217 <P>
218 For example, if you want to match a * character, you write \* in the pattern.
219 This escaping action applies whether or not the following character would
220 otherwise be interpreted as a metacharacter, so it is always safe to precede a
221 non-alphanumeric with backslash to specify that it stands for itself. In
222 particular, if you want to match a backslash, you write \\.
223 </P>
224 <P>
225 In a UTF mode, only ASCII numbers and letters have any special meaning after a
226 backslash. All other characters (in particular, those whose codepoints are
227 greater than 127) are treated as literals.
228 </P>
229 <P>
230 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
231 pattern (other than in a character class) and characters between a # outside
232 a character class and the next newline are ignored. An escaping backslash can
233 be used to include a whitespace or # character as part of the pattern.
234 </P>
235 <P>
236 If you want to remove the special meaning from a sequence of characters, you
237 can do so by putting them between \Q and \E. This is different from Perl in
238 that $ and @ are handled as literals in \Q...\E sequences in PCRE, whereas in
239 Perl, $ and @ cause variable interpolation. Note the following examples:
240 <pre>
241 Pattern PCRE matches Perl matches
242
243 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
244 \Qabc\$xyz\E abc\$xyz abc\$xyz
245 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
246 </pre>
247 The \Q...\E sequence is recognized both inside and outside character classes.
248 An isolated \E that is not preceded by \Q is ignored. If \Q is not followed
249 by \E later in the pattern, the literal interpretation continues to the end of
250 the pattern (that is, \E is assumed at the end). If the isolated \Q is inside
251 a character class, this causes an error, because the character class is not
252 terminated.
253 <a name="digitsafterbackslash"></a></P>
254 <br><b>
255 Non-printing characters
256 </b><br>
257 <P>
258 A second use of backslash provides a way of encoding non-printing characters
259 in patterns in a visible manner. There is no restriction on the appearance of
260 non-printing characters, apart from the binary zero that terminates a pattern,
261 but when a pattern is being prepared by text editing, it is often easier to use
262 one of the following escape sequences than the binary character it represents:
263 <pre>
264 \a alarm, that is, the BEL character (hex 07)
265 \cx "control-x", where x is any ASCII character
266 \e escape (hex 1B)
267 \f formfeed (hex 0C)
268 \n linefeed (hex 0A)
269 \r carriage return (hex 0D)
270 \t tab (hex 09)
271 \ddd character with octal code ddd, or back reference
272 \xhh character with hex code hh
273 \x{hhh..} character with hex code hhh.. (non-JavaScript mode)
274 \uhhhh character with hex code hhhh (JavaScript mode only)
275 </pre>
276 The precise effect of \cx is as follows: if x is a lower case letter, it
277 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
278 Thus \cz becomes hex 1A (z is 7A), but \c{ becomes hex 3B ({ is 7B), while
279 \c; becomes hex 7B (; is 3B). If the byte following \c has a value greater
280 than 127, a compile-time error occurs. This locks out non-ASCII characters in
281 all modes. (When PCRE is compiled in EBCDIC mode, all byte values are valid. A
282 lower case letter is converted to upper case, and then the 0xc0 bits are
283 flipped.)
284 </P>
285 <P>
286 By default, after \x, from zero to two hexadecimal digits are read (letters
287 can be in upper or lower case). Any number of hexadecimal digits may appear
288 between \x{ and }, but the character code is constrained as follows:
289 <pre>
290 8-bit non-UTF mode less than 0x100
291 8-bit UTF-8 mode less than 0x10ffff and a valid codepoint
292 16-bit non-UTF mode less than 0x10000
293 16-bit UTF-16 mode less than 0x10ffff and a valid codepoint
294 </pre>
295 Invalid Unicode codepoints are the range 0xd800 to 0xdfff (the so-called
296 "surrogate" codepoints).
297 </P>
298 <P>
299 If characters other than hexadecimal digits appear between \x{ and }, or if
300 there is no terminating }, this form of escape is not recognized. Instead, the
301 initial \x will be interpreted as a basic hexadecimal escape, with no
302 following digits, giving a character whose value is zero.
303 </P>
304 <P>
305 If the PCRE_JAVASCRIPT_COMPAT option is set, the interpretation of \x is
306 as just described only when it is followed by two hexadecimal digits.
307 Otherwise, it matches a literal "x" character. In JavaScript mode, support for
308 code points greater than 256 is provided by \u, which must be followed by
309 four hexadecimal digits; otherwise it matches a literal "u" character.
310 </P>
311 <P>
312 Characters whose value is less than 256 can be defined by either of the two
313 syntaxes for \x (or by \u in JavaScript mode). There is no difference in the
314 way they are handled. For example, \xdc is exactly the same as \x{dc} (or
315 \u00dc in JavaScript mode).
316 </P>
317 <P>
318 After \0 up to two further octal digits are read. If there are fewer than two
319 digits, just those that are present are used. Thus the sequence \0\x\07
320 specifies two binary zeros followed by a BEL character (code value 7). Make
321 sure you supply two digits after the initial zero if the pattern character that
322 follows is itself an octal digit.
323 </P>
324 <P>
325 The handling of a backslash followed by a digit other than 0 is complicated.
326 Outside a character class, PCRE reads it and any following digits as a decimal
327 number. If the number is less than 10, or if there have been at least that many
328 previous capturing left parentheses in the expression, the entire sequence is
329 taken as a <i>back reference</i>. A description of how this works is given
330 <a href="#backreferences">later,</a>
331 following the discussion of
332 <a href="#subpattern">parenthesized subpatterns.</a>
333 </P>
334 <P>
335 Inside a character class, or if the decimal number is greater than 9 and there
336 have not been that many capturing subpatterns, PCRE re-reads up to three octal
337 digits following the backslash, and uses them to generate a data character. Any
338 subsequent digits stand for themselves. The value of the character is
339 constrained in the same way as characters specified in hexadecimal.
340 For example:
341 <pre>
342 \040 is another way of writing a space
343 \40 is the same, provided there are fewer than 40 previous capturing subpatterns
344 \7 is always a back reference
345 \11 might be a back reference, or another way of writing a tab
346 \011 is always a tab
347 \0113 is a tab followed by the character "3"
348 \113 might be a back reference, otherwise the character with octal code 113
349 \377 might be a back reference, otherwise the value 255 (decimal)
350 \81 is either a back reference, or a binary zero followed by the two characters "8" and "1"
351 </pre>
352 Note that octal values of 100 or greater must not be introduced by a leading
353 zero, because no more than three octal digits are ever read.
354 </P>
355 <P>
356 All the sequences that define a single character value can be used both inside
357 and outside character classes. In addition, inside a character class, \b is
358 interpreted as the backspace character (hex 08).
359 </P>
360 <P>
361 \N is not allowed in a character class. \B, \R, and \X are not special
362 inside a character class. Like other unrecognized escape sequences, they are
363 treated as the literal characters "B", "R", and "X" by default, but cause an
364 error if the PCRE_EXTRA option is set. Outside a character class, these
365 sequences have different meanings.
366 </P>
367 <br><b>
368 Unsupported escape sequences
369 </b><br>
370 <P>
371 In Perl, the sequences \l, \L, \u, and \U are recognized by its string
372 handler and used to modify the case of following characters. By default, PCRE
373 does not support these escape sequences. However, if the PCRE_JAVASCRIPT_COMPAT
374 option is set, \U matches a "U" character, and \u can be used to define a
375 character by code point, as described in the previous section.
376 </P>
377 <br><b>
378 Absolute and relative back references
379 </b><br>
380 <P>
381 The sequence \g followed by an unsigned or a negative number, optionally
382 enclosed in braces, is an absolute or relative back reference. A named back
383 reference can be coded as \g{name}. Back references are discussed
384 <a href="#backreferences">later,</a>
385 following the discussion of
386 <a href="#subpattern">parenthesized subpatterns.</a>
387 </P>
388 <br><b>
389 Absolute and relative subroutine calls
390 </b><br>
391 <P>
392 For compatibility with Oniguruma, the non-Perl syntax \g followed by a name or
393 a number enclosed either in angle brackets or single quotes, is an alternative
394 syntax for referencing a subpattern as a "subroutine". Details are discussed
395 <a href="#onigurumasubroutines">later.</a>
396 Note that \g{...} (Perl syntax) and \g&#60;...&#62; (Oniguruma syntax) are <i>not</i>
397 synonymous. The former is a back reference; the latter is a
398 <a href="#subpatternsassubroutines">subroutine</a>
399 call.
400 <a name="genericchartypes"></a></P>
401 <br><b>
402 Generic character types
403 </b><br>
404 <P>
405 Another use of backslash is for specifying generic character types:
406 <pre>
407 \d any decimal digit
408 \D any character that is not a decimal digit
409 \h any horizontal whitespace character
410 \H any character that is not a horizontal whitespace character
411 \s any whitespace character
412 \S any character that is not a whitespace character
413 \v any vertical whitespace character
414 \V any character that is not a vertical whitespace character
415 \w any "word" character
416 \W any "non-word" character
417 </pre>
418 There is also the single sequence \N, which matches a non-newline character.
419 This is the same as
420 <a href="#fullstopdot">the "." metacharacter</a>
421 when PCRE_DOTALL is not set. Perl also uses \N to match characters by name;
422 PCRE does not support this.
423 </P>
424 <P>
425 Each pair of lower and upper case escape sequences partitions the complete set
426 of characters into two disjoint sets. Any given character matches one, and only
427 one, of each pair. The sequences can appear both inside and outside character
428 classes. They each match one character of the appropriate type. If the current
429 matching point is at the end of the subject string, all of them fail, because
430 there is no character to match.
431 </P>
432 <P>
433 For compatibility with Perl, \s does not match the VT character (code 11).
434 This makes it different from the the POSIX "space" class. The \s characters
435 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
436 included in a Perl script, \s may match the VT character. In PCRE, it never
437 does.
438 </P>
439 <P>
440 A "word" character is an underscore or any character that is a letter or digit.
441 By default, the definition of letters and digits is controlled by PCRE's
442 low-valued character tables, and may vary if locale-specific matching is taking
443 place (see
444 <a href="pcreapi.html#localesupport">"Locale support"</a>
445 in the
446 <a href="pcreapi.html"><b>pcreapi</b></a>
447 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
448 or "french" in Windows, some character codes greater than 128 are used for
449 accented letters, and these are then matched by \w. The use of locales with
450 Unicode is discouraged.
451 </P>
452 <P>
453 By default, in a UTF mode, characters with values greater than 128 never match
454 \d, \s, or \w, and always match \D, \S, and \W. These sequences retain
455 their original meanings from before UTF support was available, mainly for
456 efficiency reasons. However, if PCRE is compiled with Unicode property support,
457 and the PCRE_UCP option is set, the behaviour is changed so that Unicode
458 properties are used to determine character types, as follows:
459 <pre>
460 \d any character that \p{Nd} matches (decimal digit)
461 \s any character that \p{Z} matches, plus HT, LF, FF, CR
462 \w any character that \p{L} or \p{N} matches, plus underscore
463 </pre>
464 The upper case escapes match the inverse sets of characters. Note that \d
465 matches only decimal digits, whereas \w matches any Unicode digit, as well as
466 any Unicode letter, and underscore. Note also that PCRE_UCP affects \b, and
467 \B because they are defined in terms of \w and \W. Matching these sequences
468 is noticeably slower when PCRE_UCP is set.
469 </P>
470 <P>
471 The sequences \h, \H, \v, and \V are features that were added to Perl at
472 release 5.10. In contrast to the other sequences, which match only ASCII
473 characters by default, these always match certain high-valued codepoints,
474 whether or not PCRE_UCP is set. The horizontal space characters are:
475 <pre>
476 U+0009 Horizontal tab
477 U+0020 Space
478 U+00A0 Non-break space
479 U+1680 Ogham space mark
480 U+180E Mongolian vowel separator
481 U+2000 En quad
482 U+2001 Em quad
483 U+2002 En space
484 U+2003 Em space
485 U+2004 Three-per-em space
486 U+2005 Four-per-em space
487 U+2006 Six-per-em space
488 U+2007 Figure space
489 U+2008 Punctuation space
490 U+2009 Thin space
491 U+200A Hair space
492 U+202F Narrow no-break space
493 U+205F Medium mathematical space
494 U+3000 Ideographic space
495 </pre>
496 The vertical space characters are:
497 <pre>
498 U+000A Linefeed
499 U+000B Vertical tab
500 U+000C Formfeed
501 U+000D Carriage return
502 U+0085 Next line
503 U+2028 Line separator
504 U+2029 Paragraph separator
505 </pre>
506 In 8-bit, non-UTF-8 mode, only the characters with codepoints less than 256 are
507 relevant.
508 <a name="newlineseq"></a></P>
509 <br><b>
510 Newline sequences
511 </b><br>
512 <P>
513 Outside a character class, by default, the escape sequence \R matches any
514 Unicode newline sequence. In 8-bit non-UTF-8 mode \R is equivalent to the
515 following:
516 <pre>
517 (?&#62;\r\n|\n|\x0b|\f|\r|\x85)
518 </pre>
519 This is an example of an "atomic group", details of which are given
520 <a href="#atomicgroup">below.</a>
521 This particular group matches either the two-character sequence CR followed by
522 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
523 U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
524 line, U+0085). The two-character sequence is treated as a single unit that
525 cannot be split.
526 </P>
527 <P>
528 In other modes, two additional characters whose codepoints are greater than 255
529 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
530 Unicode character property support is not needed for these characters to be
531 recognized.
