ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 1298 - (show annotations)
Fri Mar 22 16:13:13 2013 UTC (6 years, 5 months ago) by ph10
File MIME type: text/html
File size: 132035 byte(s)
Fix COMMIT in recursion; document backtracking verbs in assertions and 
1 <html>
2 <head>
3 <title>pcrepattern specification</title>
4 </head>
5 <body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB">
6 <h1>pcrepattern man page</h1>
7 <p>
8 Return to the <a href="index.html">PCRE index page</a>.
9 </p>
10 <p>
11 This page is part of the PCRE HTML documentation. It was generated automatically
12 from the original man page. If there is any nonsense in it, please consult the
13 man page, in case the conversion went wrong.
14 <br>
15 <ul>
16 <li><a name="TOC1" href="#SEC1">PCRE REGULAR EXPRESSION DETAILS</a>
17 <li><a name="TOC2" href="#SEC2">EBCDIC CHARACTER CODES</a>
18 <li><a name="TOC3" href="#SEC3">NEWLINE CONVENTIONS</a>
19 <li><a name="TOC4" href="#SEC4">CHARACTERS AND METACHARACTERS</a>
20 <li><a name="TOC5" href="#SEC5">BACKSLASH</a>
21 <li><a name="TOC6" href="#SEC6">CIRCUMFLEX AND DOLLAR</a>
22 <li><a name="TOC7" href="#SEC7">FULL STOP (PERIOD, DOT) AND \N</a>
23 <li><a name="TOC8" href="#SEC8">MATCHING A SINGLE DATA UNIT</a>
25 <li><a name="TOC10" href="#SEC10">POSIX CHARACTER CLASSES</a>
26 <li><a name="TOC11" href="#SEC11">VERTICAL BAR</a>
27 <li><a name="TOC12" href="#SEC12">INTERNAL OPTION SETTING</a>
28 <li><a name="TOC13" href="#SEC13">SUBPATTERNS</a>
29 <li><a name="TOC14" href="#SEC14">DUPLICATE SUBPATTERN NUMBERS</a>
30 <li><a name="TOC15" href="#SEC15">NAMED SUBPATTERNS</a>
31 <li><a name="TOC16" href="#SEC16">REPETITION</a>
33 <li><a name="TOC18" href="#SEC18">BACK REFERENCES</a>
34 <li><a name="TOC19" href="#SEC19">ASSERTIONS</a>
35 <li><a name="TOC20" href="#SEC20">CONDITIONAL SUBPATTERNS</a>
36 <li><a name="TOC21" href="#SEC21">COMMENTS</a>
37 <li><a name="TOC22" href="#SEC22">RECURSIVE PATTERNS</a>
38 <li><a name="TOC23" href="#SEC23">SUBPATTERNS AS SUBROUTINES</a>
39 <li><a name="TOC24" href="#SEC24">ONIGURUMA SUBROUTINE SYNTAX</a>
40 <li><a name="TOC25" href="#SEC25">CALLOUTS</a>
41 <li><a name="TOC26" href="#SEC26">BACKTRACKING CONTROL</a>
42 <li><a name="TOC27" href="#SEC27">SEE ALSO</a>
43 <li><a name="TOC28" href="#SEC28">AUTHOR</a>
44 <li><a name="TOC29" href="#SEC29">REVISION</a>
45 </ul>
46 <br><a name="SEC1" href="#TOC1">PCRE REGULAR EXPRESSION DETAILS</a><br>
47 <P>
48 The syntax and semantics of the regular expressions that are supported by PCRE
49 are described in detail below. There is a quick-reference syntax summary in the
50 <a href="pcresyntax.html"><b>pcresyntax</b></a>
51 page. PCRE tries to match Perl syntax and semantics as closely as it can. PCRE
52 also supports some alternative regular expression syntax (which does not
53 conflict with the Perl syntax) in order to provide some compatibility with
54 regular expressions in Python, .NET, and Oniguruma.
55 </P>
56 <P>
57 Perl's regular expressions are described in its own documentation, and
58 regular expressions in general are covered in a number of books, some of which
59 have copious examples. Jeffrey Friedl's "Mastering Regular Expressions",
60 published by O'Reilly, covers regular expressions in great detail. This
61 description of PCRE's regular expressions is intended as reference material.
62 </P>
63 <P>
64 The original operation of PCRE was on strings of one-byte characters. However,
65 there is now also support for UTF-8 strings in the original library, an
66 extra library that supports 16-bit and UTF-16 character strings, and a
67 third library that supports 32-bit and UTF-32 character strings. To use these
68 features, PCRE must be built to include appropriate support. When using UTF
69 strings you must either call the compiling function with the PCRE_UTF8,
70 PCRE_UTF16, or PCRE_UTF32 option, or the pattern must start with one of
71 these special sequences:
72 <pre>
73 (*UTF8)
74 (*UTF16)
75 (*UTF32)
76 (*UTF)
77 </pre>
78 (*UTF) is a generic sequence that can be used with any of the libraries.
79 Starting a pattern with such a sequence is equivalent to setting the relevant
80 option. This feature is not Perl-compatible. How setting a UTF mode affects
81 pattern matching is mentioned in several places below. There is also a summary
82 of features in the
83 <a href="pcreunicode.html"><b>pcreunicode</b></a>
84 page.
85 </P>
86 <P>
87 Another special sequence that may appear at the start of a pattern or in
88 combination with (*UTF8), (*UTF16), (*UTF32) or (*UTF) is:
89 <pre>
90 (*UCP)
91 </pre>
92 This has the same effect as setting the PCRE_UCP option: it causes sequences
93 such as \d and \w to use Unicode properties to determine character types,
94 instead of recognizing only characters with codes less than 128 via a lookup
95 table.
96 </P>
97 <P>
98 If a pattern starts with (*NO_START_OPT), it has the same effect as setting the
99 PCRE_NO_START_OPTIMIZE option either at compile or matching time. There are
100 also some more of these special sequences that are concerned with the handling
101 of newlines; they are described below.
102 </P>
103 <P>
104 The remainder of this document discusses the patterns that are supported by
105 PCRE when one its main matching functions, <b>pcre_exec()</b> (8-bit) or
106 <b>pcre[16|32]_exec()</b> (16- or 32-bit), is used. PCRE also has alternative
107 matching functions, <b>pcre_dfa_exec()</b> and <b>pcre[16|32_dfa_exec()</b>,
108 which match using a different algorithm that is not Perl-compatible. Some of
109 the features discussed below are not available when DFA matching is used. The
110 advantages and disadvantages of the alternative functions, and how they differ
111 from the normal functions, are discussed in the
112 <a href="pcrematching.html"><b>pcrematching</b></a>
113 page.
114 </P>
115 <br><a name="SEC2" href="#TOC1">EBCDIC CHARACTER CODES</a><br>
116 <P>
117 PCRE can be compiled to run in an environment that uses EBCDIC as its character
118 code rather than ASCII or Unicode (typically a mainframe system). In the
119 sections below, character code values are ASCII or Unicode; in an EBCDIC
120 environment these characters may have different code values, and there are no
121 code points greater than 255.
122 <a name="newlines"></a></P>
123 <br><a name="SEC3" href="#TOC1">NEWLINE CONVENTIONS</a><br>
124 <P>
125 PCRE supports five different conventions for indicating line breaks in
126 strings: a single CR (carriage return) character, a single LF (linefeed)
127 character, the two-character sequence CRLF, any of the three preceding, or any
128 Unicode newline sequence. The
129 <a href="pcreapi.html"><b>pcreapi</b></a>
130 page has
131 <a href="pcreapi.html#newlines">further discussion</a>
132 about newlines, and shows how to set the newline convention in the
133 <i>options</i> arguments for the compiling and matching functions.
134 </P>
135 <P>
136 It is also possible to specify a newline convention by starting a pattern
137 string with one of the following five sequences:
138 <pre>
139 (*CR) carriage return
140 (*LF) linefeed
141 (*CRLF) carriage return, followed by linefeed
142 (*ANYCRLF) any of the three above
143 (*ANY) all Unicode newline sequences
144 </pre>
145 These override the default and the options given to the compiling function. For
146 example, on a Unix system where LF is the default newline sequence, the pattern
147 <pre>
148 (*CR)a.b
149 </pre>
150 changes the convention to CR. That pattern matches "a\nb" because LF is no
151 longer a newline. Note that these special settings, which are not
152 Perl-compatible, are recognized only at the very start of a pattern, and that
153 they must be in upper case. If more than one of them is present, the last one
154 is used.
155 </P>
156 <P>
157 The newline convention affects where the circumflex and dollar assertions are
158 true. It also affects the interpretation of the dot metacharacter when
159 PCRE_DOTALL is not set, and the behaviour of \N. However, it does not affect
160 what the \R escape sequence matches. By default, this is any Unicode newline
161 sequence, for Perl compatibility. However, this can be changed; see the
162 description of \R in the section entitled
163 <a href="#newlineseq">"Newline sequences"</a>
164 below. A change of \R setting can be combined with a change of newline
165 convention.
166 </P>
167 <br><a name="SEC4" href="#TOC1">CHARACTERS AND METACHARACTERS</a><br>
168 <P>
169 A regular expression is a pattern that is matched against a subject string from
170 left to right. Most characters stand for themselves in a pattern, and match the
171 corresponding characters in the subject. As a trivial example, the pattern
172 <pre>
173 The quick brown fox
174 </pre>
175 matches a portion of a subject string that is identical to itself. When
176 caseless matching is specified (the PCRE_CASELESS option), letters are matched
177 independently of case. In a UTF mode, PCRE always understands the concept of
178 case for characters whose values are less than 128, so caseless matching is
179 always possible. For characters with higher values, the concept of case is
180 supported if PCRE is compiled with Unicode property support, but not otherwise.
181 If you want to use caseless matching for characters 128 and above, you must
182 ensure that PCRE is compiled with Unicode property support as well as with
183 UTF support.
184 </P>
185 <P>
186 The power of regular expressions comes from the ability to include alternatives
187 and repetitions in the pattern. These are encoded in the pattern by the use of
188 <i>metacharacters</i>, which do not stand for themselves but instead are
189 interpreted in some special way.
190 </P>
191 <P>
192 There are two different sets of metacharacters: those that are recognized
193 anywhere in the pattern except within square brackets, and those that are
194 recognized within square brackets. Outside square brackets, the metacharacters
195 are as follows:
196 <pre>
197 \ general escape character with several uses
198 ^ assert start of string (or line, in multiline mode)
199 $ assert end of string (or line, in multiline mode)
200 . match any character except newline (by default)
201 [ start character class definition
202 | start of alternative branch
203 ( start subpattern
204 ) end subpattern
205 ? extends the meaning of (
206 also 0 or 1 quantifier
207 also quantifier minimizer
208 * 0 or more quantifier
209 + 1 or more quantifier
210 also "possessive quantifier"
211 { start min/max quantifier
212 </pre>
213 Part of a pattern that is in square brackets is called a "character class". In
214 a character class the only metacharacters are:
215 <pre>
216 \ general escape character
217 ^ negate the class, but only if the first character
218 - indicates character range
219 [ POSIX character class (only if followed by POSIX syntax)
220 ] terminates the character class
221 </pre>
222 The following sections describe the use of each of the metacharacters.
223 </P>
224 <br><a name="SEC5" href="#TOC1">BACKSLASH</a><br>
225 <P>
226 The backslash character has several uses. Firstly, if it is followed by a
227 character that is not a number or a letter, it takes away any special meaning
228 that character may have. This use of backslash as an escape character applies
229 both inside and outside character classes.