532 </P>
533 <P>
534 It is possible to restrict \R to match only CR, LF, or CRLF (instead of the
535 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
536 either at compile time or when the pattern is matched. (BSR is an abbrevation
537 for "backslash R".) This can be made the default when PCRE is built; if this is
538 the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
539 It is also possible to specify these settings by starting a pattern string with
540 one of the following sequences:
541 <pre>
542 (*BSR_ANYCRLF) CR, LF, or CRLF only
543 (*BSR_UNICODE) any Unicode newline sequence
544 </pre>
545 These override the default and the options given to the compiling function, but
546 they can themselves be overridden by options given to a matching function. Note
547 that these special settings, which are not Perl-compatible, are recognized only
548 at the very start of a pattern, and that they must be in upper case. If more
549 than one of them is present, the last one is used. They can be combined with a
550 change of newline convention; for example, a pattern can start with:
551 <pre>
552 (*ANY)(*BSR_ANYCRLF)
553 </pre>
554 They can also be combined with the (*UTF8), (*UTF16), or (*UCP) special
555 sequences. Inside a character class, \R is treated as an unrecognized escape
556 sequence, and so matches the letter "R" by default, but causes an error if
557 PCRE_EXTRA is set.
558 <a name="uniextseq"></a></P>
559 <br><b>
560 Unicode character properties
561 </b><br>
562 <P>
563 When PCRE is built with Unicode character property support, three additional
564 escape sequences that match characters with specific properties are available.
565 When in 8-bit non-UTF-8 mode, these sequences are of course limited to testing
566 characters whose codepoints are less than 256, but they do work in this mode.
567 The extra escape sequences are:
568 <pre>
569 \p{<i>xx</i>} a character with the <i>xx</i> property
570 \P{<i>xx</i>} a character without the <i>xx</i> property
571 \X an extended Unicode sequence
572 </pre>
573 The property names represented by <i>xx</i> above are limited to the Unicode
574 script names, the general category properties, "Any", which matches any
575 character (including newline), and some special PCRE properties (described
576 in the
577 <a href="#extraprops">next section).</a>
578 Other Perl properties such as "InMusicalSymbols" are not currently supported by
579 PCRE. Note that \P{Any} does not match any characters, so always causes a
580 match failure.
581 </P>
582 <P>
583 Sets of Unicode characters are defined as belonging to certain scripts. A
584 character from one of these sets can be matched using a script name. For
585 example:
586 <pre>
587 \p{Greek}
588 \P{Han}
589 </pre>
590 Those that are not part of an identified script are lumped together as
591 "Common". The current list of scripts is:
592 </P>
593 <P>
594 Arabic,
595 Armenian,
596 Avestan,
597 Balinese,
598 Bamum,
599 Batak,
600 Bengali,
601 Bopomofo,
602 Brahmi,
603 Braille,
604 Buginese,
605 Buhid,
606 Canadian_Aboriginal,
607 Carian,
608 Chakma,
609 Cham,
610 Cherokee,
611 Common,
612 Coptic,
613 Cuneiform,
614 Cypriot,
615 Cyrillic,
616 Deseret,
617 Devanagari,
618 Egyptian_Hieroglyphs,
619 Ethiopic,
620 Georgian,
621 Glagolitic,
622 Gothic,
623 Greek,
624 Gujarati,
625 Gurmukhi,
626 Han,
627 Hangul,
628 Hanunoo,
629 Hebrew,
630 Hiragana,
631 Imperial_Aramaic,
632 Inherited,
633 Inscriptional_Pahlavi,
634 Inscriptional_Parthian,
635 Javanese,
636 Kaithi,
637 Kannada,
638 Katakana,
639 Kayah_Li,
640 Kharoshthi,
641 Khmer,
642 Lao,
643 Latin,
644 Lepcha,
645 Limbu,
646 Linear_B,
647 Lisu,
648 Lycian,
649 Lydian,
650 Malayalam,
651 Mandaic,
652 Meetei_Mayek,
653 Meroitic_Cursive,
654 Meroitic_Hieroglyphs,
655 Miao,
656 Mongolian,
657 Myanmar,
658 New_Tai_Lue,
659 Nko,
660 Ogham,
661 Old_Italic,
662 Old_Persian,
663 Old_South_Arabian,
664 Old_Turkic,
665 Ol_Chiki,
666 Oriya,
667 Osmanya,
668 Phags_Pa,
669 Phoenician,
670 Rejang,
671 Runic,
672 Samaritan,
673 Saurashtra,
674 Sharada,
675 Shavian,
676 Sinhala,
677 Sora_Sompeng,
678 Sundanese,
679 Syloti_Nagri,
680 Syriac,
681 Tagalog,
682 Tagbanwa,
683 Tai_Le,
684 Tai_Tham,
685 Tai_Viet,
686 Takri,
687 Tamil,
688 Telugu,
689 Thaana,
690 Thai,
691 Tibetan,
692 Tifinagh,
693 Ugaritic,
694 Vai,
695 Yi.
696 </P>
697 <P>
698 Each character has exactly one Unicode general category property, specified by
699 a two-letter abbreviation. For compatibility with Perl, negation can be
700 specified by including a circumflex between the opening brace and the property
701 name. For example, \p{^Lu} is the same as \P{Lu}.
702 </P>
703 <P>
704 If only one letter is specified with \p or \P, it includes all the general
705 category properties that start with that letter. In this case, in the absence
706 of negation, the curly brackets in the escape sequence are optional; these two
707 examples have the same effect:
708 <pre>
709 \p{L}
710 \pL
711 </pre>
712 The following general category property codes are supported:
713 <pre>
714 C Other
715 Cc Control
716 Cf Format
717 Cn Unassigned
718 Co Private use
719 Cs Surrogate
720
721 L Letter
722 Ll Lower case letter
723 Lm Modifier letter
724 Lo Other letter
725 Lt Title case letter
726 Lu Upper case letter
727
728 M Mark
729 Mc Spacing mark
730 Me Enclosing mark
731 Mn Non-spacing mark
732
733 N Number
734 Nd Decimal number
735 Nl Letter number
736 No Other number
737
738 P Punctuation
739 Pc Connector punctuation
740 Pd Dash punctuation
741 Pe Close punctuation
742 Pf Final punctuation
743 Pi Initial punctuation
744 Po Other punctuation
745 Ps Open punctuation
746
747 S Symbol
748 Sc Currency symbol
749 Sk Modifier symbol
750 Sm Mathematical symbol
751 So Other symbol
752
753 Z Separator
754 Zl Line separator
755 Zp Paragraph separator
756 Zs Space separator
757 </pre>
758 The special property L& is also supported: it matches a character that has
759 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
760 a modifier or "other".
761 </P>
762 <P>
763 The Cs (Surrogate) property applies only to characters in the range U+D800 to
764 U+DFFF. Such characters are not valid in Unicode strings and so
765 cannot be tested by PCRE, unless UTF validity checking has been turned off
766 (see the discussion of PCRE_NO_UTF8_CHECK and PCRE_NO_UTF16_CHECK in the
767 <a href="pcreapi.html"><b>pcreapi</b></a>
768 page). Perl does not support the Cs property.
769 </P>
770 <P>
771 The long synonyms for property names that Perl supports (such as \p{Letter})
772 are not supported by PCRE, nor is it permitted to prefix any of these
773 properties with "Is".
774 </P>
775 <P>
776 No character that is in the Unicode table has the Cn (unassigned) property.
777 Instead, this property is assumed for any code point that is not in the
778 Unicode table.
779 </P>
780 <P>
781 Specifying caseless matching does not affect these escape sequences. For
782 example, \p{Lu} always matches only upper case letters.
783 </P>
784 <P>
785 The \X escape matches any number of Unicode characters that form an extended
786 Unicode sequence. \X is equivalent to
787 <pre>
788 (?&#62;\PM\pM*)
789 </pre>
790 That is, it matches a character without the "mark" property, followed by zero
791 or more characters with the "mark" property, and treats the sequence as an
792 atomic group
793 <a href="#atomicgroup">(see below).</a>
794 Characters with the "mark" property are typically accents that affect the
795 preceding character. None of them have codepoints less than 256, so in
796 8-bit non-UTF-8 mode \X matches any one character.
797 </P>
798 <P>
799 Note that recent versions of Perl have changed \X to match what Unicode calls
800 an "extended grapheme cluster", which has a more complicated definition.
801 </P>
802 <P>
803 Matching characters by Unicode property is not fast, because PCRE has to search
804 a structure that contains data for over fifteen thousand characters. That is
805 why the traditional escape sequences such as \d and \w do not use Unicode
806 properties in PCRE by default, though you can make them do so by setting the
807 PCRE_UCP option or by starting the pattern with (*UCP).
808 <a name="extraprops"></a></P>
809 <br><b>
810 PCRE's additional properties
811 </b><br>
812 <P>
813 As well as the standard Unicode properties described in the previous
814 section, PCRE supports four more that make it possible to convert traditional
815 escape sequences such as \w and \s and POSIX character classes to use Unicode
816 properties. PCRE uses these non-standard, non-Perl properties internally when
817 PCRE_UCP is set. They are:
818 <pre>
819 Xan Any alphanumeric character
820 Xps Any POSIX space character
821 Xsp Any Perl space character
822 Xwd Any Perl "word" character
823 </pre>
824 Xan matches characters that have either the L (letter) or the N (number)
825 property. Xps matches the characters tab, linefeed, vertical tab, formfeed, or
826 carriage return, and any other character that has the Z (separator) property.
827 Xsp is the same as Xps, except that vertical tab is excluded. Xwd matches the
828 same characters as Xan, plus underscore.
829 <a name="resetmatchstart"></a></P>
830 <br><b>
831 Resetting the match start
832 </b><br>
833 <P>
834 The escape sequence \K causes any previously matched characters not to be
835 included in the final matched sequence. For example, the pattern:
836 <pre>
837 foo\Kbar
838 </pre>
839 matches "foobar", but reports that it has matched "bar". This feature is
840 similar to a lookbehind assertion
841 <a href="#lookbehind">(described below).</a>
842 However, in this case, the part of the subject before the real match does not
843 have to be of fixed length, as lookbehind assertions do. The use of \K does
844 not interfere with the setting of
845 <a href="#subpattern">captured substrings.</a>
846 For example, when the pattern
847 <pre>
848 (foo)\Kbar
849 </pre>
850 matches "foobar", the first substring is still set to "foo".
851 </P>
852 <P>
853 Perl documents that the use of \K within assertions is "not well defined". In
854 PCRE, \K is acted upon when it occurs inside positive assertions, but is
855 ignored in negative assertions.
856 <a name="smallassertions"></a></P>
857 <br><b>
858 Simple assertions
859 </b><br>
860 <P>
861 The final use of backslash is for certain simple assertions. An assertion
862 specifies a condition that has to be met at a particular point in a match,
863 without consuming any characters from the subject string. The use of
864 subpatterns for more complicated assertions is described
865 <a href="#bigassertions">below.</a>
866 The backslashed assertions are:
867 <pre>
868 \b matches at a word boundary
869 \B matches when not at a word boundary
870 \A matches at the start of the subject
871 \Z matches at the end of the subject
872 also matches before a newline at the end of the subject
873 \z matches only at the end of the subject
874 \G matches at the first matching position in the subject
875 </pre>
876 Inside a character class, \b has a different meaning; it matches the backspace
877 character. If any other of these assertions appears in a character class, by
878 default it matches the corresponding literal character (for example, \B
879 matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
880 escape sequence" error is generated instead.
881 </P>
882 <P>
883 A word boundary is a position in the subject string where the current character
884 and the previous character do not both match \w or \W (i.e. one matches
885 \w and the other matches \W), or the start or end of the string if the
886 first or last character matches \w, respectively. In a UTF mode, the meanings
887 of \w and \W can be changed by setting the PCRE_UCP option. When this is
888 done, it also affects \b and \B. Neither PCRE nor Perl has a separate "start
889 of word" or "end of word" metasequence. However, whatever follows \b normally
890 determines which it is. For example, the fragment \ba matches "a" at the start
891 of a word.
892 </P>
893 <P>
894 The \A, \Z, and \z assertions differ from the traditional circumflex and
895 dollar (described in the next section) in that they only ever match at the very
896 start and end of the subject string, whatever options are set. Thus, they are
897 independent of multiline mode. These three assertions are not affected by the
898 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
899 circumflex and dollar metacharacters. However, if the <i>startoffset</i>
900 argument of <b>pcre_exec()</b> is non-zero, indicating that matching is to start
901 at a point other than the beginning of the subject, \A can never match. The
902 difference between \Z and \z is that \Z matches before a newline at the end
903 of the string as well as at the very end, whereas \z matches only at the end.
904 </P>
905 <P>
906 The \G assertion is true only when the current matching position is at the
907 start point of the match, as specified by the <i>startoffset</i> argument of
908 <b>pcre_exec()</b>. It differs from \A when the value of <i>startoffset</i> is
909 non-zero. By calling <b>pcre_exec()</b> multiple times with appropriate
910 arguments, you can mimic Perl's /g option, and it is in this kind of
911 implementation where \G can be useful.
912 </P>
913 <P>
914 Note, however, that PCRE's interpretation of \G, as the start of the current
915 match, is subtly different from Perl's, which defines it as the end of the
916 previous match. In Perl, these can be different when the previously matched
917 string was empty. Because PCRE does just one match at a time, it cannot
918 reproduce this behaviour.
919 </P>
920 <P>
921 If all the alternatives of a pattern begin with \G, the expression is anchored
922 to the starting match position, and the "anchored" flag is set in the compiled
923 regular expression.
924 </P>
925 <br><a name="SEC5" href="#TOC1">CIRCUMFLEX AND DOLLAR</a><br>
926 <P>
927 Outside a character class, in the default matching mode, the circumflex
928 character is an assertion that is true only if the current matching point is
929 at the start of the subject string. If the <i>startoffset</i> argument of
930 <b>pcre_exec()</b> is non-zero, circumflex can never match if the PCRE_MULTILINE
931 option is unset. Inside a character class, circumflex has an entirely different
932 meaning
933 <a href="#characterclass">(see below).</a>
934 </P>
935 <P>
936 Circumflex need not be the first character of the pattern if a number of
937 alternatives are involved, but it should be the first thing in each alternative
938 in which it appears if the pattern is ever to match that branch. If all
939 possible alternatives start with a circumflex, that is, if the pattern is
940 constrained to match only at the start of the subject, it is said to be an
941 "anchored" pattern. (There are also other constructs that can cause a pattern
942 to be anchored.)
943 </P>
944 <P>
945 A dollar character is an assertion that is true only if the current matching
946 point is at the end of the subject string, or immediately before a newline
947 at the end of the string (by default). Dollar need not be the last character of
948 the pattern if a number of alternatives are involved, but it should be the last
949 item in any branch in which it appears. Dollar has no special meaning in a
950 character class.