230 </P>
231 <P>
232 For example, if you want to match a * character, you write \* in the pattern.
233 This escaping action applies whether or not the following character would
234 otherwise be interpreted as a metacharacter, so it is always safe to precede a
235 non-alphanumeric with backslash to specify that it stands for itself. In
236 particular, if you want to match a backslash, you write \\.
237 </P>
238 <P>
239 In a UTF mode, only ASCII numbers and letters have any special meaning after a
240 backslash. All other characters (in particular, those whose codepoints are
241 greater than 127) are treated as literals.
242 </P>
243 <P>
244 If a pattern is compiled with the PCRE_EXTENDED option, white space in the
245 pattern (other than in a character class) and characters between a # outside
246 a character class and the next newline are ignored. An escaping backslash can
247 be used to include a white space or # character as part of the pattern.
248 </P>
249 <P>
250 If you want to remove the special meaning from a sequence of characters, you
251 can do so by putting them between \Q and \E. This is different from Perl in
252 that $ and @ are handled as literals in \Q...\E sequences in PCRE, whereas in
253 Perl, $ and @ cause variable interpolation. Note the following examples:
254 <pre>
255 Pattern PCRE matches Perl matches
257 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
258 \Qabc\$xyz\E abc\$xyz abc\$xyz
259 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
260 </pre>
261 The \Q...\E sequence is recognized both inside and outside character classes.
262 An isolated \E that is not preceded by \Q is ignored. If \Q is not followed
263 by \E later in the pattern, the literal interpretation continues to the end of
264 the pattern (that is, \E is assumed at the end). If the isolated \Q is inside
265 a character class, this causes an error, because the character class is not
266 terminated.
267 <a name="digitsafterbackslash"></a></P>
268 <br><b>
269 Non-printing characters
270 </b><br>
271 <P>
272 A second use of backslash provides a way of encoding non-printing characters
273 in patterns in a visible manner. There is no restriction on the appearance of
274 non-printing characters, apart from the binary zero that terminates a pattern,
275 but when a pattern is being prepared by text editing, it is often easier to use
276 one of the following escape sequences than the binary character it represents:
277 <pre>
278 \a alarm, that is, the BEL character (hex 07)
279 \cx "control-x", where x is any ASCII character
280 \e escape (hex 1B)
281 \f form feed (hex 0C)
282 \n linefeed (hex 0A)
283 \r carriage return (hex 0D)
284 \t tab (hex 09)
285 \ddd character with octal code ddd, or back reference
286 \xhh character with hex code hh
287 \x{hhh..} character with hex code hhh.. (non-JavaScript mode)
288 \uhhhh character with hex code hhhh (JavaScript mode only)
289 </pre>
290 The precise effect of \cx on ASCII characters is as follows: if x is a lower
291 case letter, it is converted to upper case. Then bit 6 of the character (hex
292 40) is inverted. Thus \cA to \cZ become hex 01 to hex 1A (A is 41, Z is 5A),
293 but \c{ becomes hex 3B ({ is 7B), and \c; becomes hex 7B (; is 3B). If the
294 data item (byte or 16-bit value) following \c has a value greater than 127, a
295 compile-time error occurs. This locks out non-ASCII characters in all modes.
296 </P>
297 <P>
298 The \c facility was designed for use with ASCII characters, but with the
299 extension to Unicode it is even less useful than it once was. It is, however,
300 recognized when PCRE is compiled in EBCDIC mode, where data items are always
301 bytes. In this mode, all values are valid after \c. If the next character is a
302 lower case letter, it is converted to upper case. Then the 0xc0 bits of the
303 byte are inverted. Thus \cA becomes hex 01, as in ASCII (A is C1), but because
304 the EBCDIC letters are disjoint, \cZ becomes hex 29 (Z is E9), and other
305 characters also generate different values.
306 </P>
307 <P>
308 By default, after \x, from zero to two hexadecimal digits are read (letters
309 can be in upper or lower case). Any number of hexadecimal digits may appear
310 between \x{ and }, but the character code is constrained as follows:
311 <pre>
312 8-bit non-UTF mode less than 0x100
313 8-bit UTF-8 mode less than 0x10ffff and a valid codepoint
314 16-bit non-UTF mode less than 0x10000
315 16-bit UTF-16 mode less than 0x10ffff and a valid codepoint
316 32-bit non-UTF mode less than 0x80000000
317 32-bit UTF-32 mode less than 0x10ffff and a valid codepoint
318 </pre>
319 Invalid Unicode codepoints are the range 0xd800 to 0xdfff (the so-called
320 "surrogate" codepoints), and 0xffef.
321 </P>
322 <P>
323 If characters other than hexadecimal digits appear between \x{ and }, or if
324 there is no terminating }, this form of escape is not recognized. Instead, the
325 initial \x will be interpreted as a basic hexadecimal escape, with no
326 following digits, giving a character whose value is zero.
327 </P>
328 <P>
329 If the PCRE_JAVASCRIPT_COMPAT option is set, the interpretation of \x is
330 as just described only when it is followed by two hexadecimal digits.
331 Otherwise, it matches a literal "x" character. In JavaScript mode, support for
332 code points greater than 256 is provided by \u, which must be followed by
333 four hexadecimal digits; otherwise it matches a literal "u" character.
334 Character codes specified by \u in JavaScript mode are constrained in the same
335 was as those specified by \x in non-JavaScript mode.
336 </P>
337 <P>
338 Characters whose value is less than 256 can be defined by either of the two
339 syntaxes for \x (or by \u in JavaScript mode). There is no difference in the
340 way they are handled. For example, \xdc is exactly the same as \x{dc} (or
341 \u00dc in JavaScript mode).
342 </P>
343 <P>
344 After \0 up to two further octal digits are read. If there are fewer than two
345 digits, just those that are present are used. Thus the sequence \0\x\07
346 specifies two binary zeros followed by a BEL character (code value 7). Make
347 sure you supply two digits after the initial zero if the pattern character that
348 follows is itself an octal digit.
349 </P>
350 <P>
351 The handling of a backslash followed by a digit other than 0 is complicated.
352 Outside a character class, PCRE reads it and any following digits as a decimal
353 number. If the number is less than 10, or if there have been at least that many
354 previous capturing left parentheses in the expression, the entire sequence is
355 taken as a <i>back reference</i>. A description of how this works is given
356 <a href="#backreferences">later,</a>
357 following the discussion of
358 <a href="#subpattern">parenthesized subpatterns.</a>
359 </P>
360 <P>
361 Inside a character class, or if the decimal number is greater than 9 and there
362 have not been that many capturing subpatterns, PCRE re-reads up to three octal
363 digits following the backslash, and uses them to generate a data character. Any
364 subsequent digits stand for themselves. The value of the character is
365 constrained in the same way as characters specified in hexadecimal.
366 For example:
367 <pre>
368 \040 is another way of writing an ASCII space
369 \40 is the same, provided there are fewer than 40 previous capturing subpatterns
370 \7 is always a back reference
371 \11 might be a back reference, or another way of writing a tab
372 \011 is always a tab
373 \0113 is a tab followed by the character "3"
374 \113 might be a back reference, otherwise the character with octal code 113
375 \377 might be a back reference, otherwise the value 255 (decimal)
376 \81 is either a back reference, or a binary zero followed by the two characters "8" and "1"
377 </pre>
378 Note that octal values of 100 or greater must not be introduced by a leading
379 zero, because no more than three octal digits are ever read.
380 </P>
381 <P>
382 All the sequences that define a single character value can be used both inside
383 and outside character classes. In addition, inside a character class, \b is
384 interpreted as the backspace character (hex 08).
385 </P>
386 <P>
387 \N is not allowed in a character class. \B, \R, and \X are not special
388 inside a character class. Like other unrecognized escape sequences, they are
389 treated as the literal characters "B", "R", and "X" by default, but cause an
390 error if the PCRE_EXTRA option is set. Outside a character class, these
391 sequences have different meanings.
392 </P>
393 <br><b>
394 Unsupported escape sequences
395 </b><br>
396 <P>
397 In Perl, the sequences \l, \L, \u, and \U are recognized by its string
398 handler and used to modify the case of following characters. By default, PCRE
399 does not support these escape sequences. However, if the PCRE_JAVASCRIPT_COMPAT
400 option is set, \U matches a "U" character, and \u can be used to define a
401 character by code point, as described in the previous section.
402 </P>
403 <br><b>
404 Absolute and relative back references
405 </b><br>
406 <P>
407 The sequence \g followed by an unsigned or a negative number, optionally
408 enclosed in braces, is an absolute or relative back reference. A named back
409 reference can be coded as \g{name}. Back references are discussed
410 <a href="#backreferences">later,</a>
411 following the discussion of
412 <a href="#subpattern">parenthesized subpatterns.</a>
413 </P>
414 <br><b>
415 Absolute and relative subroutine calls
416 </b><br>
417 <P>
418 For compatibility with Oniguruma, the non-Perl syntax \g followed by a name or
419 a number enclosed either in angle brackets or single quotes, is an alternative
420 syntax for referencing a subpattern as a "subroutine". Details are discussed
421 <a href="#onigurumasubroutines">later.</a>
422 Note that \g{...} (Perl syntax) and \g&#60;...&#62; (Oniguruma syntax) are <i>not</i>
423 synonymous. The former is a back reference; the latter is a
424 <a href="#subpatternsassubroutines">subroutine</a>
425 call.
426 <a name="genericchartypes"></a></P>
427 <br><b>
428 Generic character types
429 </b><br>
430 <P>
431 Another use of backslash is for specifying generic character types:
432 <pre>
433 \d any decimal digit
434 \D any character that is not a decimal digit
435 \h any horizontal white space character
436 \H any character that is not a horizontal white space character
437 \s any white space character
438 \S any character that is not a white space character
439 \v any vertical white space character
440 \V any character that is not a vertical white space character
441 \w any "word" character
442 \W any "non-word" character
443 </pre>
444 There is also the single sequence \N, which matches a non-newline character.
445 This is the same as
446 <a href="#fullstopdot">the "." metacharacter</a>
447 when PCRE_DOTALL is not set. Perl also uses \N to match characters by name;
448 PCRE does not support this.
449 </P>
450 <P>
451 Each pair of lower and upper case escape sequences partitions the complete set
452 of characters into two disjoint sets. Any given character matches one, and only
453 one, of each pair. The sequences can appear both inside and outside character
454 classes. They each match one character of the appropriate type. If the current
455 matching point is at the end of the subject string, all of them fail, because
456 there is no character to match.