951 </P>
952 <P>
953 The meaning of dollar can be changed so that it matches only at the very end of
954 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
955 does not affect the \Z assertion.
956 </P>
957 <P>
958 The meanings of the circumflex and dollar characters are changed if the
959 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
960 immediately after internal newlines as well as at the start of the subject
961 string. It does not match after a newline that ends the string. A dollar
962 matches before any newlines in the string, as well as at the very end, when
963 PCRE_MULTILINE is set. When newline is specified as the two-character
964 sequence CRLF, isolated CR and LF characters do not indicate newlines.
965 </P>
966 <P>
967 For example, the pattern /^abc$/ matches the subject string "def\nabc" (where
968 \n represents a newline) in multiline mode, but not otherwise. Consequently,
969 patterns that are anchored in single line mode because all branches start with
970 ^ are not anchored in multiline mode, and a match for circumflex is possible
971 when the <i>startoffset</i> argument of <b>pcre_exec()</b> is non-zero. The
972 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
973 </P>
974 <P>
975 Note that the sequences \A, \Z, and \z can be used to match the start and
976 end of the subject in both modes, and if all branches of a pattern start with
977 \A it is always anchored, whether or not PCRE_MULTILINE is set.
978 <a name="fullstopdot"></a></P>
979 <br><a name="SEC6" href="#TOC1">FULL STOP (PERIOD, DOT) AND \N</a><br>
980 <P>
981 Outside a character class, a dot in the pattern matches any one character in
982 the subject string except (by default) a character that signifies the end of a
983 line.
984 </P>
985 <P>
986 When a line ending is defined as a single character, dot never matches that
987 character; when the two-character sequence CRLF is used, dot does not match CR
988 if it is immediately followed by LF, but otherwise it matches all characters
989 (including isolated CRs and LFs). When any Unicode line endings are being
990 recognized, dot does not match CR or LF or any of the other line ending
991 characters.
992 </P>
993 <P>
994 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
995 option is set, a dot matches any one character, without exception. If the
996 two-character sequence CRLF is present in the subject string, it takes two dots
997 to match it.
998 </P>
999 <P>
1000 The handling of dot is entirely independent of the handling of circumflex and
1001 dollar, the only relationship being that they both involve newlines. Dot has no
1002 special meaning in a character class.
1003 </P>
1004 <P>
1005 The escape sequence \N behaves like a dot, except that it is not affected by
1006 the PCRE_DOTALL option. In other words, it matches any character except one
1007 that signifies the end of a line. Perl also uses \N to match characters by
1008 name; PCRE does not support this.
1009 </P>
1010 <br><a name="SEC7" href="#TOC1">MATCHING A SINGLE DATA UNIT</a><br>
1011 <P>
1012 Outside a character class, the escape sequence \C matches any one data unit,
1013 whether or not a UTF mode is set. In the 8-bit library, one data unit is one
1014 byte; in the 16-bit library it is a 16-bit unit. Unlike a dot, \C always
1015 matches line-ending characters. The feature is provided in Perl in order to
1016 match individual bytes in UTF-8 mode, but it is unclear how it can usefully be
1017 used. Because \C breaks up characters into individual data units, matching one
1018 unit with \C in a UTF mode means that the rest of the string may start with a
1019 malformed UTF character. This has undefined results, because PCRE assumes that
1020 it is dealing with valid UTF strings (and by default it checks this at the
1021 start of processing unless the PCRE_NO_UTF8_CHECK option is used).
1022 </P>
1023 <P>
1024 PCRE does not allow \C to appear in lookbehind assertions
1025 <a href="#lookbehind">(described below)</a>
1026 in a UTF mode, because this would make it impossible to calculate the length of
1027 the lookbehind.
1028 </P>
1029 <P>
1030 In general, the \C escape sequence is best avoided. However, one
1031 way of using it that avoids the problem of malformed UTF characters is to use a
1032 lookahead to check the length of the next character, as in this pattern, which
1033 could be used with a UTF-8 string (ignore white space and line breaks):
1034 <pre>
1035 (?| (?=[\x00-\x7f])(\C) |
1036 (?=[\x80-\x{7ff}])(\C)(\C) |
1037 (?=[\x{800}-\x{ffff}])(\C)(\C)(\C) |
1038 (?=[\x{10000}-\x{1fffff}])(\C)(\C)(\C)(\C))
1039 </pre>
1040 A group that starts with (?| resets the capturing parentheses numbers in each
1041 alternative (see
1042 <a href="#dupsubpatternnumber">"Duplicate Subpattern Numbers"</a>
1043 below). The assertions at the start of each branch check the next UTF-8
1044 character for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The
1045 character's individual bytes are then captured by the appropriate number of
1046 groups.
1047 <a name="characterclass"></a></P>
1048 <br><a name="SEC8" href="#TOC1">SQUARE BRACKETS AND CHARACTER CLASSES</a><br>
1049 <P>
1050 An opening square bracket introduces a character class, terminated by a closing
1051 square bracket. A closing square bracket on its own is not special by default.
1052 However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
1053 bracket causes a compile-time error. If a closing square bracket is required as
1054 a member of the class, it should be the first data character in the class
1055 (after an initial circumflex, if present) or escaped with a backslash.
1056 </P>
1057 <P>
1058 A character class matches a single character in the subject. In a UTF mode, the
1059 character may be more than one data unit long. A matched character must be in
1060 the set of characters defined by the class, unless the first character in the
1061 class definition is a circumflex, in which case the subject character must not
1062 be in the set defined by the class. If a circumflex is actually required as a
1063 member of the class, ensure it is not the first character, or escape it with a
1064 backslash.
1065 </P>
1066 <P>
1067 For example, the character class [aeiou] matches any lower case vowel, while
1068 [^aeiou] matches any character that is not a lower case vowel. Note that a
1069 circumflex is just a convenient notation for specifying the characters that
1070 are in the class by enumerating those that are not. A class that starts with a
1071 circumflex is not an assertion; it still consumes a character from the subject
1072 string, and therefore it fails if the current pointer is at the end of the
1073 string.
1074 </P>
1075 <P>
1076 In UTF-8 (UTF-16) mode, characters with values greater than 255 (0xffff) can be
1077 included in a class as a literal string of data units, or by using the \x{
1078 escaping mechanism.
1079 </P>
1080 <P>
1081 When caseless matching is set, any letters in a class represent both their
1082 upper case and lower case versions, so for example, a caseless [aeiou] matches
1083 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1084 caseful version would. In a UTF mode, PCRE always understands the concept of
1085 case for characters whose values are less than 128, so caseless matching is
1086 always possible. For characters with higher values, the concept of case is
1087 supported if PCRE is compiled with Unicode property support, but not otherwise.
1088 If you want to use caseless matching in a UTF mode for characters 128 and
1089 above, you must ensure that PCRE is compiled with Unicode property support as
1090 well as with UTF support.
1091 </P>
1092 <P>
1093 Characters that might indicate line breaks are never treated in any special way
1094 when matching character classes, whatever line-ending sequence is in use, and
1095 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
1096 such as [^a] always matches one of these characters.
1097 </P>
1098 <P>
1099 The minus (hyphen) character can be used to specify a range of characters in a
1100 character class. For example, [d-m] matches any letter between d and m,
1101 inclusive. If a minus character is required in a class, it must be escaped with
1102 a backslash or appear in a position where it cannot be interpreted as
1103 indicating a range, typically as the first or last character in the class.
1104 </P>
1105 <P>
1106 It is not possible to have the literal character "]" as the end character of a
1107 range. A pattern such as [W-]46] is interpreted as a class of two characters
1108 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1109 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1110 the end of range, so [W-\]46] is interpreted as a class containing a range
1111 followed by two other characters. The octal or hexadecimal representation of
1112 "]" can also be used to end a range.
1113 </P>
1114 <P>
1115 Ranges operate in the collating sequence of character values. They can also be
1116 used for characters specified numerically, for example [\000-\037]. Ranges
1117 can include any characters that are valid for the current mode.
1118 </P>
1119 <P>
1120 If a range that includes letters is used when caseless matching is set, it
1121 matches the letters in either case. For example, [W-c] is equivalent to
1122 [][\\^_`wxyzabc], matched caselessly, and in a non-UTF mode, if character
1123 tables for a French locale are in use, [\xc8-\xcb] matches accented E
1124 characters in both cases. In UTF modes, PCRE supports the concept of case for
1125 characters with values greater than 128 only when it is compiled with Unicode
1126 property support.
1127 </P>
1128 <P>
1129 The character escape sequences \d, \D, \h, \H, \p, \P, \s, \S, \v,
1130 \V, \w, and \W may appear in a character class, and add the characters that
1131 they match to the class. For example, [\dABCDEF] matches any hexadecimal
1132 digit. In UTF modes, the PCRE_UCP option affects the meanings of \d, \s, \w
1133 and their upper case partners, just as it does when they appear outside a
1134 character class, as described in the section entitled
1135 <a href="#genericchartypes">"Generic character types"</a>
1136 above. The escape sequence \b has a different meaning inside a character
1137 class; it matches the backspace character. The sequences \B, \N, \R, and \X
1138 are not special inside a character class. Like any other unrecognized escape
1139 sequences, they are treated as the literal characters "B", "N", "R", and "X" by
1140 default, but cause an error if the PCRE_EXTRA option is set.
1141 </P>
1142 <P>
1143 A circumflex can conveniently be used with the upper case character types to
1144 specify a more restricted set of characters than the matching lower case type.
1145 For example, the class [^\W_] matches any letter or digit, but not underscore,
1146 whereas [\w] includes underscore. A positive character class should be read as
1147 "something OR something OR ..." and a negative class as "NOT something AND NOT
1148 something AND NOT ...".
1149 </P>
1150 <P>
1151 The only metacharacters that are recognized in character classes are backslash,
1152 hyphen (only where it can be interpreted as specifying a range), circumflex
1153 (only at the start), opening square bracket (only when it can be interpreted as
1154 introducing a POSIX class name - see the next section), and the terminating
1155 closing square bracket. However, escaping other non-alphanumeric characters
1156 does no harm.
1157 </P>
1158 <br><a name="SEC9" href="#TOC1">POSIX CHARACTER CLASSES</a><br>
1159 <P>
1160 Perl supports the POSIX notation for character classes. This uses names
1161 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1162 this notation. For example,
1163 <pre>
1164 [01[:alpha:]%]
1165 </pre>
1166 matches "0", "1", any alphabetic character, or "%". The supported class names
1167 are:
1168 <pre>
1169 alnum letters and digits
1170 alpha letters
1171 ascii character codes 0 - 127
1172 blank space or tab only
1173 cntrl control characters
1174 digit decimal digits (same as \d)
1175 graph printing characters, excluding space
1176 lower lower case letters
1177 print printing characters, including space
1178 punct printing characters, excluding letters and digits and space
1179 space white space (not quite the same as \s)
1180 upper upper case letters
1181 word "word" characters (same as \w)
1182 xdigit hexadecimal digits
1183 </pre>
1184 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1185 space (32). Notice that this list includes the VT character (code 11). This
1186 makes "space" different to \s, which does not include VT (for Perl
1187 compatibility).
1188 </P>
1189 <P>
1190 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1191 5.8. Another Perl extension is negation, which is indicated by a ^ character
1192 after the colon. For example,
1193 <pre>
1194 [12[:^digit:]]
1195 </pre>
1196 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1197 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1198 supported, and an error is given if they are encountered.
1199 </P>
1200 <P>
1201 By default, in UTF modes, characters with values greater than 128 do not match
1202 any of the POSIX character classes. However, if the PCRE_UCP option is passed
1203 to <b>pcre_compile()</b>, some of the classes are changed so that Unicode
1204 character properties are used. This is achieved by replacing the POSIX classes
1205 by other sequences, as follows:
1206 <pre>
1207 [:alnum:] becomes \p{Xan}
1208 [:alpha:] becomes \p{L}
1209 [:blank:] becomes \h
1210 [:digit:] becomes \p{Nd}
1211 [:lower:] becomes \p{Ll}
1212 [:space:] becomes \p{Xps}
1213 [:upper:] becomes \p{Lu}
1214 [:word:] becomes \p{Xwd}
1215 </pre>
1216 Negated versions, such as [:^alpha:] use \P instead of \p. The other POSIX
1217 classes are unchanged, and match only characters with code points less than
1218 128.
1219 </P>
1220 <br><a name="SEC10" href="#TOC1">VERTICAL BAR</a><br>
1221 <P>
1222 Vertical bar characters are used to separate alternative patterns. For example,
1223 the pattern
1224 <pre>
1225 gilbert|sullivan
1226 </pre>
1227 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1228 and an empty alternative is permitted (matching the empty string). The matching
1229 process tries each alternative in turn, from left to right, and the first one
1230 that succeeds is used. If the alternatives are within a subpattern
1231 <a href="#subpattern">(defined below),</a>
1232 "succeeds" means matching the rest of the main pattern as well as the
1233 alternative in the subpattern.
1234 </P>
1235 <br><a name="SEC11" href="#TOC1">INTERNAL OPTION SETTING</a><br>
1236 <P>
1237 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1238 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1239 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1240 The option letters are
1241 <pre>
1242 i for PCRE_CASELESS
1243 m for PCRE_MULTILINE
1244 s for PCRE_DOTALL
1245 x for PCRE_EXTENDED
1246 </pre>
1247 For example, (?im) sets caseless, multiline matching. It is also possible to
1248 unset these options by preceding the letter with a hyphen, and a combined
1249 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1250 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1251 permitted. If a letter appears both before and after the hyphen, the option is
1252 unset.
1253 </P>
1254 <P>
1255 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1256 changed in the same way as the Perl-compatible options by using the characters
1257 J, U and X respectively.
1258 </P>
1259 <P>
1260 When one of these option changes occurs at top level (that is, not inside
1261 subpattern parentheses), the change applies to the remainder of the pattern
1262 that follows. If the change is placed right at the start of a pattern, PCRE
1263 extracts it into the global options (and it will therefore show up in data
1264 extracted by the <b>pcre_fullinfo()</b> function).