457 </P>
458 <P>
459 For compatibility with Perl, \s does not match the VT character (code 11).
460 This makes it different from the the POSIX "space" class. The \s characters
461 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
462 included in a Perl script, \s may match the VT character. In PCRE, it never
463 does.
464 </P>
465 <P>
466 A "word" character is an underscore or any character that is a letter or digit.
467 By default, the definition of letters and digits is controlled by PCRE's
468 low-valued character tables, and may vary if locale-specific matching is taking
469 place (see
470 <a href="pcreapi.html#localesupport">"Locale support"</a>
471 in the
472 <a href="pcreapi.html"><b>pcreapi</b></a>
473 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
474 or "french" in Windows, some character codes greater than 128 are used for
475 accented letters, and these are then matched by \w. The use of locales with
476 Unicode is discouraged.
477 </P>
478 <P>
479 By default, in a UTF mode, characters with values greater than 128 never match
480 \d, \s, or \w, and always match \D, \S, and \W. These sequences retain
481 their original meanings from before UTF support was available, mainly for
482 efficiency reasons. However, if PCRE is compiled with Unicode property support,
483 and the PCRE_UCP option is set, the behaviour is changed so that Unicode
484 properties are used to determine character types, as follows:
485 <pre>
486 \d any character that \p{Nd} matches (decimal digit)
487 \s any character that \p{Z} matches, plus HT, LF, FF, CR
488 \w any character that \p{L} or \p{N} matches, plus underscore
489 </pre>
490 The upper case escapes match the inverse sets of characters. Note that \d
491 matches only decimal digits, whereas \w matches any Unicode digit, as well as
492 any Unicode letter, and underscore. Note also that PCRE_UCP affects \b, and
493 \B because they are defined in terms of \w and \W. Matching these sequences
494 is noticeably slower when PCRE_UCP is set.
495 </P>
496 <P>
497 The sequences \h, \H, \v, and \V are features that were added to Perl at
498 release 5.10. In contrast to the other sequences, which match only ASCII
499 characters by default, these always match certain high-valued codepoints,
500 whether or not PCRE_UCP is set. The horizontal space characters are:
501 <pre>
502 U+0009 Horizontal tab (HT)
503 U+0020 Space
504 U+00A0 Non-break space
505 U+1680 Ogham space mark
506 U+180E Mongolian vowel separator
507 U+2000 En quad
508 U+2001 Em quad
509 U+2002 En space
510 U+2003 Em space
511 U+2004 Three-per-em space
512 U+2005 Four-per-em space
513 U+2006 Six-per-em space
514 U+2007 Figure space
515 U+2008 Punctuation space
516 U+2009 Thin space
517 U+200A Hair space
518 U+202F Narrow no-break space
519 U+205F Medium mathematical space
520 U+3000 Ideographic space
521 </pre>
522 The vertical space characters are:
523 <pre>
524 U+000A Linefeed (LF)
525 U+000B Vertical tab (VT)
526 U+000C Form feed (FF)
527 U+000D Carriage return (CR)
528 U+0085 Next line (NEL)
529 U+2028 Line separator
530 U+2029 Paragraph separator
531 </pre>
532 In 8-bit, non-UTF-8 mode, only the characters with codepoints less than 256 are
533 relevant.
534 <a name="newlineseq"></a></P>
535 <br><b>
536 Newline sequences
537 </b><br>
538 <P>
539 Outside a character class, by default, the escape sequence \R matches any
540 Unicode newline sequence. In 8-bit non-UTF-8 mode \R is equivalent to the
541 following:
542 <pre>
543 (?&#62;\r\n|\n|\x0b|\f|\r|\x85)
544 </pre>
545 This is an example of an "atomic group", details of which are given
546 <a href="#atomicgroup">below.</a>
547 This particular group matches either the two-character sequence CR followed by
548 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
549 U+000B), FF (form feed, U+000C), CR (carriage return, U+000D), or NEL (next
550 line, U+0085). The two-character sequence is treated as a single unit that
551 cannot be split.
552 </P>
553 <P>
554 In other modes, two additional characters whose codepoints are greater than 255
555 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
556 Unicode character property support is not needed for these characters to be
557 recognized.
558 </P>
559 <P>
560 It is possible to restrict \R to match only CR, LF, or CRLF (instead of the
561 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
562 either at compile time or when the pattern is matched. (BSR is an abbrevation
563 for "backslash R".) This can be made the default when PCRE is built; if this is
564 the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
565 It is also possible to specify these settings by starting a pattern string with
566 one of the following sequences:
567 <pre>
568 (*BSR_ANYCRLF) CR, LF, or CRLF only
569 (*BSR_UNICODE) any Unicode newline sequence
570 </pre>
571 These override the default and the options given to the compiling function, but
572 they can themselves be overridden by options given to a matching function. Note
573 that these special settings, which are not Perl-compatible, are recognized only
574 at the very start of a pattern, and that they must be in upper case. If more
575 than one of them is present, the last one is used. They can be combined with a
576 change of newline convention; for example, a pattern can start with:
577 <pre>
579 </pre>
580 They can also be combined with the (*UTF8), (*UTF16), (*UTF32), (*UTF) or
581 (*UCP) special sequences. Inside a character class, \R is treated as an
582 unrecognized escape sequence, and so matches the letter "R" by default, but
583 causes an error if PCRE_EXTRA is set.
584 <a name="uniextseq"></a></P>
585 <br><b>
586 Unicode character properties
587 </b><br>
588 <P>
589 When PCRE is built with Unicode character property support, three additional
590 escape sequences that match characters with specific properties are available.
591 When in 8-bit non-UTF-8 mode, these sequences are of course limited to testing
592 characters whose codepoints are less than 256, but they do work in this mode.
593 The extra escape sequences are:
594 <pre>
595 \p{<i>xx</i>} a character with the <i>xx</i> property
596 \P{<i>xx</i>} a character without the <i>xx</i> property
597 \X a Unicode extended grapheme cluster
598 </pre>
599 The property names represented by <i>xx</i> above are limited to the Unicode
600 script names, the general category properties, "Any", which matches any
601 character (including newline), and some special PCRE properties (described
602 in the
603 <a href="#extraprops">next section).</a>
604 Other Perl properties such as "InMusicalSymbols" are not currently supported by
605 PCRE. Note that \P{Any} does not match any characters, so always causes a
606 match failure.
607 </P>
608 <P>
609 Sets of Unicode characters are defined as belonging to certain scripts. A
610 character from one of these sets can be matched using a script name. For
611 example:
612 <pre>
613 \p{Greek}
614 \P{Han}
615 </pre>
616 Those that are not part of an identified script are lumped together as
617 "Common". The current list of scripts is:
618 </P>
619 <P>
620 Arabic,
621 Armenian,
622 Avestan,
623 Balinese,
624 Bamum,
625 Batak,
626 Bengali,
627 Bopomofo,
628 Brahmi,
629 Braille,
630 Buginese,
631 Buhid,
632 Canadian_Aboriginal,
633 Carian,
634 Chakma,
635 Cham,
636 Cherokee,
637 Common,
638 Coptic,
639 Cuneiform,
640 Cypriot,
641 Cyrillic,
642 Deseret,
643 Devanagari,
644 Egyptian_Hieroglyphs,
645 Ethiopic,
646 Georgian,
647 Glagolitic,
648 Gothic,
649 Greek,
650 Gujarati,
651 Gurmukhi,
652 Han,
653 Hangul,
654 Hanunoo,
655 Hebrew,
656 Hiragana,
657 Imperial_Aramaic,
658 Inherited,
659 Inscriptional_Pahlavi,
660 Inscriptional_Parthian,
661 Javanese,
662 Kaithi,
663 Kannada,
664 Katakana,
665 Kayah_Li,
666 Kharoshthi,
667 Khmer,
668 Lao,
669 Latin,
670 Lepcha,
671 Limbu,
672 Linear_B,
673 Lisu,
674 Lycian,
675 Lydian,
676 Malayalam,
677 Mandaic,
678 Meetei_Mayek,
679 Meroitic_Cursive,
680 Meroitic_Hieroglyphs,
681 Miao,
682 Mongolian,
683 Myanmar,
684 New_Tai_Lue,
685 Nko,
686 Ogham,
687 Old_Italic,
688 Old_Persian,
689 Old_South_Arabian,
690 Old_Turkic,
691 Ol_Chiki,
692 Oriya,
693 Osmanya,
694 Phags_Pa,
695 Phoenician,
696 Rejang,
697 Runic,
698 Samaritan,
699 Saurashtra,
700 Sharada,
701 Shavian,
702 Sinhala,
703 Sora_Sompeng,
704 Sundanese,
705 Syloti_Nagri,
706 Syriac,
707 Tagalog,
708 Tagbanwa,
709 Tai_Le,
710 Tai_Tham,
711 Tai_Viet,
712 Takri,
713 Tamil,
714 Telugu,
715 Thaana,
716 Thai,
717 Tibetan,
718 Tifinagh,
719 Ugaritic,
720 Vai,
721 Yi.
722 </P>
723 <P>
724 Each character has exactly one Unicode general category property, specified by
725 a two-letter abbreviation. For compatibility with Perl, negation can be
726 specified by including a circumflex between the opening brace and the property
727 name. For example, \p{^Lu} is the same as \P{Lu}.
728 </P>
729 <P>
730 If only one letter is specified with \p or \P, it includes all the general
731 category properties that start with that letter. In this case, in the absence
732 of negation, the curly brackets in the escape sequence are optional; these two
733 examples have the same effect:
734 <pre>
735 \p{L}
736 \pL
737 </pre>
738 The following general category property codes are supported:
739 <pre>
740 C Other
741 Cc Control
742 Cf Format
743 Cn Unassigned
744 Co Private use
745 Cs Surrogate
747 L Letter
748 Ll Lower case letter
749 Lm Modifier letter
750 Lo Other letter
751 Lt Title case letter
752 Lu Upper case letter
754 M Mark
755 Mc Spacing mark
756 Me Enclosing mark
757 Mn Non-spacing mark
759 N Number
760 Nd Decimal number
761 Nl Letter number
762 No Other number
764 P Punctuation
765 Pc Connector punctuation
766 Pd Dash punctuation
767 Pe Close punctuation
768 Pf Final punctuation
769 Pi Initial punctuation
770 Po Other punctuation
771 Ps Open punctuation
773 S Symbol
774 Sc Currency symbol
775 Sk Modifier symbol
776 Sm Mathematical symbol
777 So Other symbol
779 Z Separator
780 Zl Line separator
781 Zp Paragraph separator
782 Zs Space separator
783 </pre>
784 The special property L& is also supported: it matches a character that has
785 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
786 a modifier or "other".