1265 </P>
1266 <P>
1267 An option change within a subpattern (see below for a description of
1268 subpatterns) affects only that part of the subpattern that follows it, so
1269 <pre>
1270 (a(?i)b)c
1271 </pre>
1272 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1273 By this means, options can be made to have different settings in different
1274 parts of the pattern. Any changes made in one alternative do carry on
1275 into subsequent branches within the same subpattern. For example,
1276 <pre>
1277 (a(?i)b|c)
1278 </pre>
1279 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1280 branch is abandoned before the option setting. This is because the effects of
1281 option settings happen at compile time. There would be some very weird
1282 behaviour otherwise.
1283 </P>
1284 <P>
1285 <b>Note:</b> There are other PCRE-specific options that can be set by the
1286 application when the compiling or matching functions are called. In some cases
1287 the pattern can contain special leading sequences such as (*CRLF) to override
1288 what the application has set or what has been defaulted. Details are given in
1289 the section entitled
1290 <a href="#newlineseq">"Newline sequences"</a>
1291 above. There are also the (*UTF8), (*UTF16), and (*UCP) leading sequences that
1292 can be used to set UTF and Unicode property modes; they are equivalent to
1293 setting the PCRE_UTF8, PCRE_UTF16, and the PCRE_UCP options, respectively.
1294 <a name="subpattern"></a></P>
1295 <br><a name="SEC12" href="#TOC1">SUBPATTERNS</a><br>
1296 <P>
1297 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1298 Turning part of a pattern into a subpattern does two things:
1299 <br>
1300 <br>
1301 1. It localizes a set of alternatives. For example, the pattern
1302 <pre>
1303 cat(aract|erpillar|)
1304 </pre>
1305 matches "cataract", "caterpillar", or "cat". Without the parentheses, it would
1306 match "cataract", "erpillar" or an empty string.
1307 <br>
1308 <br>
1309 2. It sets up the subpattern as a capturing subpattern. This means that, when
1310 the whole pattern matches, that portion of the subject string that matched the
1311 subpattern is passed back to the caller via the <i>ovector</i> argument of the
1312 matching function. (This applies only to the traditional matching functions;
1313 the DFA matching functions do not support capturing.)
1314 </P>
1315 <P>
1316 Opening parentheses are counted from left to right (starting from 1) to obtain
1317 numbers for the capturing subpatterns. For example, if the string "the red
1318 king" is matched against the pattern
1319 <pre>
1320 the ((red|white) (king|queen))
1321 </pre>
1322 the captured substrings are "red king", "red", and "king", and are numbered 1,
1323 2, and 3, respectively.
1324 </P>
1325 <P>
1326 The fact that plain parentheses fulfil two functions is not always helpful.
1327 There are often times when a grouping subpattern is required without a
1328 capturing requirement. If an opening parenthesis is followed by a question mark
1329 and a colon, the subpattern does not do any capturing, and is not counted when
1330 computing the number of any subsequent capturing subpatterns. For example, if
1331 the string "the white queen" is matched against the pattern
1332 <pre>
1333 the ((?:red|white) (king|queen))
1334 </pre>
1335 the captured substrings are "white queen" and "queen", and are numbered 1 and
1336 2. The maximum number of capturing subpatterns is 65535.
1337 </P>
1338 <P>
1339 As a convenient shorthand, if any option settings are required at the start of
1340 a non-capturing subpattern, the option letters may appear between the "?" and
1341 the ":". Thus the two patterns
1342 <pre>
1343 (?i:saturday|sunday)
1344 (?:(?i)saturday|sunday)
1345 </pre>
1346 match exactly the same set of strings. Because alternative branches are tried
1347 from left to right, and options are not reset until the end of the subpattern
1348 is reached, an option setting in one branch does affect subsequent branches, so
1349 the above patterns match "SUNDAY" as well as "Saturday".
1350 <a name="dupsubpatternnumber"></a></P>
1351 <br><a name="SEC13" href="#TOC1">DUPLICATE SUBPATTERN NUMBERS</a><br>
1352 <P>
1353 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1354 the same numbers for its capturing parentheses. Such a subpattern starts with
1355 (?| and is itself a non-capturing subpattern. For example, consider this
1356 pattern:
1357 <pre>
1358 (?|(Sat)ur|(Sun))day
1359 </pre>
1360 Because the two alternatives are inside a (?| group, both sets of capturing
1361 parentheses are numbered one. Thus, when the pattern matches, you can look
1362 at captured substring number one, whichever alternative matched. This construct
1363 is useful when you want to capture part, but not all, of one of a number of
1364 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1365 number is reset at the start of each branch. The numbers of any capturing
1366 parentheses that follow the subpattern start after the highest number used in
1367 any branch. The following example is taken from the Perl documentation. The
1368 numbers underneath show in which buffer the captured content will be stored.
1369 <pre>
1370 # before ---------------branch-reset----------- after
1371 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1372 # 1 2 2 3 2 3 4
1373 </pre>
1374 A back reference to a numbered subpattern uses the most recent value that is
1375 set for that number by any subpattern. The following pattern matches "abcabc"
1376 or "defdef":
1377 <pre>
1378 /(?|(abc)|(def))\1/
1379 </pre>
1380 In contrast, a subroutine call to a numbered subpattern always refers to the
1381 first one in the pattern with the given number. The following pattern matches
1382 "abcabc" or "defabc":
1383 <pre>
1384 /(?|(abc)|(def))(?1)/
1385 </pre>
1386 If a
1387 <a href="#conditions">condition test</a>
1388 for a subpattern's having matched refers to a non-unique number, the test is
1389 true if any of the subpatterns of that number have matched.
1390 </P>
1391 <P>
1392 An alternative approach to using this "branch reset" feature is to use
1393 duplicate named subpatterns, as described in the next section.
1394 </P>
1395 <br><a name="SEC14" href="#TOC1">NAMED SUBPATTERNS</a><br>
1396 <P>
1397 Identifying capturing parentheses by number is simple, but it can be very hard
1398 to keep track of the numbers in complicated regular expressions. Furthermore,
1399 if an expression is modified, the numbers may change. To help with this
1400 difficulty, PCRE supports the naming of subpatterns. This feature was not
1401 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1402 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1403 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1404 have different names, but PCRE does not.
1405 </P>
1406 <P>
1407 In PCRE, a subpattern can be named in one of three ways: (?&#60;name&#62;...) or
1408 (?'name'...) as in Perl, or (?P&#60;name&#62;...) as in Python. References to capturing
1409 parentheses from other parts of the pattern, such as
1410 <a href="#backreferences">back references,</a>
1411 <a href="#recursion">recursion,</a>
1412 and
1413 <a href="#conditions">conditions,</a>
1414 can be made by name as well as by number.
1415 </P>
1416 <P>
1417 Names consist of up to 32 alphanumeric characters and underscores. Named
1418 capturing parentheses are still allocated numbers as well as names, exactly as
1419 if the names were not present. The PCRE API provides function calls for
1420 extracting the name-to-number translation table from a compiled pattern. There
1421 is also a convenience function for extracting a captured substring by name.
1422 </P>
1423 <P>
1424 By default, a name must be unique within a pattern, but it is possible to relax
1425 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1426 names are also always permitted for subpatterns with the same number, set up as
1427 described in the previous section.) Duplicate names can be useful for patterns
1428 where only one instance of the named parentheses can match. Suppose you want to
1429 match the name of a weekday, either as a 3-letter abbreviation or as the full
1430 name, and in both cases you want to extract the abbreviation. This pattern
1431 (ignoring the line breaks) does the job:
1432 <pre>
1433 (?&#60;DN&#62;Mon|Fri|Sun)(?:day)?|
1434 (?&#60;DN&#62;Tue)(?:sday)?|
1435 (?&#60;DN&#62;Wed)(?:nesday)?|
1436 (?&#60;DN&#62;Thu)(?:rsday)?|
1437 (?&#60;DN&#62;Sat)(?:urday)?
1438 </pre>
1439 There are five capturing substrings, but only one is ever set after a match.
1440 (An alternative way of solving this problem is to use a "branch reset"
1441 subpattern, as described in the previous section.)
1442 </P>
1443 <P>
1444 The convenience function for extracting the data by name returns the substring
1445 for the first (and in this example, the only) subpattern of that name that
1446 matched. This saves searching to find which numbered subpattern it was.
1447 </P>
1448 <P>
1449 If you make a back reference to a non-unique named subpattern from elsewhere in
1450 the pattern, the one that corresponds to the first occurrence of the name is
1451 used. In the absence of duplicate numbers (see the previous section) this is
1452 the one with the lowest number. If you use a named reference in a condition
1453 test (see the
1454 <a href="#conditions">section about conditions</a>
1455 below), either to check whether a subpattern has matched, or to check for
1456 recursion, all subpatterns with the same name are tested. If the condition is
1457 true for any one of them, the overall condition is true. This is the same
1458 behaviour as testing by number. For further details of the interfaces for
1459 handling named subpatterns, see the
1460 <a href="pcreapi.html"><b>pcreapi</b></a>
1461 documentation.
1462 </P>
1463 <P>
1464 <b>Warning:</b> You cannot use different names to distinguish between two
1465 subpatterns with the same number because PCRE uses only the numbers when
1466 matching. For this reason, an error is given at compile time if different names
1467 are given to subpatterns with the same number. However, you can give the same
1468 name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
1469 </P>
1470 <br><a name="SEC15" href="#TOC1">REPETITION</a><br>
1471 <P>
1472 Repetition is specified by quantifiers, which can follow any of the following
1473 items:
1474 <pre>
1475 a literal data character
1476 the dot metacharacter
1477 the \C escape sequence
1478 the \X escape sequence
1479 the \R escape sequence
1480 an escape such as \d or \pL that matches a single character
1481 a character class
1482 a back reference (see next section)
1483 a parenthesized subpattern (including assertions)
1484 a subroutine call to a subpattern (recursive or otherwise)
1485 </pre>
1486 The general repetition quantifier specifies a minimum and maximum number of
1487 permitted matches, by giving the two numbers in curly brackets (braces),
1488 separated by a comma. The numbers must be less than 65536, and the first must
1489 be less than or equal to the second. For example:
1490 <pre>
1491 z{2,4}
1492 </pre>
1493 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1494 character. If the second number is omitted, but the comma is present, there is
1495 no upper limit; if the second number and the comma are both omitted, the
1496 quantifier specifies an exact number of required matches. Thus
1497 <pre>
1498 [aeiou]{3,}
1499 </pre>
1500 matches at least 3 successive vowels, but may match many more, while
1501 <pre>
1502 \d{8}
1503 </pre>
1504 matches exactly 8 digits. An opening curly bracket that appears in a position
1505 where a quantifier is not allowed, or one that does not match the syntax of a
1506 quantifier, is taken as a literal character. For example, {,6} is not a
1507 quantifier, but a literal string of four characters.
1508 </P>
1509 <P>
1510 In UTF modes, quantifiers apply to characters rather than to individual data
1511 units. Thus, for example, \x{100}{2} matches two characters, each of
1512 which is represented by a two-byte sequence in a UTF-8 string. Similarly,
1513 \X{3} matches three Unicode extended sequences, each of which may be several
1514 data units long (and they may be of different lengths).
1515 </P>
1516 <P>
1517 The quantifier {0} is permitted, causing the expression to behave as if the
1518 previous item and the quantifier were not present. This may be useful for
1519 subpatterns that are referenced as
1520 <a href="#subpatternsassubroutines">subroutines</a>
1521 from elsewhere in the pattern (but see also the section entitled
1522 <a href="#subdefine">"Defining subpatterns for use by reference only"</a>
1523 below). Items other than subpatterns that have a {0} quantifier are omitted
1524 from the compiled pattern.
1525 </P>
1526 <P>
1527 For convenience, the three most common quantifiers have single-character
1528 abbreviations:
1529 <pre>
1530 * is equivalent to {0,}
1531 + is equivalent to {1,}
1532 ? is equivalent to {0,1}
1533 </pre>
1534 It is possible to construct infinite loops by following a subpattern that can
1535 match no characters with a quantifier that has no upper limit, for example:
1536 <pre>
1537 (a?)*
1538 </pre>
1539 Earlier versions of Perl and PCRE used to give an error at compile time for
1540 such patterns. However, because there are cases where this can be useful, such
1541 patterns are now accepted, but if any repetition of the subpattern does in fact
1542 match no characters, the loop is forcibly broken.
1543 </P>
1544 <P>
1545 By default, the quantifiers are "greedy", that is, they match as much as
1546 possible (up to the maximum number of permitted times), without causing the
1547 rest of the pattern to fail. The classic example of where this gives problems
1548 is in trying to match comments in C programs. These appear between /* and */
1549 and within the comment, individual * and / characters may appear. An attempt to
1550 match C comments by applying the pattern
1551 <pre>
1552 /\*.*\*/
1553 </pre>
1554 to the string
1555 <pre>
1556 /* first comment */ not comment /* second comment */
1557 </pre>
1558 fails, because it matches the entire string owing to the greediness of the .*
1559 item.
1560 </P>
1561 <P>
1562 However, if a quantifier is followed by a question mark, it ceases to be
1563 greedy, and instead matches the minimum number of times possible, so the
1564 pattern
1565 <pre>
1566 /\*.*?\*/
1567 </pre>
1568 does the right thing with the C comments. The meaning of the various
1569 quantifiers is not otherwise changed, just the preferred number of matches.
1570 Do not confuse this use of question mark with its use as a quantifier in its
1571 own right. Because it has two uses, it can sometimes appear doubled, as in
1572 <pre>
1573 \d??\d
1574 </pre>
1575 which matches one digit by preference, but can match two if that is the only
1576 way the rest of the pattern matches.
1577 </P>
1578 <P>
1579 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1580 the quantifiers are not greedy by default, but individual ones can be made
1581 greedy by following them with a question mark. In other words, it inverts the
1582 default behaviour.
1583 </P>
1584 <P>
1585 When a parenthesized subpattern is quantified with a minimum repeat count that
1586 is greater than 1 or with a limited maximum, more memory is required for the
1587 compiled pattern, in proportion to the size of the minimum or maximum.
1588 </P>
1589 <P>
1590 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1591 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1592 implicitly anchored, because whatever follows will be tried against every
1593 character position in the subject string, so there is no point in retrying the
1594 overall match at any position after the first. PCRE normally treats such a
1595 pattern as though it were preceded by \A.