787 </P>
788 <P>
789 The Cs (Surrogate) property applies only to characters in the range U+D800 to
790 U+DFFF. Such characters are not valid in Unicode strings and so
791 cannot be tested by PCRE, unless UTF validity checking has been turned off
792 (see the discussion of PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK and
793 PCRE_NO_UTF32_CHECK in the
794 <a href="pcreapi.html"><b>pcreapi</b></a>
795 page). Perl does not support the Cs property.
796 </P>
797 <P>
798 The long synonyms for property names that Perl supports (such as \p{Letter})
799 are not supported by PCRE, nor is it permitted to prefix any of these
800 properties with "Is".
801 </P>
802 <P>
803 No character that is in the Unicode table has the Cn (unassigned) property.
804 Instead, this property is assumed for any code point that is not in the
805 Unicode table.
806 </P>
807 <P>
808 Specifying caseless matching does not affect these escape sequences. For
809 example, \p{Lu} always matches only upper case letters. This is different from
810 the behaviour of current versions of Perl.
811 </P>
812 <P>
813 Matching characters by Unicode property is not fast, because PCRE has to do a
814 multistage table lookup in order to find a character's property. That is why
815 the traditional escape sequences such as \d and \w do not use Unicode
816 properties in PCRE by default, though you can make them do so by setting the
817 PCRE_UCP option or by starting the pattern with (*UCP).
818 </P>
819 <br><b>
820 Extended grapheme clusters
821 </b><br>
822 <P>
823 The \X escape matches any number of Unicode characters that form an "extended
824 grapheme cluster", and treats the sequence as an atomic group
825 <a href="#atomicgroup">(see below).</a>
826 Up to and including release 8.31, PCRE matched an earlier, simpler definition
827 that was equivalent to
828 <pre>
829 (?&#62;\PM\pM*)
830 </pre>
831 That is, it matched a character without the "mark" property, followed by zero
832 or more characters with the "mark" property. Characters with the "mark"
833 property are typically non-spacing accents that affect the preceding character.
834 </P>
835 <P>
836 This simple definition was extended in Unicode to include more complicated
837 kinds of composite character by giving each character a grapheme breaking
838 property, and creating rules that use these properties to define the boundaries
839 of extended grapheme clusters. In releases of PCRE later than 8.31, \X matches
840 one of these clusters.
841 </P>
842 <P>
843 \X always matches at least one character. Then it decides whether to add
844 additional characters according to the following rules for ending a cluster:
845 </P>
846 <P>
847 1. End at the end of the subject string.
848 </P>
849 <P>
850 2. Do not end between CR and LF; otherwise end after any control character.
851 </P>
852 <P>
853 3. Do not break Hangul (a Korean script) syllable sequences. Hangul characters
854 are of five types: L, V, T, LV, and LVT. An L character may be followed by an
855 L, V, LV, or LVT character; an LV or V character may be followed by a V or T
856 character; an LVT or T character may be follwed only by a T character.
857 </P>
858 <P>
859 4. Do not end before extending characters or spacing marks. Characters with
860 the "mark" property always have the "extend" grapheme breaking property.
861 </P>
862 <P>
863 5. Do not end after prepend characters.
864 </P>
865 <P>
866 6. Otherwise, end the cluster.
867 <a name="extraprops"></a></P>
868 <br><b>
869 PCRE's additional properties
870 </b><br>
871 <P>
872 As well as the standard Unicode properties described above, PCRE supports four
873 more that make it possible to convert traditional escape sequences such as \w
874 and \s and POSIX character classes to use Unicode properties. PCRE uses these
875 non-standard, non-Perl properties internally when PCRE_UCP is set. However,
876 they may also be used explicitly. These properties are:
877 <pre>
878 Xan Any alphanumeric character
879 Xps Any POSIX space character
880 Xsp Any Perl space character
881 Xwd Any Perl "word" character
882 </pre>
883 Xan matches characters that have either the L (letter) or the N (number)
884 property. Xps matches the characters tab, linefeed, vertical tab, form feed, or
885 carriage return, and any other character that has the Z (separator) property.
886 Xsp is the same as Xps, except that vertical tab is excluded. Xwd matches the
887 same characters as Xan, plus underscore.
888 </P>
889 <P>
890 There is another non-standard property, Xuc, which matches any character that
891 can be represented by a Universal Character Name in C++ and other programming
892 languages. These are the characters $, @, ` (grave accent), and all characters
893 with Unicode code points greater than or equal to U+00A0, except for the
894 surrogates U+D800 to U+DFFF. Note that most base (ASCII) characters are
895 excluded. (Universal Character Names are of the form \uHHHH or \UHHHHHHHH
896 where H is a hexadecimal digit. Note that the Xuc property does not match these
897 sequences but the characters that they represent.)
898 <a name="resetmatchstart"></a></P>
899 <br><b>
900 Resetting the match start
901 </b><br>
902 <P>
903 The escape sequence \K causes any previously matched characters not to be
904 included in the final matched sequence. For example, the pattern:
905 <pre>
906 foo\Kbar
907 </pre>
908 matches "foobar", but reports that it has matched "bar". This feature is
909 similar to a lookbehind assertion
910 <a href="#lookbehind">(described below).</a>
911 However, in this case, the part of the subject before the real match does not
912 have to be of fixed length, as lookbehind assertions do. The use of \K does
913 not interfere with the setting of
914 <a href="#subpattern">captured substrings.</a>
915 For example, when the pattern
916 <pre>
917 (foo)\Kbar
918 </pre>
919 matches "foobar", the first substring is still set to "foo".
920 </P>
921 <P>
922 Perl documents that the use of \K within assertions is "not well defined". In
923 PCRE, \K is acted upon when it occurs inside positive assertions, but is
924 ignored in negative assertions.
925 <a name="smallassertions"></a></P>
926 <br><b>
927 Simple assertions
928 </b><br>
929 <P>
930 The final use of backslash is for certain simple assertions. An assertion
931 specifies a condition that has to be met at a particular point in a match,
932 without consuming any characters from the subject string. The use of
933 subpatterns for more complicated assertions is described
934 <a href="#bigassertions">below.</a>
935 The backslashed assertions are:
936 <pre>
937 \b matches at a word boundary
938 \B matches when not at a word boundary
939 \A matches at the start of the subject
940 \Z matches at the end of the subject
941 also matches before a newline at the end of the subject
942 \z matches only at the end of the subject
943 \G matches at the first matching position in the subject
944 </pre>
945 Inside a character class, \b has a different meaning; it matches the backspace
946 character. If any other of these assertions appears in a character class, by
947 default it matches the corresponding literal character (for example, \B
948 matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
949 escape sequence" error is generated instead.
950 </P>
951 <P>
952 A word boundary is a position in the subject string where the current character
953 and the previous character do not both match \w or \W (i.e. one matches
954 \w and the other matches \W), or the start or end of the string if the
955 first or last character matches \w, respectively. In a UTF mode, the meanings
956 of \w and \W can be changed by setting the PCRE_UCP option. When this is
957 done, it also affects \b and \B. Neither PCRE nor Perl has a separate "start
958 of word" or "end of word" metasequence. However, whatever follows \b normally
959 determines which it is. For example, the fragment \ba matches "a" at the start
960 of a word.
961 </P>
962 <P>
963 The \A, \Z, and \z assertions differ from the traditional circumflex and
964 dollar (described in the next section) in that they only ever match at the very
965 start and end of the subject string, whatever options are set. Thus, they are
966 independent of multiline mode. These three assertions are not affected by the
967 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
968 circumflex and dollar metacharacters. However, if the <i>startoffset</i>
969 argument of <b>pcre_exec()</b> is non-zero, indicating that matching is to start
970 at a point other than the beginning of the subject, \A can never match. The
971 difference between \Z and \z is that \Z matches before a newline at the end
972 of the string as well as at the very end, whereas \z matches only at the end.
973 </P>
974 <P>
975 The \G assertion is true only when the current matching position is at the
976 start point of the match, as specified by the <i>startoffset</i> argument of
977 <b>pcre_exec()</b>. It differs from \A when the value of <i>startoffset</i> is
978 non-zero. By calling <b>pcre_exec()</b> multiple times with appropriate
979 arguments, you can mimic Perl's /g option, and it is in this kind of
980 implementation where \G can be useful.
981 </P>
982 <P>
983 Note, however, that PCRE's interpretation of \G, as the start of the current
984 match, is subtly different from Perl's, which defines it as the end of the
985 previous match. In Perl, these can be different when the previously matched
986 string was empty. Because PCRE does just one match at a time, it cannot
987 reproduce this behaviour.
988 </P>
989 <P>
990 If all the alternatives of a pattern begin with \G, the expression is anchored
991 to the starting match position, and the "anchored" flag is set in the compiled
992 regular expression.
993 </P>
994 <br><a name="SEC6" href="#TOC1">CIRCUMFLEX AND DOLLAR</a><br>
995 <P>
996 The circumflex and dollar metacharacters are zero-width assertions. That is,
997 they test for a particular condition being true without consuming any
998 characters from the subject string.
999 </P>
1000 <P>
1001 Outside a character class, in the default matching mode, the circumflex
1002 character is an assertion that is true only if the current matching point is at
1003 the start of the subject string. If the <i>startoffset</i> argument of
1004 <b>pcre_exec()</b> is non-zero, circumflex can never match if the PCRE_MULTILINE
1005 option is unset. Inside a character class, circumflex has an entirely different
1006 meaning
1007 <a href="#characterclass">(see below).</a>
1008 </P>
1009 <P>
1010 Circumflex need not be the first character of the pattern if a number of
1011 alternatives are involved, but it should be the first thing in each alternative
1012 in which it appears if the pattern is ever to match that branch. If all
1013 possible alternatives start with a circumflex, that is, if the pattern is
1014 constrained to match only at the start of the subject, it is said to be an
1015 "anchored" pattern. (There are also other constructs that can cause a pattern
1016 to be anchored.)
1017 </P>
1018 <P>
1019 The dollar character is an assertion that is true only if the current matching
1020 point is at the end of the subject string, or immediately before a newline at
1021 the end of the string (by default). Note, however, that it does not actually
1022 match the newline. Dollar need not be the last character of the pattern if a
1023 number of alternatives are involved, but it should be the last item in any
1024 branch in which it appears. Dollar has no special meaning in a character class.
1025 </P>
1026 <P>
1027 The meaning of dollar can be changed so that it matches only at the very end of
1028 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
1029 does not affect the \Z assertion.
1030 </P>
1031 <P>
1032 The meanings of the circumflex and dollar characters are changed if the
1033 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
1034 immediately after internal newlines as well as at the start of the subject
1035 string. It does not match after a newline that ends the string. A dollar
1036 matches before any newlines in the string, as well as at the very end, when
1037 PCRE_MULTILINE is set. When newline is specified as the two-character
1038 sequence CRLF, isolated CR and LF characters do not indicate newlines.