1596 </P>
1597 <P>
1598 In cases where it is known that the subject string contains no newlines, it is
1599 worth setting PCRE_DOTALL in order to obtain this optimization, or
1600 alternatively using ^ to indicate anchoring explicitly.
1601 </P>
1602 <P>
1603 However, there is one situation where the optimization cannot be used. When .*
1604 is inside capturing parentheses that are the subject of a back reference
1605 elsewhere in the pattern, a match at the start may fail where a later one
1606 succeeds. Consider, for example:
1607 <pre>
1608 (.*)abc\1
1609 </pre>
1610 If the subject is "xyz123abc123" the match point is the fourth character. For
1611 this reason, such a pattern is not implicitly anchored.
1612 </P>
1613 <P>
1614 When a capturing subpattern is repeated, the value captured is the substring
1615 that matched the final iteration. For example, after
1616 <pre>
1617 (tweedle[dume]{3}\s*)+
1618 </pre>
1619 has matched "tweedledum tweedledee" the value of the captured substring is
1620 "tweedledee". However, if there are nested capturing subpatterns, the
1621 corresponding captured values may have been set in previous iterations. For
1622 example, after
1623 <pre>
1624 /(a|(b))+/
1625 </pre>
1626 matches "aba" the value of the second captured substring is "b".
1627 <a name="atomicgroup"></a></P>
1628 <br><a name="SEC16" href="#TOC1">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a><br>
1629 <P>
1630 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1631 repetition, failure of what follows normally causes the repeated item to be
1632 re-evaluated to see if a different number of repeats allows the rest of the
1633 pattern to match. Sometimes it is useful to prevent this, either to change the
1634 nature of the match, or to cause it fail earlier than it otherwise might, when
1635 the author of the pattern knows there is no point in carrying on.
1636 </P>
1637 <P>
1638 Consider, for example, the pattern \d+foo when applied to the subject line
1639 <pre>
1640 123456bar
1641 </pre>
1642 After matching all 6 digits and then failing to match "foo", the normal
1643 action of the matcher is to try again with only 5 digits matching the \d+
1644 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1645 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1646 that once a subpattern has matched, it is not to be re-evaluated in this way.
1647 </P>
1648 <P>
1649 If we use atomic grouping for the previous example, the matcher gives up
1650 immediately on failing to match "foo" the first time. The notation is a kind of
1651 special parenthesis, starting with (?&#62; as in this example:
1652 <pre>
1653 (?&#62;\d+)foo
1654 </pre>
1655 This kind of parenthesis "locks up" the part of the pattern it contains once
1656 it has matched, and a failure further into the pattern is prevented from
1657 backtracking into it. Backtracking past it to previous items, however, works as
1658 normal.
1659 </P>
1660 <P>
1661 An alternative description is that a subpattern of this type matches the string
1662 of characters that an identical standalone pattern would match, if anchored at
1663 the current point in the subject string.
1664 </P>
1665 <P>
1666 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1667 the above example can be thought of as a maximizing repeat that must swallow
1668 everything it can. So, while both \d+ and \d+? are prepared to adjust the
1669 number of digits they match in order to make the rest of the pattern match,
1670 (?&#62;\d+) can only match an entire sequence of digits.
1671 </P>
1672 <P>
1673 Atomic groups in general can of course contain arbitrarily complicated
1674 subpatterns, and can be nested. However, when the subpattern for an atomic
1675 group is just a single repeated item, as in the example above, a simpler
1676 notation, called a "possessive quantifier" can be used. This consists of an
1677 additional + character following a quantifier. Using this notation, the
1678 previous example can be rewritten as
1679 <pre>
1680 \d++foo
1681 </pre>
1682 Note that a possessive quantifier can be used with an entire group, for
1683 example:
1684 <pre>
1685 (abc|xyz){2,3}+
1686 </pre>
1687 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1688 option is ignored. They are a convenient notation for the simpler forms of
1689 atomic group. However, there is no difference in the meaning of a possessive
1690 quantifier and the equivalent atomic group, though there may be a performance
1691 difference; possessive quantifiers should be slightly faster.
1692 </P>
1693 <P>
1694 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1695 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1696 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1697 package, and PCRE copied it from there. It ultimately found its way into Perl
1698 at release 5.10.
1699 </P>
1700 <P>
1701 PCRE has an optimization that automatically "possessifies" certain simple
1702 pattern constructs. For example, the sequence A+B is treated as A++B because
1703 there is no point in backtracking into a sequence of A's when B must follow.
1704 </P>
1705 <P>
1706 When a pattern contains an unlimited repeat inside a subpattern that can itself
1707 be repeated an unlimited number of times, the use of an atomic group is the
1708 only way to avoid some failing matches taking a very long time indeed. The
1709 pattern
1710 <pre>
1711 (\D+|&#60;\d+&#62;)*[!?]
1712 </pre>
1713 matches an unlimited number of substrings that either consist of non-digits, or
1714 digits enclosed in &#60;&#62;, followed by either ! or ?. When it matches, it runs
1715 quickly. However, if it is applied to
1716 <pre>
1717 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1718 </pre>
1719 it takes a long time before reporting failure. This is because the string can
1720 be divided between the internal \D+ repeat and the external * repeat in a
1721 large number of ways, and all have to be tried. (The example uses [!?] rather
1722 than a single character at the end, because both PCRE and Perl have an
1723 optimization that allows for fast failure when a single character is used. They
1724 remember the last single character that is required for a match, and fail early
1725 if it is not present in the string.) If the pattern is changed so that it uses
1726 an atomic group, like this:
1727 <pre>
1728 ((?&#62;\D+)|&#60;\d+&#62;)*[!?]
1729 </pre>
1730 sequences of non-digits cannot be broken, and failure happens quickly.
1731 <a name="backreferences"></a></P>
1732 <br><a name="SEC17" href="#TOC1">BACK REFERENCES</a><br>
1733 <P>
1734 Outside a character class, a backslash followed by a digit greater than 0 (and
1735 possibly further digits) is a back reference to a capturing subpattern earlier
1736 (that is, to its left) in the pattern, provided there have been that many
1737 previous capturing left parentheses.
1738 </P>
1739 <P>
1740 However, if the decimal number following the backslash is less than 10, it is
1741 always taken as a back reference, and causes an error only if there are not
1742 that many capturing left parentheses in the entire pattern. In other words, the
1743 parentheses that are referenced need not be to the left of the reference for
1744 numbers less than 10. A "forward back reference" of this type can make sense
1745 when a repetition is involved and the subpattern to the right has participated
1746 in an earlier iteration.
1747 </P>
1748 <P>
1749 It is not possible to have a numerical "forward back reference" to a subpattern
1750 whose number is 10 or more using this syntax because a sequence such as \50 is
1751 interpreted as a character defined in octal. See the subsection entitled
1752 "Non-printing characters"
1753 <a href="#digitsafterbackslash">above</a>
1754 for further details of the handling of digits following a backslash. There is
1755 no such problem when named parentheses are used. A back reference to any
1756 subpattern is possible using named parentheses (see below).
1757 </P>
1758 <P>
1759 Another way of avoiding the ambiguity inherent in the use of digits following a
1760 backslash is to use the \g escape sequence. This escape must be followed by an
1761 unsigned number or a negative number, optionally enclosed in braces. These
1762 examples are all identical:
1763 <pre>
1764 (ring), \1
1765 (ring), \g1
1766 (ring), \g{1}
1767 </pre>
1768 An unsigned number specifies an absolute reference without the ambiguity that
1769 is present in the older syntax. It is also useful when literal digits follow
1770 the reference. A negative number is a relative reference. Consider this
1771 example:
1772 <pre>
1773 (abc(def)ghi)\g{-1}
1774 </pre>
1775 The sequence \g{-1} is a reference to the most recently started capturing
1776 subpattern before \g, that is, is it equivalent to \2 in this example.
1777 Similarly, \g{-2} would be equivalent to \1. The use of relative references
1778 can be helpful in long patterns, and also in patterns that are created by
1779 joining together fragments that contain references within themselves.
1780 </P>
1781 <P>
1782 A back reference matches whatever actually matched the capturing subpattern in
1783 the current subject string, rather than anything matching the subpattern
1784 itself (see
1785 <a href="#subpatternsassubroutines">"Subpatterns as subroutines"</a>
1786 below for a way of doing that). So the pattern
1787 <pre>
1788 (sens|respons)e and \1ibility
1789 </pre>
1790 matches "sense and sensibility" and "response and responsibility", but not
1791 "sense and responsibility". If caseful matching is in force at the time of the
1792 back reference, the case of letters is relevant. For example,
1793 <pre>
1794 ((?i)rah)\s+\1
1795 </pre>
1796 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1797 capturing subpattern is matched caselessly.
1798 </P>
1799 <P>
1800 There are several different ways of writing back references to named
1801 subpatterns. The .NET syntax \k{name} and the Perl syntax \k&#60;name&#62; or
1802 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1803 back reference syntax, in which \g can be used for both numeric and named
1804 references, is also supported. We could rewrite the above example in any of
1805 the following ways:
1806 <pre>
1807 (?&#60;p1&#62;(?i)rah)\s+\k&#60;p1&#62;
1808 (?'p1'(?i)rah)\s+\k{p1}
1809 (?P&#60;p1&#62;(?i)rah)\s+(?P=p1)
1810 (?&#60;p1&#62;(?i)rah)\s+\g{p1}
1811 </pre>
1812 A subpattern that is referenced by name may appear in the pattern before or
1813 after the reference.
1814 </P>
1815 <P>
1816 There may be more than one back reference to the same subpattern. If a
1817 subpattern has not actually been used in a particular match, any back
1818 references to it always fail by default. For example, the pattern
1819 <pre>
1820 (a|(bc))\2
1821 </pre>
1822 always fails if it starts to match "a" rather than "bc". However, if the
1823 PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1824 unset value matches an empty string.
1825 </P>
1826 <P>
1827 Because there may be many capturing parentheses in a pattern, all digits
1828 following a backslash are taken as part of a potential back reference number.
1829 If the pattern continues with a digit character, some delimiter must be used to
1830 terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1831 whitespace. Otherwise, the \g{ syntax or an empty comment (see
1832 <a href="#comments">"Comments"</a>
1833 below) can be used.
1834 </P>
1835 <br><b>
1836 Recursive back references
1837 </b><br>
1838 <P>
1839 A back reference that occurs inside the parentheses to which it refers fails
1840 when the subpattern is first used, so, for example, (a\1) never matches.
1841 However, such references can be useful inside repeated subpatterns. For
1842 example, the pattern
1843 <pre>
1844 (a|b\1)+
1845 </pre>
1846 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1847 the subpattern, the back reference matches the character string corresponding
1848 to the previous iteration. In order for this to work, the pattern must be such
1849 that the first iteration does not need to match the back reference. This can be
1850 done using alternation, as in the example above, or by a quantifier with a
1851 minimum of zero.
1852 </P>
1853 <P>
1854 Back references of this type cause the group that they reference to be treated
1855 as an
1856 <a href="#atomicgroup">atomic group.</a>
1857 Once the whole group has been matched, a subsequent matching failure cannot
1858 cause backtracking into the middle of the group.
1859 <a name="bigassertions"></a></P>
1860 <br><a name="SEC18" href="#TOC1">ASSERTIONS</a><br>
1861 <P>
1862 An assertion is a test on the characters following or preceding the current
1863 matching point that does not actually consume any characters. The simple
1864 assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are described
1865 <a href="#smallassertions">above.</a>
1866 </P>
1867 <P>
1868 More complicated assertions are coded as subpatterns. There are two kinds:
1869 those that look ahead of the current position in the subject string, and those
1870 that look behind it. An assertion subpattern is matched in the normal way,
1871 except that it does not cause the current matching position to be changed.
1872 </P>
1873 <P>
1874 Assertion subpatterns are not capturing subpatterns. If such an assertion
1875 contains capturing subpatterns within it, these are counted for the purposes of
1876 numbering the capturing subpatterns in the whole pattern. However, substring
1877 capturing is carried out only for positive assertions, because it does not make
1878 sense for negative assertions.
1879 </P>
1880 <P>
1881 For compatibility with Perl, assertion subpatterns may be repeated; though
1882 it makes no sense to assert the same thing several times, the side effect of
1883 capturing parentheses may occasionally be useful. In practice, there only three
1884 cases:
1885 <br>
1886 <br>
1887 (1) If the quantifier is {0}, the assertion is never obeyed during matching.
1888 However, it may contain internal capturing parenthesized groups that are called
1889 from elsewhere via the
1890 <a href="#subpatternsassubroutines">subroutine mechanism.</a>
1891 <br>
1892 <br>
1893 (2) If quantifier is {0,n} where n is greater than zero, it is treated as if it
1894 were {0,1}. At run time, the rest of the pattern match is tried with and
1895 without the assertion, the order depending on the greediness of the quantifier.
1896 <br>
1897 <br>
1898 (3) If the minimum repetition is greater than zero, the quantifier is ignored.
1899 The assertion is obeyed just once when encountered during matching.
1900 </P>
1901 <br><b>
1902 Lookahead assertions
1903 </b><br>
1904 <P>
1905 Lookahead assertions start with (?= for positive assertions and (?! for
1906 negative assertions. For example,
1907 <pre>
1908 \w+(?=;)
1909 </pre>
1910 matches a word followed by a semicolon, but does not include the semicolon in
1911 the match, and
1912 <pre>
1913 foo(?!bar)
1914 </pre>
1915 matches any occurrence of "foo" that is not followed by "bar". Note that the
1916 apparently similar pattern
1917 <pre>
1918 (?!foo)bar
1919 </pre>
1920 does not find an occurrence of "bar" that is preceded by something other than
1921 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1922 (?!foo) is always true when the next three characters are "bar". A
1923 lookbehind assertion is needed to achieve the other effect.
1924 </P>
1925 <P>
1926 If you want to force a matching failure at some point in a pattern, the most
1927 convenient way to do it is with (?!) because an empty string always matches, so
1928 an assertion that requires there not to be an empty string must always fail.