1039 </P>
1040 <P>
1041 For example, the pattern /^abc$/ matches the subject string "def\nabc" (where
1042 \n represents a newline) in multiline mode, but not otherwise. Consequently,
1043 patterns that are anchored in single line mode because all branches start with
1044 ^ are not anchored in multiline mode, and a match for circumflex is possible
1045 when the <i>startoffset</i> argument of <b>pcre_exec()</b> is non-zero. The
1046 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
1047 </P>
1048 <P>
1049 Note that the sequences \A, \Z, and \z can be used to match the start and
1050 end of the subject in both modes, and if all branches of a pattern start with
1051 \A it is always anchored, whether or not PCRE_MULTILINE is set.
1052 <a name="fullstopdot"></a></P>
1053 <br><a name="SEC7" href="#TOC1">FULL STOP (PERIOD, DOT) AND \N</a><br>
1054 <P>
1055 Outside a character class, a dot in the pattern matches any one character in
1056 the subject string except (by default) a character that signifies the end of a
1057 line.
1058 </P>
1059 <P>
1060 When a line ending is defined as a single character, dot never matches that
1061 character; when the two-character sequence CRLF is used, dot does not match CR
1062 if it is immediately followed by LF, but otherwise it matches all characters
1063 (including isolated CRs and LFs). When any Unicode line endings are being
1064 recognized, dot does not match CR or LF or any of the other line ending
1065 characters.
1066 </P>
1067 <P>
1068 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
1069 option is set, a dot matches any one character, without exception. If the
1070 two-character sequence CRLF is present in the subject string, it takes two dots
1071 to match it.
1072 </P>
1073 <P>
1074 The handling of dot is entirely independent of the handling of circumflex and
1075 dollar, the only relationship being that they both involve newlines. Dot has no
1076 special meaning in a character class.
1077 </P>
1078 <P>
1079 The escape sequence \N behaves like a dot, except that it is not affected by
1080 the PCRE_DOTALL option. In other words, it matches any character except one
1081 that signifies the end of a line. Perl also uses \N to match characters by
1082 name; PCRE does not support this.
1083 </P>
1084 <br><a name="SEC8" href="#TOC1">MATCHING A SINGLE DATA UNIT</a><br>
1085 <P>
1086 Outside a character class, the escape sequence \C matches any one data unit,
1087 whether or not a UTF mode is set. In the 8-bit library, one data unit is one
1088 byte; in the 16-bit library it is a 16-bit unit; in the 32-bit library it is
1089 a 32-bit unit. Unlike a dot, \C always
1090 matches line-ending characters. The feature is provided in Perl in order to
1091 match individual bytes in UTF-8 mode, but it is unclear how it can usefully be
1092 used. Because \C breaks up characters into individual data units, matching one
1093 unit with \C in a UTF mode means that the rest of the string may start with a
1094 malformed UTF character. This has undefined results, because PCRE assumes that
1095 it is dealing with valid UTF strings (and by default it checks this at the
1096 start of processing unless the PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK or
1097 PCRE_NO_UTF32_CHECK option is used).
1098 </P>
1099 <P>
1100 PCRE does not allow \C to appear in lookbehind assertions
1101 <a href="#lookbehind">(described below)</a>
1102 in a UTF mode, because this would make it impossible to calculate the length of
1103 the lookbehind.
1104 </P>
1105 <P>
1106 In general, the \C escape sequence is best avoided. However, one
1107 way of using it that avoids the problem of malformed UTF characters is to use a
1108 lookahead to check the length of the next character, as in this pattern, which
1109 could be used with a UTF-8 string (ignore white space and line breaks):
1110 <pre>
1111 (?| (?=[\x00-\x7f])(\C) |
1112 (?=[\x80-\x{7ff}])(\C)(\C) |
1113 (?=[\x{800}-\x{ffff}])(\C)(\C)(\C) |
1114 (?=[\x{10000}-\x{1fffff}])(\C)(\C)(\C)(\C))
1115 </pre>
1116 A group that starts with (?| resets the capturing parentheses numbers in each
1117 alternative (see
1118 <a href="#dupsubpatternnumber">"Duplicate Subpattern Numbers"</a>
1119 below). The assertions at the start of each branch check the next UTF-8
1120 character for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The
1121 character's individual bytes are then captured by the appropriate number of
1122 groups.
1123 <a name="characterclass"></a></P>
1124 <br><a name="SEC9" href="#TOC1">SQUARE BRACKETS AND CHARACTER CLASSES</a><br>
1125 <P>
1126 An opening square bracket introduces a character class, terminated by a closing
1127 square bracket. A closing square bracket on its own is not special by default.
1128 However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
1129 bracket causes a compile-time error. If a closing square bracket is required as
1130 a member of the class, it should be the first data character in the class
1131 (after an initial circumflex, if present) or escaped with a backslash.
1132 </P>
1133 <P>
1134 A character class matches a single character in the subject. In a UTF mode, the
1135 character may be more than one data unit long. A matched character must be in
1136 the set of characters defined by the class, unless the first character in the
1137 class definition is a circumflex, in which case the subject character must not
1138 be in the set defined by the class. If a circumflex is actually required as a
1139 member of the class, ensure it is not the first character, or escape it with a
1140 backslash.
1141 </P>
1142 <P>
1143 For example, the character class [aeiou] matches any lower case vowel, while
1144 [^aeiou] matches any character that is not a lower case vowel. Note that a
1145 circumflex is just a convenient notation for specifying the characters that
1146 are in the class by enumerating those that are not. A class that starts with a
1147 circumflex is not an assertion; it still consumes a character from the subject
1148 string, and therefore it fails if the current pointer is at the end of the
1149 string.
1150 </P>
1151 <P>
1152 In UTF-8 (UTF-16, UTF-32) mode, characters with values greater than 255 (0xffff)
1153 can be included in a class as a literal string of data units, or by using the
1154 \x{ escaping mechanism.
1155 </P>
1156 <P>
1157 When caseless matching is set, any letters in a class represent both their
1158 upper case and lower case versions, so for example, a caseless [aeiou] matches
1159 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1160 caseful version would. In a UTF mode, PCRE always understands the concept of
1161 case for characters whose values are less than 128, so caseless matching is
1162 always possible. For characters with higher values, the concept of case is
1163 supported if PCRE is compiled with Unicode property support, but not otherwise.
1164 If you want to use caseless matching in a UTF mode for characters 128 and
1165 above, you must ensure that PCRE is compiled with Unicode property support as
1166 well as with UTF support.
1167 </P>
1168 <P>
1169 Characters that might indicate line breaks are never treated in any special way
1170 when matching character classes, whatever line-ending sequence is in use, and
1171 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
1172 such as [^a] always matches one of these characters.
1173 </P>
1174 <P>
1175 The minus (hyphen) character can be used to specify a range of characters in a
1176 character class. For example, [d-m] matches any letter between d and m,
1177 inclusive. If a minus character is required in a class, it must be escaped with
1178 a backslash or appear in a position where it cannot be interpreted as
1179 indicating a range, typically as the first or last character in the class.
1180 </P>
1181 <P>
1182 It is not possible to have the literal character "]" as the end character of a
1183 range. A pattern such as [W-]46] is interpreted as a class of two characters
1184 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1185 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1186 the end of range, so [W-\]46] is interpreted as a class containing a range
1187 followed by two other characters. The octal or hexadecimal representation of
1188 "]" can also be used to end a range.
1189 </P>
1190 <P>
1191 Ranges operate in the collating sequence of character values. They can also be
1192 used for characters specified numerically, for example [\000-\037]. Ranges
1193 can include any characters that are valid for the current mode.
1194 </P>
1195 <P>
1196 If a range that includes letters is used when caseless matching is set, it
1197 matches the letters in either case. For example, [W-c] is equivalent to
1198 [][\\^_`wxyzabc], matched caselessly, and in a non-UTF mode, if character
1199 tables for a French locale are in use, [\xc8-\xcb] matches accented E
1200 characters in both cases. In UTF modes, PCRE supports the concept of case for
1201 characters with values greater than 128 only when it is compiled with Unicode
1202 property support.
1203 </P>
1204 <P>
1205 The character escape sequences \d, \D, \h, \H, \p, \P, \s, \S, \v,
1206 \V, \w, and \W may appear in a character class, and add the characters that
1207 they match to the class. For example, [\dABCDEF] matches any hexadecimal
1208 digit. In UTF modes, the PCRE_UCP option affects the meanings of \d, \s, \w
1209 and their upper case partners, just as it does when they appear outside a
1210 character class, as described in the section entitled
1211 <a href="#genericchartypes">"Generic character types"</a>
1212 above. The escape sequence \b has a different meaning inside a character
1213 class; it matches the backspace character. The sequences \B, \N, \R, and \X
1214 are not special inside a character class. Like any other unrecognized escape
1215 sequences, they are treated as the literal characters "B", "N", "R", and "X" by
1216 default, but cause an error if the PCRE_EXTRA option is set.
1217 </P>
1218 <P>
1219 A circumflex can conveniently be used with the upper case character types to
1220 specify a more restricted set of characters than the matching lower case type.
1221 For example, the class [^\W_] matches any letter or digit, but not underscore,
1222 whereas [\w] includes underscore. A positive character class should be read as
1223 "something OR something OR ..." and a negative class as "NOT something AND NOT
1224 something AND NOT ...".
1225 </P>
1226 <P>
1227 The only metacharacters that are recognized in character classes are backslash,
1228 hyphen (only where it can be interpreted as specifying a range), circumflex
1229 (only at the start), opening square bracket (only when it can be interpreted as
1230 introducing a POSIX class name - see the next section), and the terminating
1231 closing square bracket. However, escaping other non-alphanumeric characters
1232 does no harm.
1233 </P>
1234 <br><a name="SEC10" href="#TOC1">POSIX CHARACTER CLASSES</a><br>
1235 <P>
1236 Perl supports the POSIX notation for character classes. This uses names
1237 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1238 this notation. For example,
1239 <pre>
1240 [01[:alpha:]%]
1241 </pre>
1242 matches "0", "1", any alphabetic character, or "%". The supported class names
1243 are:
1244 <pre>
1245 alnum letters and digits
1246 alpha letters
1247 ascii character codes 0 - 127
1248 blank space or tab only
1249 cntrl control characters
1250 digit decimal digits (same as \d)
1251 graph printing characters, excluding space
1252 lower lower case letters
1253 print printing characters, including space
1254 punct printing characters, excluding letters and digits and space
1255 space white space (not quite the same as \s)
1256 upper upper case letters
1257 word "word" characters (same as \w)
1258 xdigit hexadecimal digits
1259 </pre>
1260 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1261 space (32). Notice that this list includes the VT character (code 11). This
1262 makes "space" different to \s, which does not include VT (for Perl
1263 compatibility).
1264 </P>
1265 <P>
1266 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1267 5.8. Another Perl extension is negation, which is indicated by a ^ character
1268 after the colon. For example,
1269 <pre>
1270 [12[:^digit:]]
1271 </pre>
1272 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1273 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1274 supported, and an error is given if they are encountered.
1275 </P>
1276 <P>
1277 By default, in UTF modes, characters with values greater than 128 do not match
1278 any of the POSIX character classes. However, if the PCRE_UCP option is passed
1279 to <b>pcre_compile()</b>, some of the classes are changed so that Unicode
1280 character properties are used. This is achieved by replacing the POSIX classes
1281 by other sequences, as follows:
1282 <pre>
1283 [:alnum:] becomes \p{Xan}
1284 [:alpha:] becomes \p{L}
1285 [:blank:] becomes \h
1286 [:digit:] becomes \p{Nd}
1287 [:lower:] becomes \p{Ll}
1288 [:space:] becomes \p{Xps}
1289 [:upper:] becomes \p{Lu}
1290 [:word:] becomes \p{Xwd}
1291 </pre>
1292 Negated versions, such as [:^alpha:] use \P instead of \p. The other POSIX
1293 classes are unchanged, and match only characters with code points less than
1294 128.
1295 </P>
1296 <br><a name="SEC11" href="#TOC1">VERTICAL BAR</a><br>
1297 <P>
1298 Vertical bar characters are used to separate alternative patterns. For example,
1299 the pattern
1300 <pre>
1301 gilbert|sullivan
1302 </pre>
1303 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1304 and an empty alternative is permitted (matching the empty string). The matching
1305 process tries each alternative in turn, from left to right, and the first one
1306 that succeeds is used. If the alternatives are within a subpattern
1307 <a href="#subpattern">(defined below),</a>
1308 "succeeds" means matching the rest of the main pattern as well as the
1309 alternative in the subpattern.
1310 </P>
1311 <br><a name="SEC12" href="#TOC1">INTERNAL OPTION SETTING</a><br>
1312 <P>
1313 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1314 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1315 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1316 The option letters are
1317 <pre>
1318 i for PCRE_CASELESS
1320 s for PCRE_DOTALL
1321 x for PCRE_EXTENDED
1322 </pre>
1323 For example, (?im) sets caseless, multiline matching. It is also possible to
1324 unset these options by preceding the letter with a hyphen, and a combined
1325 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1326 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1327 permitted. If a letter appears both before and after the hyphen, the option is
1328 unset.
1329 </P>
1330 <P>
1331 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1332 changed in the same way as the Perl-compatible options by using the characters
1333 J, U and X respectively.
1334 </P>
1335 <P>
1336 When one of these option changes occurs at top level (that is, not inside
1337 subpattern parentheses), the change applies to the remainder of the pattern
1338 that follows. If the change is placed right at the start of a pattern, PCRE
1339 extracts it into the global options (and it will therefore show up in data
1340 extracted by the <b>pcre_fullinfo()</b> function).
1341 </P>
1342 <P>
1343 An option change within a subpattern (see below for a description of
1344 subpatterns) affects only that part of the subpattern that follows it, so
1345 <pre>
1346 (a(?i)b)c
1347 </pre>
1348 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1349 By this means, options can be made to have different settings in different
1350 parts of the pattern. Any changes made in one alternative do carry on
1351 into subsequent branches within the same subpattern. For example,
1352 <pre>
1353 (a(?i)b|c)
1354 </pre>
1355 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1356 branch is abandoned before the option setting. This is because the effects of
1357 option settings happen at compile time. There would be some very weird
1358 behaviour otherwise.
1359 </P>
1360 <P>
1361 <b>Note:</b> There are other PCRE-specific options that can be set by the
1362 application when the compiling or matching functions are called. In some cases
1363 the pattern can contain special leading sequences such as (*CRLF) to override
1364 what the application has set or what has been defaulted. Details are given in
1365 the section entitled
1366 <a href="#newlineseq">"Newline sequences"</a>
1367 above. There are also the (*UTF8), (*UTF16),(*UTF32), and (*UCP) leading
1368 sequences that can be used to set UTF and Unicode property modes; they are
1369 equivalent to setting the PCRE_UTF8, PCRE_UTF16, PCRE_UTF32 and the PCRE_UCP
1370 options, respectively. The (*UTF) sequence is a generic version that can be
1371 used with any of the libraries.
1372 <a name="subpattern"></a></P>
1373 <br><a name="SEC13" href="#TOC1">SUBPATTERNS</a><br>
1374 <P>
1375 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1376 Turning part of a pattern into a subpattern does two things:
1377 <br>
1378 <br>
1379 1. It localizes a set of alternatives. For example, the pattern
1380 <pre>
1381 cat(aract|erpillar|)
1382 </pre>
1383 matches "cataract", "caterpillar", or "cat". Without the parentheses, it would
1384 match "cataract", "erpillar" or an empty string.
1385 <br>
1386 <br>
1387 2. It sets up the subpattern as a capturing subpattern. This means that, when
1388 the whole pattern matches, that portion of the subject string that matched the
1389 subpattern is passed back to the caller via the <i>ovector</i> argument of the
1390 matching function. (This applies only to the traditional matching functions;
1391 the DFA matching functions do not support capturing.)
1392 </P>
1393 <P>
1394 Opening parentheses are counted from left to right (starting from 1) to obtain
1395 numbers for the capturing subpatterns. For example, if the string "the red
1396 king" is matched against the pattern
1397 <pre>
1398 the ((red|white) (king|queen))
1399 </pre>
1400 the captured substrings are "red king", "red", and "king", and are numbered 1,
1401 2, and 3, respectively.
1402 </P>
1403 <P>
1404 The fact that plain parentheses fulfil two functions is not always helpful.
1405 There are often times when a grouping subpattern is required without a
1406 capturing requirement. If an opening parenthesis is followed by a question mark
1407 and a colon, the subpattern does not do any capturing, and is not counted when
1408 computing the number of any subsequent capturing subpatterns. For example, if
1409 the string "the white queen" is matched against the pattern
1410 <pre>
1411 the ((?:red|white) (king|queen))
1412 </pre>
1413 the captured substrings are "white queen" and "queen", and are numbered 1 and
1414 2. The maximum number of capturing subpatterns is 65535.
1415 </P>
1416 <P>
1417 As a convenient shorthand, if any option settings are required at the start of
1418 a non-capturing subpattern, the option letters may appear between the "?" and
1419 the ":". Thus the two patterns
1420 <pre>
1421 (?i:saturday|sunday)
1422 (?:(?i)saturday|sunday)
1423 </pre>
1424 match exactly the same set of strings. Because alternative branches are tried
1425 from left to right, and options are not reset until the end of the subpattern
1426 is reached, an option setting in one branch does affect subsequent branches, so
1427 the above patterns match "SUNDAY" as well as "Saturday".
1428 <a name="dupsubpatternnumber"></a></P>
1429 <br><a name="SEC14" href="#TOC1">DUPLICATE SUBPATTERN NUMBERS</a><br>
1430 <P>
1431 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1432 the same numbers for its capturing parentheses. Such a subpattern starts with
1433 (?| and is itself a non-capturing subpattern. For example, consider this
1434 pattern:
1435 <pre>
1436 (?|(Sat)ur|(Sun))day
1437 </pre>
1438 Because the two alternatives are inside a (?| group, both sets of capturing
1439 parentheses are numbered one. Thus, when the pattern matches, you can look
1440 at captured substring number one, whichever alternative matched. This construct
1441 is useful when you want to capture part, but not all, of one of a number of
1442 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1443 number is reset at the start of each branch. The numbers of any capturing
1444 parentheses that follow the subpattern start after the highest number used in
1445 any branch. The following example is taken from the Perl documentation. The
1446 numbers underneath show in which buffer the captured content will be stored.
1447 <pre>
1448 # before ---------------branch-reset----------- after
1449 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1450 # 1 2 2 3 2 3 4
1451 </pre>
1452 A back reference to a numbered subpattern uses the most recent value that is
1453 set for that number by any subpattern. The following pattern matches "abcabc"
1454 or "defdef":
1455 <pre>
1456 /(?|(abc)|(def))\1/
1457 </pre>
1458 In contrast, a subroutine call to a numbered subpattern always refers to the
1459 first one in the pattern with the given number. The following pattern matches
1460 "abcabc" or "defabc":
1461 <pre>
1462 /(?|(abc)|(def))(?1)/
1463 </pre>
1464 If a
1465 <a href="#conditions">condition test</a>
1466 for a subpattern's having matched refers to a non-unique number, the test is
1467 true if any of the subpatterns of that number have matched.
1468 </P>
1469 <P>
1470 An alternative approach to using this "branch reset" feature is to use
1471 duplicate named subpatterns, as described in the next section.
1472 </P>
1473 <br><a name="SEC15" href="#TOC1">NAMED SUBPATTERNS</a><br>
1474 <P>
1475 Identifying capturing parentheses by number is simple, but it can be very hard
1476 to keep track of the numbers in complicated regular expressions. Furthermore,
1477 if an expression is modified, the numbers may change. To help with this
1478 difficulty, PCRE supports the naming of subpatterns. This feature was not
1479 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1480 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1481 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1482 have different names, but PCRE does not.
1483 </P>
1484 <P>
1485 In PCRE, a subpattern can be named in one of three ways: (?&#60;name&#62;...) or
1486 (?'name'...) as in Perl, or (?P&#60;name&#62;...) as in Python. References to capturing
1487 parentheses from other parts of the pattern, such as
1488 <a href="#backreferences">back references,</a>
1489 <a href="#recursion">recursion,</a>
1490 and
1491 <a href="#conditions">conditions,</a>
1492 can be made by name as well as by number.
1493 </P>
1494 <P>
1495 Names consist of up to 32 alphanumeric characters and underscores. Named
1496 capturing parentheses are still allocated numbers as well as names, exactly as
1497 if the names were not present. The PCRE API provides function calls for
1498 extracting the name-to-number translation table from a compiled pattern. There
1499 is also a convenience function for extracting a captured substring by name.
1500 </P>
1501 <P>
1502 By default, a name must be unique within a pattern, but it is possible to relax
1503 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1504 names are also always permitted for subpatterns with the same number, set up as
1505 described in the previous section.) Duplicate names can be useful for patterns
1506 where only one instance of the named parentheses can match. Suppose you want to
1507 match the name of a weekday, either as a 3-letter abbreviation or as the full
1508 name, and in both cases you want to extract the abbreviation. This pattern
1509 (ignoring the line breaks) does the job:
1510 <pre>
1511 (?&#60;DN&#62;Mon|Fri|Sun)(?:day)?|
1512 (?&#60;DN&#62;Tue)(?:sday)?|
1513 (?&#60;DN&#62;Wed)(?:nesday)?|
1514 (?&#60;DN&#62;Thu)(?:rsday)?|
1515 (?&#60;DN&#62;Sat)(?:urday)?
1516 </pre>
1517 There are five capturing substrings, but only one is ever set after a match.
1518 (An alternative way of solving this problem is to use a "branch reset"
1519 subpattern, as described in the previous section.)
1520 </P>
1521 <P>
1522 The convenience function for extracting the data by name returns the substring
1523 for the first (and in this example, the only) subpattern of that name that
1524 matched. This saves searching to find which numbered subpattern it was.