1929 The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).
1930 <a name="lookbehind"></a></P>
1931 <br><b>
1932 Lookbehind assertions
1933 </b><br>
1934 <P>
1935 Lookbehind assertions start with (?&#60;= for positive assertions and (?&#60;! for
1936 negative assertions. For example,
1937 <pre>
1938 (?&#60;!foo)bar
1939 </pre>
1940 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1941 a lookbehind assertion are restricted such that all the strings it matches must
1942 have a fixed length. However, if there are several top-level alternatives, they
1943 do not all have to have the same fixed length. Thus
1944 <pre>
1945 (?&#60;=bullock|donkey)
1946 </pre>
1947 is permitted, but
1948 <pre>
1949 (?&#60;!dogs?|cats?)
1950 </pre>
1951 causes an error at compile time. Branches that match different length strings
1952 are permitted only at the top level of a lookbehind assertion. This is an
1953 extension compared with Perl, which requires all branches to match the same
1954 length of string. An assertion such as
1955 <pre>
1956 (?&#60;=ab(c|de))
1957 </pre>
1958 is not permitted, because its single top-level branch can match two different
1959 lengths, but it is acceptable to PCRE if rewritten to use two top-level
1960 branches:
1961 <pre>
1962 (?&#60;=abc|abde)
1963 </pre>
1964 In some cases, the escape sequence \K
1965 <a href="#resetmatchstart">(see above)</a>
1966 can be used instead of a lookbehind assertion to get round the fixed-length
1967 restriction.
1968 </P>
1969 <P>
1970 The implementation of lookbehind assertions is, for each alternative, to
1971 temporarily move the current position back by the fixed length and then try to
1972 match. If there are insufficient characters before the current position, the
1973 assertion fails.
1974 </P>
1975 <P>
1976 In a UTF mode, PCRE does not allow the \C escape (which matches a single data
1977 unit even in a UTF mode) to appear in lookbehind assertions, because it makes
1978 it impossible to calculate the length of the lookbehind. The \X and \R
1979 escapes, which can match different numbers of data units, are also not
1980 permitted.
1981 </P>
1982 <P>
1983 <a href="#subpatternsassubroutines">"Subroutine"</a>
1984 calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
1985 as the subpattern matches a fixed-length string.
1986 <a href="#recursion">Recursion,</a>
1987 however, is not supported.
1988 </P>
1989 <P>
1990 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1991 specify efficient matching of fixed-length strings at the end of subject
1992 strings. Consider a simple pattern such as
1993 <pre>
1994 abcd$
1995 </pre>
1996 when applied to a long string that does not match. Because matching proceeds
1997 from left to right, PCRE will look for each "a" in the subject and then see if
1998 what follows matches the rest of the pattern. If the pattern is specified as
1999 <pre>
2000 ^.*abcd$
2001 </pre>
2002 the initial .* matches the entire string at first, but when this fails (because
2003 there is no following "a"), it backtracks to match all but the last character,
2004 then all but the last two characters, and so on. Once again the search for "a"
2005 covers the entire string, from right to left, so we are no better off. However,
2006 if the pattern is written as
2007 <pre>
2008 ^.*+(?&#60;=abcd)
2009 </pre>
2010 there can be no backtracking for the .*+ item; it can match only the entire
2011 string. The subsequent lookbehind assertion does a single test on the last four
2012 characters. If it fails, the match fails immediately. For long strings, this
2013 approach makes a significant difference to the processing time.
2014 </P>
2015 <br><b>
2016 Using multiple assertions
2017 </b><br>
2018 <P>
2019 Several assertions (of any sort) may occur in succession. For example,
2020 <pre>
2021 (?&#60;=\d{3})(?&#60;!999)foo
2022 </pre>
2023 matches "foo" preceded by three digits that are not "999". Notice that each of
2024 the assertions is applied independently at the same point in the subject
2025 string. First there is a check that the previous three characters are all
2026 digits, and then there is a check that the same three characters are not "999".
2027 This pattern does <i>not</i> match "foo" preceded by six characters, the first
2028 of which are digits and the last three of which are not "999". For example, it
2029 doesn't match "123abcfoo". A pattern to do that is
2030 <pre>
2031 (?&#60;=\d{3}...)(?&#60;!999)foo
2032 </pre>
2033 This time the first assertion looks at the preceding six characters, checking
2034 that the first three are digits, and then the second assertion checks that the
2035 preceding three characters are not "999".
2036 </P>
2037 <P>
2038 Assertions can be nested in any combination. For example,
2039 <pre>
2040 (?&#60;=(?&#60;!foo)bar)baz
2041 </pre>
2042 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
2043 preceded by "foo", while
2044 <pre>
2045 (?&#60;=\d{3}(?!999)...)foo
2046 </pre>
2047 is another pattern that matches "foo" preceded by three digits and any three
2048 characters that are not "999".
2049 <a name="conditions"></a></P>
2050 <br><a name="SEC19" href="#TOC1">CONDITIONAL SUBPATTERNS</a><br>
2051 <P>
2052 It is possible to cause the matching process to obey a subpattern
2053 conditionally or to choose between two alternative subpatterns, depending on
2054 the result of an assertion, or whether a specific capturing subpattern has
2055 already been matched. The two possible forms of conditional subpattern are:
2056 <pre>
2057 (?(condition)yes-pattern)
2058 (?(condition)yes-pattern|no-pattern)
2059 </pre>
2060 If the condition is satisfied, the yes-pattern is used; otherwise the
2061 no-pattern (if present) is used. If there are more than two alternatives in the
2062 subpattern, a compile-time error occurs. Each of the two alternatives may
2063 itself contain nested subpatterns of any form, including conditional
2064 subpatterns; the restriction to two alternatives applies only at the level of
2065 the condition. This pattern fragment is an example where the alternatives are
2066 complex:
2067 <pre>
2068 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
2069
2070 </PRE>
2071 </P>
2072 <P>
2073 There are four kinds of condition: references to subpatterns, references to
2074 recursion, a pseudo-condition called DEFINE, and assertions.
2075 </P>
2076 <br><b>
2077 Checking for a used subpattern by number
2078 </b><br>
2079 <P>
2080 If the text between the parentheses consists of a sequence of digits, the
2081 condition is true if a capturing subpattern of that number has previously
2082 matched. If there is more than one capturing subpattern with the same number
2083 (see the earlier
2084 <a href="#recursion">section about duplicate subpattern numbers),</a>
2085 the condition is true if any of them have matched. An alternative notation is
2086 to precede the digits with a plus or minus sign. In this case, the subpattern
2087 number is relative rather than absolute. The most recently opened parentheses
2088 can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside
2089 loops it can also make sense to refer to subsequent groups. The next
2090 parentheses to be opened can be referenced as (?(+1), and so on. (The value
2091 zero in any of these forms is not used; it provokes a compile-time error.)
2092 </P>
2093 <P>
2094 Consider the following pattern, which contains non-significant white space to
2095 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
2096 three parts for ease of discussion:
2097 <pre>
2098 ( \( )? [^()]+ (?(1) \) )
2099 </pre>
2100 The first part matches an optional opening parenthesis, and if that
2101 character is present, sets it as the first captured substring. The second part
2102 matches one or more characters that are not parentheses. The third part is a
2103 conditional subpattern that tests whether or not the first set of parentheses
2104 matched. If they did, that is, if subject started with an opening parenthesis,
2105 the condition is true, and so the yes-pattern is executed and a closing
2106 parenthesis is required. Otherwise, since no-pattern is not present, the
2107 subpattern matches nothing. In other words, this pattern matches a sequence of
2108 non-parentheses, optionally enclosed in parentheses.
2109 </P>
2110 <P>
2111 If you were embedding this pattern in a larger one, you could use a relative
2112 reference:
2113 <pre>
2114 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
2115 </pre>
2116 This makes the fragment independent of the parentheses in the larger pattern.
2117 </P>
2118 <br><b>
2119 Checking for a used subpattern by name
2120 </b><br>
2121 <P>
2122 Perl uses the syntax (?(&#60;name&#62;)...) or (?('name')...) to test for a used
2123 subpattern by name. For compatibility with earlier versions of PCRE, which had
2124 this facility before Perl, the syntax (?(name)...) is also recognized. However,
2125 there is a possible ambiguity with this syntax, because subpattern names may
2126 consist entirely of digits. PCRE looks first for a named subpattern; if it
2127 cannot find one and the name consists entirely of digits, PCRE looks for a
2128 subpattern of that number, which must be greater than zero. Using subpattern
2129 names that consist entirely of digits is not recommended.
2130 </P>
2131 <P>
2132 Rewriting the above example to use a named subpattern gives this:
2133 <pre>
2134 (?&#60;OPEN&#62; \( )? [^()]+ (?(&#60;OPEN&#62;) \) )
2135 </pre>
2136 If the name used in a condition of this kind is a duplicate, the test is
2137 applied to all subpatterns of the same name, and is true if any one of them has
2138 matched.
2139 </P>
2140 <br><b>
2141 Checking for pattern recursion
2142 </b><br>
2143 <P>
2144 If the condition is the string (R), and there is no subpattern with the name R,
2145 the condition is true if a recursive call to the whole pattern or any
2146 subpattern has been made. If digits or a name preceded by ampersand follow the
2147 letter R, for example:
2148 <pre>
2149 (?(R3)...) or (?(R&name)...)
2150 </pre>
2151 the condition is true if the most recent recursion is into a subpattern whose
2152 number or name is given. This condition does not check the entire recursion
2153 stack. If the name used in a condition of this kind is a duplicate, the test is
2154 applied to all subpatterns of the same name, and is true if any one of them is
2155 the most recent recursion.
2156 </P>
2157 <P>
2158 At "top level", all these recursion test conditions are false.
2159 <a href="#recursion">The syntax for recursive patterns</a>
2160 is described below.
2161 <a name="subdefine"></a></P>
2162 <br><b>
2163 Defining subpatterns for use by reference only
2164 </b><br>
2165 <P>
2166 If the condition is the string (DEFINE), and there is no subpattern with the
2167 name DEFINE, the condition is always false. In this case, there may be only one
2168 alternative in the subpattern. It is always skipped if control reaches this
2169 point in the pattern; the idea of DEFINE is that it can be used to define
2170 subroutines that can be referenced from elsewhere. (The use of
2171 <a href="#subpatternsassubroutines">subroutines</a>
2172 is described below.) For example, a pattern to match an IPv4 address such as
2173 "192.168.23.245" could be written like this (ignore whitespace and line
2174 breaks):
2175 <pre>
2176 (?(DEFINE) (?&#60;byte&#62; 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
2177 \b (?&byte) (\.(?&byte)){3} \b
2178 </pre>
2179 The first part of the pattern is a DEFINE group inside which a another group
2180 named "byte" is defined. This matches an individual component of an IPv4
2181 address (a number less than 256). When matching takes place, this part of the
2182 pattern is skipped because DEFINE acts like a false condition. The rest of the
2183 pattern uses references to the named group to match the four dot-separated
2184 components of an IPv4 address, insisting on a word boundary at each end.
2185 </P>
2186 <br><b>
2187 Assertion conditions
2188 </b><br>
2189 <P>
2190 If the condition is not in any of the above formats, it must be an assertion.
2191 This may be a positive or negative lookahead or lookbehind assertion. Consider
2192 this pattern, again containing non-significant white space, and with the two
2193 alternatives on the second line:
2194 <pre>
2195 (?(?=[^a-z]*[a-z])
2196 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
2197 </pre>
2198 The condition is a positive lookahead assertion that matches an optional
2199 sequence of non-letters followed by a letter. In other words, it tests for the
2200 presence of at least one letter in the subject. If a letter is found, the
2201 subject is matched against the first alternative; otherwise it is matched
2202 against the second. This pattern matches strings in one of the two forms
2203 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2204 <a name="comments"></a></P>
2205 <br><a name="SEC20" href="#TOC1">COMMENTS</a><br>
2206 <P>
2207 There are two ways of including comments in patterns that are processed by
2208 PCRE. In both cases, the start of the comment must not be in a character class,
2209 nor in the middle of any other sequence of related characters such as (?: or a
2210 subpattern name or number. The characters that make up a comment play no part
2211 in the pattern matching.
2212 </P>
2213 <P>
2214 The sequence (?# marks the start of a comment that continues up to the next
2215 closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED
2216 option is set, an unescaped # character also introduces a comment, which in
2217 this case continues to immediately after the next newline character or
2218 character sequence in the pattern. Which characters are interpreted as newlines
2219 is controlled by the options passed to a compiling function or by a special
2220 sequence at the start of the pattern, as described in the section entitled
2221 <a href="#newlines">"Newline conventions"</a>
2222 above. Note that the end of this type of comment is a literal newline sequence
2223 in the pattern; escape sequences that happen to represent a newline do not
2224 count. For example, consider this pattern when PCRE_EXTENDED is set, and the
2225 default newline convention is in force:
2226 <pre>
2227 abc #comment \n still comment
2228 </pre>
2229 On encountering the # character, <b>pcre_compile()</b> skips along, looking for
2230 a newline in the pattern. The sequence \n is still literal at this stage, so
2231 it does not terminate the comment. Only an actual character with the code value
2232 0x0a (the default newline) does so.
2233 <a name="recursion"></a></P>
2234 <br><a name="SEC21" href="#TOC1">RECURSIVE PATTERNS</a><br>
2235 <P>
2236 Consider the problem of matching a string in parentheses, allowing for
2237 unlimited nested parentheses. Without the use of recursion, the best that can
2238 be done is to use a pattern that matches up to some fixed depth of nesting. It
2239 is not possible to handle an arbitrary nesting depth.
2240 </P>
2241 <P>
2242 For some time, Perl has provided a facility that allows regular expressions to
2243 recurse (amongst other things). It does this by interpolating Perl code in the
2244 expression at run time, and the code can refer to the expression itself. A Perl
2245 pattern using code interpolation to solve the parentheses problem can be
2246 created like this:
2247 <pre>
2248 $re = qr{\( (?: (?&#62;[^()]+) | (?p{$re}) )* \)}x;
2249 </pre>
2250 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2251 recursively to the pattern in which it appears.