1525 </P>
1526 <P>
1527 If you make a back reference to a non-unique named subpattern from elsewhere in
1528 the pattern, the one that corresponds to the first occurrence of the name is
1529 used. In the absence of duplicate numbers (see the previous section) this is
1530 the one with the lowest number. If you use a named reference in a condition
1531 test (see the
1532 <a href="#conditions">section about conditions</a>
1533 below), either to check whether a subpattern has matched, or to check for
1534 recursion, all subpatterns with the same name are tested. If the condition is
1535 true for any one of them, the overall condition is true. This is the same
1536 behaviour as testing by number. For further details of the interfaces for
1537 handling named subpatterns, see the
1538 <a href="pcreapi.html"><b>pcreapi</b></a>
1539 documentation.
1540 </P>
1541 <P>
1542 <b>Warning:</b> You cannot use different names to distinguish between two
1543 subpatterns with the same number because PCRE uses only the numbers when
1544 matching. For this reason, an error is given at compile time if different names
1545 are given to subpatterns with the same number. However, you can give the same
1546 name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
1547 </P>
1548 <br><a name="SEC16" href="#TOC1">REPETITION</a><br>
1549 <P>
1550 Repetition is specified by quantifiers, which can follow any of the following
1551 items:
1552 <pre>
1553 a literal data character
1554 the dot metacharacter
1555 the \C escape sequence
1556 the \X escape sequence
1557 the \R escape sequence
1558 an escape such as \d or \pL that matches a single character
1559 a character class
1560 a back reference (see next section)
1561 a parenthesized subpattern (including assertions)
1562 a subroutine call to a subpattern (recursive or otherwise)
1563 </pre>
1564 The general repetition quantifier specifies a minimum and maximum number of
1565 permitted matches, by giving the two numbers in curly brackets (braces),
1566 separated by a comma. The numbers must be less than 65536, and the first must
1567 be less than or equal to the second. For example:
1568 <pre>
1569 z{2,4}
1570 </pre>
1571 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1572 character. If the second number is omitted, but the comma is present, there is
1573 no upper limit; if the second number and the comma are both omitted, the
1574 quantifier specifies an exact number of required matches. Thus
1575 <pre>
1576 [aeiou]{3,}
1577 </pre>
1578 matches at least 3 successive vowels, but may match many more, while
1579 <pre>
1580 \d{8}
1581 </pre>
1582 matches exactly 8 digits. An opening curly bracket that appears in a position
1583 where a quantifier is not allowed, or one that does not match the syntax of a
1584 quantifier, is taken as a literal character. For example, {,6} is not a
1585 quantifier, but a literal string of four characters.
1586 </P>
1587 <P>
1588 In UTF modes, quantifiers apply to characters rather than to individual data
1589 units. Thus, for example, \x{100}{2} matches two characters, each of
1590 which is represented by a two-byte sequence in a UTF-8 string. Similarly,
1591 \X{3} matches three Unicode extended grapheme clusters, each of which may be
1592 several data units long (and they may be of different lengths).
1593 </P>
1594 <P>
1595 The quantifier {0} is permitted, causing the expression to behave as if the
1596 previous item and the quantifier were not present. This may be useful for
1597 subpatterns that are referenced as
1598 <a href="#subpatternsassubroutines">subroutines</a>
1599 from elsewhere in the pattern (but see also the section entitled
1600 <a href="#subdefine">"Defining subpatterns for use by reference only"</a>
1601 below). Items other than subpatterns that have a {0} quantifier are omitted
1602 from the compiled pattern.
1603 </P>
1604 <P>
1605 For convenience, the three most common quantifiers have single-character
1606 abbreviations:
1607 <pre>
1608 * is equivalent to {0,}
1609 + is equivalent to {1,}
1610 ? is equivalent to {0,1}
1611 </pre>
1612 It is possible to construct infinite loops by following a subpattern that can
1613 match no characters with a quantifier that has no upper limit, for example:
1614 <pre>
1615 (a?)*
1616 </pre>
1617 Earlier versions of Perl and PCRE used to give an error at compile time for
1618 such patterns. However, because there are cases where this can be useful, such
1619 patterns are now accepted, but if any repetition of the subpattern does in fact
1620 match no characters, the loop is forcibly broken.
1621 </P>
1622 <P>
1623 By default, the quantifiers are "greedy", that is, they match as much as
1624 possible (up to the maximum number of permitted times), without causing the
1625 rest of the pattern to fail. The classic example of where this gives problems
1626 is in trying to match comments in C programs. These appear between /* and */
1627 and within the comment, individual * and / characters may appear. An attempt to
1628 match C comments by applying the pattern
1629 <pre>
1630 /\*.*\*/
1631 </pre>
1632 to the string
1633 <pre>
1634 /* first comment */ not comment /* second comment */
1635 </pre>
1636 fails, because it matches the entire string owing to the greediness of the .*
1637 item.
1638 </P>
1639 <P>
1640 However, if a quantifier is followed by a question mark, it ceases to be
1641 greedy, and instead matches the minimum number of times possible, so the
1642 pattern
1643 <pre>
1644 /\*.*?\*/
1645 </pre>
1646 does the right thing with the C comments. The meaning of the various
1647 quantifiers is not otherwise changed, just the preferred number of matches.
1648 Do not confuse this use of question mark with its use as a quantifier in its
1649 own right. Because it has two uses, it can sometimes appear doubled, as in
1650 <pre>
1651 \d??\d
1652 </pre>
1653 which matches one digit by preference, but can match two if that is the only
1654 way the rest of the pattern matches.
1655 </P>
1656 <P>
1657 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1658 the quantifiers are not greedy by default, but individual ones can be made
1659 greedy by following them with a question mark. In other words, it inverts the
1660 default behaviour.
1661 </P>
1662 <P>
1663 When a parenthesized subpattern is quantified with a minimum repeat count that
1664 is greater than 1 or with a limited maximum, more memory is required for the
1665 compiled pattern, in proportion to the size of the minimum or maximum.
1666 </P>
1667 <P>
1668 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1669 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1670 implicitly anchored, because whatever follows will be tried against every
1671 character position in the subject string, so there is no point in retrying the
1672 overall match at any position after the first. PCRE normally treats such a
1673 pattern as though it were preceded by \A.
1674 </P>
1675 <P>
1676 In cases where it is known that the subject string contains no newlines, it is
1677 worth setting PCRE_DOTALL in order to obtain this optimization, or
1678 alternatively using ^ to indicate anchoring explicitly.
1679 </P>
1680 <P>
1681 However, there are some cases where the optimization cannot be used. When .*
1682 is inside capturing parentheses that are the subject of a back reference
1683 elsewhere in the pattern, a match at the start may fail where a later one
1684 succeeds. Consider, for example:
1685 <pre>
1686 (.*)abc\1
1687 </pre>
1688 If the subject is "xyz123abc123" the match point is the fourth character. For
1689 this reason, such a pattern is not implicitly anchored.
1690 </P>
1691 <P>
1692 Another case where implicit anchoring is not applied is when the leading .* is
1693 inside an atomic group. Once again, a match at the start may fail where a later
1694 one succeeds. Consider this pattern:
1695 <pre>
1696 (?&#62;.*?a)b
1697 </pre>
1698 It matches "ab" in the subject "aab". The use of the backtracking control verbs
1699 (*PRUNE) and (*SKIP) also disable this optimization.
1700 </P>
1701 <P>
1702 When a capturing subpattern is repeated, the value captured is the substring
1703 that matched the final iteration. For example, after
1704 <pre>
1705 (tweedle[dume]{3}\s*)+
1706 </pre>
1707 has matched "tweedledum tweedledee" the value of the captured substring is
1708 "tweedledee". However, if there are nested capturing subpatterns, the
1709 corresponding captured values may have been set in previous iterations. For
1710 example, after
1711 <pre>
1712 /(a|(b))+/
1713 </pre>
1714 matches "aba" the value of the second captured substring is "b".
1715 <a name="atomicgroup"></a></P>
1716 <br><a name="SEC17" href="#TOC1">ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS</a><br>
1717 <P>
1718 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1719 repetition, failure of what follows normally causes the repeated item to be
1720 re-evaluated to see if a different number of repeats allows the rest of the
1721 pattern to match. Sometimes it is useful to prevent this, either to change the
1722 nature of the match, or to cause it fail earlier than it otherwise might, when
1723 the author of the pattern knows there is no point in carrying on.
1724 </P>
1725 <P>
1726 Consider, for example, the pattern \d+foo when applied to the subject line
1727 <pre>
1728 123456bar
1729 </pre>
1730 After matching all 6 digits and then failing to match "foo", the normal
1731 action of the matcher is to try again with only 5 digits matching the \d+
1732 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1733 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1734 that once a subpattern has matched, it is not to be re-evaluated in this way.
1735 </P>
1736 <P>
1737 If we use atomic grouping for the previous example, the matcher gives up
1738 immediately on failing to match "foo" the first time. The notation is a kind of
1739 special parenthesis, starting with (?&#62; as in this example:
1740 <pre>
1741 (?&#62;\d+)foo
1742 </pre>
1743 This kind of parenthesis "locks up" the part of the pattern it contains once
1744 it has matched, and a failure further into the pattern is prevented from
1745 backtracking into it. Backtracking past it to previous items, however, works as
1746 normal.
1747 </P>
1748 <P>
1749 An alternative description is that a subpattern of this type matches the string
1750 of characters that an identical standalone pattern would match, if anchored at
1751 the current point in the subject string.
1752 </P>
1753 <P>
1754 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1755 the above example can be thought of as a maximizing repeat that must swallow
1756 everything it can. So, while both \d+ and \d+? are prepared to adjust the
1757 number of digits they match in order to make the rest of the pattern match,
1758 (?&#62;\d+) can only match an entire sequence of digits.
1759 </P>
1760 <P>
1761 Atomic groups in general can of course contain arbitrarily complicated
1762 subpatterns, and can be nested. However, when the subpattern for an atomic
1763 group is just a single repeated item, as in the example above, a simpler
1764 notation, called a "possessive quantifier" can be used. This consists of an
1765 additional + character following a quantifier. Using this notation, the
1766 previous example can be rewritten as
1767 <pre>
1768 \d++foo
1769 </pre>
1770 Note that a possessive quantifier can be used with an entire group, for
1771 example:
1772 <pre>
1773 (abc|xyz){2,3}+
1774 </pre>
1775 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1776 option is ignored. They are a convenient notation for the simpler forms of
1777 atomic group. However, there is no difference in the meaning of a possessive
1778 quantifier and the equivalent atomic group, though there may be a performance
1779 difference; possessive quantifiers should be slightly faster.
1780 </P>
1781 <P>
1782 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1783 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1784 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1785 package, and PCRE copied it from there. It ultimately found its way into Perl
1786 at release 5.10.
1787 </P>
1788 <P>
1789 PCRE has an optimization that automatically "possessifies" certain simple
1790 pattern constructs. For example, the sequence A+B is treated as A++B because
1791 there is no point in backtracking into a sequence of A's when B must follow.