2252 </P>
2253 <P>
2254 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2255 supports special syntax for recursion of the entire pattern, and also for
2256 individual subpattern recursion. After its introduction in PCRE and Python,
2257 this kind of recursion was subsequently introduced into Perl at release 5.10.
2258 </P>
2259 <P>
2260 A special item that consists of (? followed by a number greater than zero and a
2261 closing parenthesis is a recursive subroutine call of the subpattern of the
2262 given number, provided that it occurs inside that subpattern. (If not, it is a
2263 <a href="#subpatternsassubroutines">non-recursive subroutine</a>
2264 call, which is described in the next section.) The special item (?R) or (?0) is
2265 a recursive call of the entire regular expression.
2266 </P>
2267 <P>
2268 This PCRE pattern solves the nested parentheses problem (assume the
2269 PCRE_EXTENDED option is set so that white space is ignored):
2270 <pre>
2271 \( ( [^()]++ | (?R) )* \)
2272 </pre>
2273 First it matches an opening parenthesis. Then it matches any number of
2274 substrings which can either be a sequence of non-parentheses, or a recursive
2275 match of the pattern itself (that is, a correctly parenthesized substring).
2276 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2277 to avoid backtracking into sequences of non-parentheses.
2278 </P>
2279 <P>
2280 If this were part of a larger pattern, you would not want to recurse the entire
2281 pattern, so instead you could use this:
2282 <pre>
2283 ( \( ( [^()]++ | (?1) )* \) )
2284 </pre>
2285 We have put the pattern into parentheses, and caused the recursion to refer to
2286 them instead of the whole pattern.
2287 </P>
2288 <P>
2289 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2290 is made easier by the use of relative references. Instead of (?1) in the
2291 pattern above you can write (?-2) to refer to the second most recently opened
2292 parentheses preceding the recursion. In other words, a negative number counts
2293 capturing parentheses leftwards from the point at which it is encountered.
2294 </P>
2295 <P>
2296 It is also possible to refer to subsequently opened parentheses, by writing
2297 references such as (?+2). However, these cannot be recursive because the
2298 reference is not inside the parentheses that are referenced. They are always
2299 <a href="#subpatternsassubroutines">non-recursive subroutine</a>
2300 calls, as described in the next section.
2301 </P>
2302 <P>
2303 An alternative approach is to use named parentheses instead. The Perl syntax
2304 for this is (?&name); PCRE's earlier syntax (?P&#62;name) is also supported. We
2305 could rewrite the above example as follows:
2306 <pre>
2307 (?&#60;pn&#62; \( ( [^()]++ | (?&pn) )* \) )
2308 </pre>
2309 If there is more than one subpattern with the same name, the earliest one is
2310 used.
2311 </P>
2312 <P>
2313 This particular example pattern that we have been looking at contains nested
2314 unlimited repeats, and so the use of a possessive quantifier for matching
2315 strings of non-parentheses is important when applying the pattern to strings
2316 that do not match. For example, when this pattern is applied to
2317 <pre>
2318 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2319 </pre>
2320 it yields "no match" quickly. However, if a possessive quantifier is not used,
2321 the match runs for a very long time indeed because there are so many different
2322 ways the + and * repeats can carve up the subject, and all have to be tested
2323 before failure can be reported.
2324 </P>
2325 <P>
2326 At the end of a match, the values of capturing parentheses are those from
2327 the outermost level. If you want to obtain intermediate values, a callout
2328 function can be used (see below and the
2329 <a href="pcrecallout.html"><b>pcrecallout</b></a>
2330 documentation). If the pattern above is matched against
2331 <pre>
2332 (ab(cd)ef)
2333 </pre>
2334 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2335 the last value taken on at the top level. If a capturing subpattern is not
2336 matched at the top level, its final captured value is unset, even if it was
2337 (temporarily) set at a deeper level during the matching process.
2338 </P>
2339 <P>
2340 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2341 obtain extra memory to store data during a recursion, which it does by using
2342 <b>pcre_malloc</b>, freeing it via <b>pcre_free</b> afterwards. If no memory can
2343 be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2344 </P>
2345 <P>
2346 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2347 Consider this pattern, which matches text in angle brackets, allowing for
2348 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2349 recursing), whereas any characters are permitted at the outer level.
2350 <pre>
2351 &#60; (?: (?(R) \d++ | [^&#60;&#62;]*+) | (?R)) * &#62;
2352 </pre>
2353 In this pattern, (?(R) is the start of a conditional subpattern, with two
2354 different alternatives for the recursive and non-recursive cases. The (?R) item
2355 is the actual recursive call.
2356 <a name="recursiondifference"></a></P>
2357 <br><b>
2358 Differences in recursion processing between PCRE and Perl
2359 </b><br>
2360 <P>
2361 Recursion processing in PCRE differs from Perl in two important ways. In PCRE
2362 (like Python, but unlike Perl), a recursive subpattern call is always treated
2363 as an atomic group. That is, once it has matched some of the subject string, it
2364 is never re-entered, even if it contains untried alternatives and there is a
2365 subsequent matching failure. This can be illustrated by the following pattern,
2366 which purports to match a palindromic string that contains an odd number of
2367 characters (for example, "a", "aba", "abcba", "abcdcba"):
2368 <pre>
2369 ^(.|(.)(?1)\2)$
2370 </pre>
2371 The idea is that it either matches a single character, or two identical
2372 characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2373 it does not if the pattern is longer than three characters. Consider the
2374 subject string "abcba":
2375 </P>
2376 <P>
2377 At the top level, the first character is matched, but as it is not at the end
2378 of the string, the first alternative fails; the second alternative is taken
2379 and the recursion kicks in. The recursive call to subpattern 1 successfully
2380 matches the next character ("b"). (Note that the beginning and end of line
2381 tests are not part of the recursion).
2382 </P>
2383 <P>
2384 Back at the top level, the next character ("c") is compared with what
2385 subpattern 2 matched, which was "a". This fails. Because the recursion is
2386 treated as an atomic group, there are now no backtracking points, and so the
2387 entire match fails. (Perl is able, at this point, to re-enter the recursion and
2388 try the second alternative.) However, if the pattern is written with the
2389 alternatives in the other order, things are different:
2390 <pre>
2391 ^((.)(?1)\2|.)$
2392 </pre>
2393 This time, the recursing alternative is tried first, and continues to recurse
2394 until it runs out of characters, at which point the recursion fails. But this
2395 time we do have another alternative to try at the higher level. That is the big
2396 difference: in the previous case the remaining alternative is at a deeper
2397 recursion level, which PCRE cannot use.
2398 </P>
2399 <P>
2400 To change the pattern so that it matches all palindromic strings, not just
2401 those with an odd number of characters, it is tempting to change the pattern to
2402 this:
2403 <pre>
2404 ^((.)(?1)\2|.?)$
2405 </pre>
2406 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2407 deeper recursion has matched a single character, it cannot be entered again in
2408 order to match an empty string. The solution is to separate the two cases, and
2409 write out the odd and even cases as alternatives at the higher level:
2410 <pre>
2411 ^(?:((.)(?1)\2|)|((.)(?3)\4|.))
2412 </pre>
2413 If you want to match typical palindromic phrases, the pattern has to ignore all
2414 non-word characters, which can be done like this:
2415 <pre>
2416 ^\W*+(?:((.)\W*+(?1)\W*+\2|)|((.)\W*+(?3)\W*+\4|\W*+.\W*+))\W*+$
2417 </pre>
2418 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2419 man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2420 the use of the possessive quantifier *+ to avoid backtracking into sequences of
2421 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2422 more) to match typical phrases, and Perl takes so long that you think it has
2423 gone into a loop.
2424 </P>
2425 <P>
2426 <b>WARNING</b>: The palindrome-matching patterns above work only if the subject
2427 string does not start with a palindrome that is shorter than the entire string.
2428 For example, although "abcba" is correctly matched, if the subject is "ababa",
2429 PCRE finds the palindrome "aba" at the start, then fails at top level because
2430 the end of the string does not follow. Once again, it cannot jump back into the
2431 recursion to try other alternatives, so the entire match fails.
2432 </P>
2433 <P>
2434 The second way in which PCRE and Perl differ in their recursion processing is
2435 in the handling of captured values. In Perl, when a subpattern is called
2436 recursively or as a subpattern (see the next section), it has no access to any
2437 values that were captured outside the recursion, whereas in PCRE these values
2438 can be referenced. Consider this pattern:
2439 <pre>
2440 ^(.)(\1|a(?2))
2441 </pre>
2442 In PCRE, this pattern matches "bab". The first capturing parentheses match "b",
2443 then in the second group, when the back reference \1 fails to match "b", the
2444 second alternative matches "a" and then recurses. In the recursion, \1 does
2445 now match "b" and so the whole match succeeds. In Perl, the pattern fails to
2446 match because inside the recursive call \1 cannot access the externally set
2447 value.
2448 <a name="subpatternsassubroutines"></a></P>
2449 <br><a name="SEC22" href="#TOC1">SUBPATTERNS AS SUBROUTINES</a><br>
2450 <P>
2451 If the syntax for a recursive subpattern call (either by number or by
2452 name) is used outside the parentheses to which it refers, it operates like a
2453 subroutine in a programming language. The called subpattern may be defined
2454 before or after the reference. A numbered reference can be absolute or
2455 relative, as in these examples:
2456 <pre>
2457 (...(absolute)...)...(?2)...
2458 (...(relative)...)...(?-1)...
2459 (...(?+1)...(relative)...
2460 </pre>
2461 An earlier example pointed out that the pattern
2462 <pre>
2463 (sens|respons)e and \1ibility
2464 </pre>
2465 matches "sense and sensibility" and "response and responsibility", but not
2466 "sense and responsibility". If instead the pattern
2467 <pre>
2468 (sens|respons)e and (?1)ibility
2469 </pre>
2470 is used, it does match "sense and responsibility" as well as the other two
2471 strings. Another example is given in the discussion of DEFINE above.
2472 </P>
2473 <P>
2474 All subroutine calls, whether recursive or not, are always treated as atomic
2475 groups. That is, once a subroutine has matched some of the subject string, it
2476 is never re-entered, even if it contains untried alternatives and there is a
2477 subsequent matching failure. Any capturing parentheses that are set during the
2478 subroutine call revert to their previous values afterwards.
2479 </P>
2480 <P>
2481 Processing options such as case-independence are fixed when a subpattern is
2482 defined, so if it is used as a subroutine, such options cannot be changed for
2483 different calls. For example, consider this pattern:
2484 <pre>
2485 (abc)(?i:(?-1))
2486 </pre>
2487 It matches "abcabc". It does not match "abcABC" because the change of
2488 processing option does not affect the called subpattern.
2489 <a name="onigurumasubroutines"></a></P>
2490 <br><a name="SEC23" href="#TOC1">ONIGURUMA SUBROUTINE SYNTAX</a><br>
2491 <P>
2492 For compatibility with Oniguruma, the non-Perl syntax \g followed by a name or
2493 a number enclosed either in angle brackets or single quotes, is an alternative
2494 syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2495 are two of the examples used above, rewritten using this syntax:
2496 <pre>
2497 (?&#60;pn&#62; \( ( (?&#62;[^()]+) | \g&#60;pn&#62; )* \) )
2498 (sens|respons)e and \g'1'ibility
2499 </pre>
2500 PCRE supports an extension to Oniguruma: if a number is preceded by a
2501 plus or a minus sign it is taken as a relative reference. For example:
2502 <pre>
2503 (abc)(?i:\g&#60;-1&#62;)
2504 </pre>
2505 Note that \g{...} (Perl syntax) and \g&#60;...&#62; (Oniguruma syntax) are <i>not</i>
2506 synonymous. The former is a back reference; the latter is a subroutine call.
2507 </P>
2508 <br><a name="SEC24" href="#TOC1">CALLOUTS</a><br>
2509 <P>
2510 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2511 code to be obeyed in the middle of matching a regular expression. This makes it
2512 possible, amongst other things, to extract different substrings that match the
2513 same pair of parentheses when there is a repetition.
2514 </P>
2515 <P>
2516 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2517 code. The feature is called "callout". The caller of PCRE provides an external
2518 function by putting its entry point in the global variable <i>pcre_callout</i>
2519 (8-bit library) or <i>pcre16_callout</i> (16-bit library). By default, this
2520 variable contains NULL, which disables all calling out.
2521 </P>
2522 <P>
2523 Within a regular expression, (?C) indicates the points at which the external
2524 function is to be called. If you want to identify different callout points, you
2525 can put a number less than 256 after the letter C. The default value is zero.
2526 For example, this pattern has two callout points:
2527 <pre>
2528 (?C1)abc(?C2)def
2529 </pre>
2530 If the PCRE_AUTO_CALLOUT flag is passed to a compiling function, callouts are
2531 automatically installed before each item in the pattern. They are all numbered
2532 255.
2533 </P>
2534 <P>
2535 During matching, when PCRE reaches a callout point, the external function is
2536 called. It is provided with the number of the callout, the position in the
2537 pattern, and, optionally, one item of data originally supplied by the caller of
2538 the matching function. The callout function may cause matching to proceed, to
2539 backtrack, or to fail altogether. A complete description of the interface to
2540 the callout function is given in the
2541 <a href="pcrecallout.html"><b>pcrecallout</b></a>
2542 documentation.
2543 <a name="backtrackcontrol"></a></P>
2544 <br><a name="SEC25" href="#TOC1">BACKTRACKING CONTROL</a><br>
2545 <P>
2546 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2547 are described in the Perl documentation as "experimental and subject to change
2548 or removal in a future version of Perl". It goes on to say: "Their usage in
2549 production code should be noted to avoid problems during upgrades." The same
2550 remarks apply to the PCRE features described in this section.
2551 </P>
2552 <P>
2553 Since these verbs are specifically related to backtracking, most of them can be
2554 used only when the pattern is to be matched using one of the traditional
2555 matching functions, which use a backtracking algorithm. With the exception of
2556 (*FAIL), which behaves like a failing negative assertion, they cause an error
2557 if encountered by a DFA matching function.
2558 </P>
2559 <P>
2560 If any of these verbs are used in an assertion or in a subpattern that is
2561 called as a subroutine (whether or not recursively), their effect is confined
2562 to that subpattern; it does not extend to the surrounding pattern, with one
2563 exception: the name from a *(MARK), (*PRUNE), or (*THEN) that is encountered in
2564 a successful positive assertion <i>is</i> passed back when a match succeeds
2565 (compare capturing parentheses in assertions). Note that such subpatterns are
2566 processed as anchored at the point where they are tested. Note also that Perl's
2567 treatment of subroutines is different in some cases.
2568 </P>
2569 <P>
2570 The new verbs make use of what was previously invalid syntax: an opening
2571 parenthesis followed by an asterisk. They are generally of the form
2572 (*VERB) or (*VERB:NAME). Some may take either form, with differing behaviour,
2573 depending on whether or not an argument is present. A name is any sequence of
2574 characters that does not include a closing parenthesis. If the name is empty,
2575 that is, if the closing parenthesis immediately follows the colon, the effect
2576 is as if the colon were not there. Any number of these verbs may occur in a
2577 pattern.
2578 <a name="nooptimize"></a></P>
2579 <br><b>
2580 Optimizations that affect backtracking verbs
2581 </b><br>
2582 <P>
2583 PCRE contains some optimizations that are used to speed up matching by running
2584 some checks at the start of each match attempt. For example, it may know the
2585 minimum length of matching subject, or that a particular character must be
2586 present. When one of these optimizations suppresses the running of a match, any
2587 included backtracking verbs will not, of course, be processed. You can suppress
2588 the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2589 when calling <b>pcre_compile()</b> or <b>pcre_exec()</b>, or by starting the
2590 pattern with (*NO_START_OPT). There is more discussion of this option in the
2591 section entitled
2592 <a href="pcreapi.html#execoptions">"Option bits for <b>pcre_exec()</b>"</a>
2593 in the
2594 <a href="pcreapi.html"><b>pcreapi</b></a>
2595 documentation.
2596 </P>
2597 <P>
2598 Experiments with Perl suggest that it too has similar optimizations, sometimes
2599 leading to anomalous results.
2600 </P>
2601 <br><b>
2602 Verbs that act immediately
2603 </b><br>
2604 <P>
2605 The following verbs act as soon as they are encountered. They may not be
2606 followed by a name.
2607 <pre>
2608 (*ACCEPT)
2609 </pre>
2610 This verb causes the match to end successfully, skipping the remainder of the
2611 pattern. However, when it is inside a subpattern that is called as a
2612 subroutine, only that subpattern is ended successfully. Matching then continues
2613 at the outer level. If (*ACCEPT) is inside capturing parentheses, the data so
2614 far is captured. For example:
2615 <pre>
2616 A((?:A|B(*ACCEPT)|C)D)
2617 </pre>
2618 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2619 the outer parentheses.
2620 <pre>
2621 (*FAIL) or (*F)
2622 </pre>
2623 This verb causes a matching failure, forcing backtracking to occur. It is
2624 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2625 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2626 Perl features that are not present in PCRE. The nearest equivalent is the
2627 callout feature, as for example in this pattern:
2628 <pre>
2629 a+(?C)(*FAIL)
2630 </pre>
2631 A match with the string "aaaa" always fails, but the callout is taken before
2632 each backtrack happens (in this example, 10 times).
2633 </P>
2634 <br><b>
2635 Recording which path was taken
2636 </b><br>
2637 <P>
2638 There is one verb whose main purpose is to track how a match was arrived at,
2639 though it also has a secondary use in conjunction with advancing the match
2640 starting point (see (*SKIP) below).
2641 <pre>
2642 (*MARK:NAME) or (*:NAME)
2643 </pre>
2644 A name is always required with this verb. There may be as many instances of
2645 (*MARK) as you like in a pattern, and their names do not have to be unique.
2646 </P>
2647 <P>
2648 When a match succeeds, the name of the last-encountered (*MARK) on the matching
2649 path is passed back to the caller as described in the section entitled
2650 <a href="pcreapi.html#extradata">"Extra data for <b>pcre_exec()</b>"</a>
2651 in the
2652 <a href="pcreapi.html"><b>pcreapi</b></a>
2653 documentation. Here is an example of <b>pcretest</b> output, where the /K
2654 modifier requests the retrieval and outputting of (*MARK) data:
2655 <pre>
2656 re&#62; /X(*MARK:A)Y|X(*MARK:B)Z/K
2657 data&#62; XY
2658 0: XY
2659 MK: A
2660 XZ
2661 0: XZ
2662 MK: B
2663 </pre>
2664 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2665 indicates which of the two alternatives matched. This is a more efficient way
2666 of obtaining this information than putting each alternative in its own
2667 capturing parentheses.
2668 </P>
2669 <P>
2670 If (*MARK) is encountered in a positive assertion, its name is recorded and
2671 passed back if it is the last-encountered. This does not happen for negative
2672 assertions.
2673 </P>
2674 <P>
2675 After a partial match or a failed match, the name of the last encountered
2676 (*MARK) in the entire match process is returned. For example:
2677 <pre>
2678 re&#62; /X(*MARK:A)Y|X(*MARK:B)Z/K
2679 data&#62; XP
2680 No match, mark = B
2681 </pre>
2682 Note that in this unanchored example the mark is retained from the match
2683 attempt that started at the letter "X" in the subject. Subsequent match
2684 attempts starting at "P" and then with an empty string do not get as far as the
2685 (*MARK) item, but nevertheless do not reset it.
2686 </P>
2687 <P>
2688 If you are interested in (*MARK) values after failed matches, you should
2689 probably set the PCRE_NO_START_OPTIMIZE option
2690 <a href="#nooptimize">(see above)</a>
2691 to ensure that the match is always attempted.
2692 </P>
2693 <br><b>
2694 Verbs that act after backtracking
2695 </b><br>
2696 <P>
2697 The following verbs do nothing when they are encountered. Matching continues
2698 with what follows, but if there is no subsequent match, causing a backtrack to
2699 the verb, a failure is forced. That is, backtracking cannot pass to the left of
2700 the verb. However, when one of these verbs appears inside an atomic group, its
2701 effect is confined to that group, because once the group has been matched,
2702 there is never any backtracking into it. In this situation, backtracking can
2703 "jump back" to the left of the entire atomic group. (Remember also, as stated
2704 above, that this localization also applies in subroutine calls and assertions.)
2705 </P>
2706 <P>
2707 These verbs differ in exactly what kind of failure occurs when backtracking
2708 reaches them.
2709 <pre>
2710 (*COMMIT)
2711 </pre>
2712 This verb, which may not be followed by a name, causes the whole match to fail
2713 outright if the rest of the pattern does not match. Even if the pattern is
2714 unanchored, no further attempts to find a match by advancing the starting point
2715 take place. Once (*COMMIT) has been passed, <b>pcre_exec()</b> is committed to
2716 finding a match at the current starting point, or not at all. For example:
2717 <pre>
2718 a+(*COMMIT)b
2719 </pre>
2720 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2721 dynamic anchor, or "I've started, so I must finish." The name of the most
2722 recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2723 match failure.
2724 </P>
2725 <P>
2726 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2727 unless PCRE's start-of-match optimizations are turned off, as shown in this
2728 <b>pcretest</b> example:
2729 <pre>
2730 re&#62; /(*COMMIT)abc/
2731 data&#62; xyzabc
2732 0: abc
2733 xyzabc\Y
2734 No match
2735 </pre>
2736 PCRE knows that any match must start with "a", so the optimization skips along
2737 the subject to "a" before running the first match attempt, which succeeds. When
2738 the optimization is disabled by the \Y escape in the second subject, the match
2739 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2740 starting points.
2741 <pre>
2742 (*PRUNE) or (*PRUNE:NAME)
2743 </pre>
2744 This verb causes the match to fail at the current starting position in the
2745 subject if the rest of the pattern does not match. If the pattern is
2746 unanchored, the normal "bumpalong" advance to the next starting character then
2747 happens. Backtracking can occur as usual to the left of (*PRUNE), before it is
2748 reached, or when matching to the right of (*PRUNE), but if there is no match to
2749 the right, backtracking cannot cross (*PRUNE). In simple cases, the use of
2750 (*PRUNE) is just an alternative to an atomic group or possessive quantifier,
2751 but there are some uses of (*PRUNE) that cannot be expressed in any other way.
2752 The behaviour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE). In an
2753 anchored pattern (*PRUNE) has the same effect as (*COMMIT).
2754 <pre>
2755 (*SKIP)
2756 </pre>
2757 This verb, when given without a name, is like (*PRUNE), except that if the
2758 pattern is unanchored, the "bumpalong" advance is not to the next character,
2759 but to the position in the subject where (*SKIP) was encountered. (*SKIP)
2760 signifies that whatever text was matched leading up to it cannot be part of a
2761 successful match. Consider:
2762 <pre>
2763 a+(*SKIP)b
2764 </pre>
2765 If the subject is "aaaac...", after the first match attempt fails (starting at
2766 the first character in the string), the starting point skips on to start the
2767 next attempt at "c". Note that a possessive quantifer does not have the same
2768 effect as this example; although it would suppress backtracking during the
2769 first match attempt, the second attempt would start at the second character
2770 instead of skipping on to "c".
2771 <pre>
2772 (*SKIP:NAME)
2773 </pre>
2774 When (*SKIP) has an associated name, its behaviour is modified. If the
2775 following pattern fails to match, the previous path through the pattern is
2776 searched for the most recent (*MARK) that has the same name. If one is found,
2777 the "bumpalong" advance is to the subject position that corresponds to that
2778 (*MARK) instead of to where (*SKIP) was encountered. If no (*MARK) with a
2779 matching name is found, the (*SKIP) is ignored.
2780 <pre>
2781 (*THEN) or (*THEN:NAME)
2782 </pre>
2783 This verb causes a skip to the next innermost alternative if the rest of the
2784 pattern does not match. That is, it cancels pending backtracking, but only
2785 within the current alternative. Its name comes from the observation that it can
2786 be used for a pattern-based if-then-else block:
2787 <pre>
2788 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2789 </pre>
2790 If the COND1 pattern matches, FOO is tried (and possibly further items after
2791 the end of the group if FOO succeeds); on failure, the matcher skips to the
2792 second alternative and tries COND2, without backtracking into COND1. The
2793 behaviour of (*THEN:NAME) is exactly the same as (*MARK:NAME)(*THEN).
2794 If (*THEN) is not inside an alternation, it acts like (*PRUNE).
2795 </P>
2796 <P>
2797 Note that a subpattern that does not contain a | character is just a part of
2798 the enclosing alternative; it is not a nested alternation with only one
2799 alternative. The effect of (*THEN) extends beyond such a subpattern to the
2800 enclosing alternative. Consider this pattern, where A, B, etc. are complex
2801 pattern fragments that do not contain any | characters at this level:
2802 <pre>
2803 A (B(*THEN)C) | D
2804 </pre>
2805 If A and B are matched, but there is a failure in C, matching does not
2806 backtrack into A; instead it moves to the next alternative, that is, D.
2807 However, if the subpattern containing (*THEN) is given an alternative, it
2808 behaves differently:
2809 <pre>
2810 A (B(*THEN)C | (*FAIL)) | D
2811 </pre>
2812 The effect of (*THEN) is now confined to the inner subpattern. After a failure
2813 in C, matching moves to (*FAIL), which causes the whole subpattern to fail
2814 because there are no more alternatives to try. In this case, matching does now
2815 backtrack into A.
2816 </P>
2817 <P>
2818 Note also that a conditional subpattern is not considered as having two
2819 alternatives, because only one is ever used. In other words, the | character in
2820 a conditional subpattern has a different meaning. Ignoring white space,
2821 consider:
2822 <pre>
2823 ^.*? (?(?=a) a | b(*THEN)c )
2824 </pre>
2825 If the subject is "ba", this pattern does not match. Because .*? is ungreedy,
2826 it initially matches zero characters. The condition (?=a) then fails, the
2827 character "b" is matched, but "c" is not. At this point, matching does not
2828 backtrack to .*? as might perhaps be expected from the presence of the |
2829 character. The conditional subpattern is part of the single alternative that
2830 comprises the whole pattern, and so the match fails. (If there was a backtrack
2831 into .*?, allowing it to match "b", the match would succeed.)
2832 </P>
2833 <P>
2834 The verbs just described provide four different "strengths" of control when
2835 subsequent matching fails. (*THEN) is the weakest, carrying on the match at the
2836 next alternative. (*PRUNE) comes next, failing the match at the current
2837 starting position, but allowing an advance to the next character (for an
2838 unanchored pattern). (*SKIP) is similar, except that the advance may be more
2839 than one character. (*COMMIT) is the strongest, causing the entire match to
2840 fail.
2841 </P>
2842 <P>
2843 If more than one such verb is present in a pattern, the "strongest" one wins.
2844 For example, consider this pattern, where A, B, etc. are complex pattern
2845 fragments:
2846 <pre>
2847 (A(*COMMIT)B(*THEN)C|D)
2848 </pre>
2849 Once A has matched, PCRE is committed to this match, at the current starting
2850 position. If subsequently B matches, but C does not, the normal (*THEN) action
2851 of trying the next alternative (that is, D) does not happen because (*COMMIT)
2852 overrides.
2853 </P>
2854 <br><a name="SEC26" href="#TOC1">SEE ALSO</a><br>
2855 <P>
2856 <b>pcreapi</b>(3), <b>pcrecallout</b>(3), <b>pcrematching</b>(3),
2857 <b>pcresyntax</b>(3), <b>pcre</b>(3), <b>pcre16(3)</b>.
2858 </P>
2859 <br><a name="SEC27" href="#TOC1">AUTHOR</a><br>
2860 <P>
2861 Philip Hazel
2862 <br>
2863 University Computing Service
2864 <br>
2865 Cambridge CB2 3QH, England.
2866 <br>
2867 </P>
2868 <br><a name="SEC28" href="#TOC1">REVISION</a><br>
2869 <P>
2870 Last updated: 24 February 2012
2871 <br>
2872 Copyright &copy; 1997-2012 University of Cambridge.
2873 <br>
2874 <p>
2875 Return to the <a href="index.html">PCRE index page</a>.
2876 </p>

Properties

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

  ViewVC Help
Powered by ViewVC 1.1.5