1792 </P>
1793 <P>
1794 When a pattern contains an unlimited repeat inside a subpattern that can itself
1795 be repeated an unlimited number of times, the use of an atomic group is the
1796 only way to avoid some failing matches taking a very long time indeed. The
1797 pattern
1798 <pre>
1799 (\D+|&#60;\d+&#62;)*[!?]
1800 </pre>
1801 matches an unlimited number of substrings that either consist of non-digits, or
1802 digits enclosed in &#60;&#62;, followed by either ! or ?. When it matches, it runs
1803 quickly. However, if it is applied to
1804 <pre>
1805 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1806 </pre>
1807 it takes a long time before reporting failure. This is because the string can
1808 be divided between the internal \D+ repeat and the external * repeat in a
1809 large number of ways, and all have to be tried. (The example uses [!?] rather
1810 than a single character at the end, because both PCRE and Perl have an
1811 optimization that allows for fast failure when a single character is used. They
1812 remember the last single character that is required for a match, and fail early
1813 if it is not present in the string.) If the pattern is changed so that it uses
1814 an atomic group, like this:
1815 <pre>
1816 ((?&#62;\D+)|&#60;\d+&#62;)*[!?]
1817 </pre>
1818 sequences of non-digits cannot be broken, and failure happens quickly.
1819 <a name="backreferences"></a></P>
1820 <br><a name="SEC18" href="#TOC1">BACK REFERENCES</a><br>
1821 <P>
1822 Outside a character class, a backslash followed by a digit greater than 0 (and
1823 possibly further digits) is a back reference to a capturing subpattern earlier
1824 (that is, to its left) in the pattern, provided there have been that many
1825 previous capturing left parentheses.
1826 </P>
1827 <P>
1828 However, if the decimal number following the backslash is less than 10, it is
1829 always taken as a back reference, and causes an error only if there are not
1830 that many capturing left parentheses in the entire pattern. In other words, the
1831 parentheses that are referenced need not be to the left of the reference for
1832 numbers less than 10. A "forward back reference" of this type can make sense
1833 when a repetition is involved and the subpattern to the right has participated
1834 in an earlier iteration.
1835 </P>
1836 <P>
1837 It is not possible to have a numerical "forward back reference" to a subpattern
1838 whose number is 10 or more using this syntax because a sequence such as \50 is
1839 interpreted as a character defined in octal. See the subsection entitled
1840 "Non-printing characters"
1841 <a href="#digitsafterbackslash">above</a>
1842 for further details of the handling of digits following a backslash. There is
1843 no such problem when named parentheses are used. A back reference to any
1844 subpattern is possible using named parentheses (see below).
1845 </P>
1846 <P>
1847 Another way of avoiding the ambiguity inherent in the use of digits following a
1848 backslash is to use the \g escape sequence. This escape must be followed by an
1849 unsigned number or a negative number, optionally enclosed in braces. These
1850 examples are all identical:
1851 <pre>
1852 (ring), \1
1853 (ring), \g1
1854 (ring), \g{1}
1855 </pre>
1856 An unsigned number specifies an absolute reference without the ambiguity that
1857 is present in the older syntax. It is also useful when literal digits follow
1858 the reference. A negative number is a relative reference. Consider this
1859 example:
1860 <pre>
1861 (abc(def)ghi)\g{-1}
1862 </pre>
1863 The sequence \g{-1} is a reference to the most recently started capturing
1864 subpattern before \g, that is, is it equivalent to \2 in this example.
1865 Similarly, \g{-2} would be equivalent to \1. The use of relative references
1866 can be helpful in long patterns, and also in patterns that are created by
1867 joining together fragments that contain references within themselves.
1868 </P>
1869 <P>
1870 A back reference matches whatever actually matched the capturing subpattern in
1871 the current subject string, rather than anything matching the subpattern
1872 itself (see
1873 <a href="#subpatternsassubroutines">"Subpatterns as subroutines"</a>
1874 below for a way of doing that). So the pattern
1875 <pre>
1876 (sens|respons)e and \1ibility
1877 </pre>
1878 matches "sense and sensibility" and "response and responsibility", but not
1879 "sense and responsibility". If caseful matching is in force at the time of the
1880 back reference, the case of letters is relevant. For example,
1881 <pre>
1882 ((?i)rah)\s+\1
1883 </pre>
1884 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1885 capturing subpattern is matched caselessly.
1886 </P>
1887 <P>
1888 There are several different ways of writing back references to named
1889 subpatterns. The .NET syntax \k{name} and the Perl syntax \k&#60;name&#62; or
1890 \k'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1891 back reference syntax, in which \g can be used for both numeric and named
1892 references, is also supported. We could rewrite the above example in any of
1893 the following ways:
1894 <pre>
1895 (?&#60;p1&#62;(?i)rah)\s+\k&#60;p1&#62;
1896 (?'p1'(?i)rah)\s+\k{p1}
1897 (?P&#60;p1&#62;(?i)rah)\s+(?P=p1)
1898 (?&#60;p1&#62;(?i)rah)\s+\g{p1}
1899 </pre>
1900 A subpattern that is referenced by name may appear in the pattern before or
1901 after the reference.
1902 </P>
1903 <P>
1904 There may be more than one back reference to the same subpattern. If a
1905 subpattern has not actually been used in a particular match, any back
1906 references to it always fail by default. For example, the pattern
1907 <pre>
1908 (a|(bc))\2
1909 </pre>
1910 always fails if it starts to match "a" rather than "bc". However, if the
1911 PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1912 unset value matches an empty string.
1913 </P>
1914 <P>
1915 Because there may be many capturing parentheses in a pattern, all digits
1916 following a backslash are taken as part of a potential back reference number.
1917 If the pattern continues with a digit character, some delimiter must be used to
1918 terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1919 white space. Otherwise, the \g{ syntax or an empty comment (see
1920 <a href="#comments">"Comments"</a>
1921 below) can be used.
1922 </P>
1923 <br><b>
1924 Recursive back references
1925 </b><br>
1926 <P>
1927 A back reference that occurs inside the parentheses to which it refers fails
1928 when the subpattern is first used, so, for example, (a\1) never matches.
1929 However, such references can be useful inside repeated subpatterns. For
1930 example, the pattern
1931 <pre>
1932 (a|b\1)+
1933 </pre>
1934 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1935 the subpattern, the back reference matches the character string corresponding
1936 to the previous iteration. In order for this to work, the pattern must be such
1937 that the first iteration does not need to match the back reference. This can be
1938 done using alternation, as in the example above, or by a quantifier with a
1939 minimum of zero.
1940 </P>
1941 <P>
1942 Back references of this type cause the group that they reference to be treated
1943 as an
1944 <a href="#atomicgroup">atomic group.</a>
1945 Once the whole group has been matched, a subsequent matching failure cannot
1946 cause backtracking into the middle of the group.
1947 <a name="bigassertions"></a></P>
1948 <br><a name="SEC19" href="#TOC1">ASSERTIONS</a><br>
1949 <P>
1950 An assertion is a test on the characters following or preceding the current
1951 matching point that does not actually consume any characters. The simple
1952 assertions coded as \b, \B, \A, \G, \Z, \z, ^ and $ are described
1953 <a href="#smallassertions">above.</a>
1954 </P>
1955 <P>
1956 More complicated assertions are coded as subpatterns. There are two kinds:
1957 those that look ahead of the current position in the subject string, and those
1958 that look behind it. An assertion subpattern is matched in the normal way,
1959 except that it does not cause the current matching position to be changed.
1960 </P>
1961 <P>
1962 Assertion subpatterns are not capturing subpatterns. If such an assertion
1963 contains capturing subpatterns within it, these are counted for the purposes of
1964 numbering the capturing subpatterns in the whole pattern. However, substring
1965 capturing is carried out only for positive assertions. (Perl sometimes, but not
1966 always, does do capturing in negative assertions.)
1967 </P>
1968 <P>
1969 For compatibility with Perl, assertion subpatterns may be repeated; though
1970 it makes no sense to assert the same thing several times, the side effect of
1971 capturing parentheses may occasionally be useful. In practice, there only three
1972 cases:
1973 <br>
1974 <br>
1975 (1) If the quantifier is {0}, the assertion is never obeyed during matching.
1976 However, it may contain internal capturing parenthesized groups that are called
1977 from elsewhere via the
1978 <a href="#subpatternsassubroutines">subroutine mechanism.</a>
1979 <br>
1980 <br>
1981 (2) If quantifier is {0,n} where n is greater than zero, it is treated as if it
1982 were {0,1}. At run time, the rest of the pattern match is tried with and
1983 without the assertion, the order depending on the greediness of the quantifier.
1984 <br>
1985 <br>
1986 (3) If the minimum repetition is greater than zero, the quantifier is ignored.
1987 The assertion is obeyed just once when encountered during matching.
1988 </P>
1989 <br><b>
1990 Lookahead assertions
1991 </b><br>
1992 <P>
1993 Lookahead assertions start with (?= for positive assertions and (?! for
1994 negative assertions. For example,
1995 <pre>
1996 \w+(?=;)
1997 </pre>
1998 matches a word followed by a semicolon, but does not include the semicolon in
1999 the match, and
2000 <pre>
2001 foo(?!bar)
2002 </pre>
2003 matches any occurrence of "foo" that is not followed by "bar". Note that the
2004 apparently similar pattern
2005 <pre>
2006 (?!foo)bar
2007 </pre>
2008 does not find an occurrence of "bar" that is preceded by something other than
2009 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
2010 (?!foo) is always true when the next three characters are "bar". A
2011 lookbehind assertion is needed to achieve the other effect.
2012 </P>
2013 <P>
2014 If you want to force a matching failure at some point in a pattern, the most
2015 convenient way to do it is with (?!) because an empty string always matches, so
2016 an assertion that requires there not to be an empty string must always fail.
2017 The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).
2018 <a name="lookbehind"></a></P>
2019 <br><b>
2020 Lookbehind assertions
2021 </b><br>
2022 <P>
2023 Lookbehind assertions start with (?&#60;= for positive assertions and (?&#60;! for
2024 negative assertions. For example,
2025 <pre>
2026 (?&#60;!foo)bar
2027 </pre>
2028 does find an occurrence of "bar" that is not preceded by "foo". The contents of
2029 a lookbehind assertion are restricted such that all the strings it matches must
2030 have a fixed length. However, if there are several top-level alternatives, they
2031 do not all have to have the same fixed length. Thus
2032 <pre>
2033 (?&#60;=bullock|donkey)
2034 </pre>
2035 is permitted, but
2036 <pre>
2037 (?&#60;!dogs?|cats?)
2038 </pre>
2039 causes an error at compile time. Branches that match different length strings
2040 are permitted only at the top level of a lookbehind assertion. This is an
2041 extension compared with Perl, which requires all branches to match the same
2042 length of string. An assertion such as
2043 <pre>
2044 (?&#60;=ab(c|de))
2045 </pre>
2046 is not permitted, because its single top-level branch can match two different
2047 lengths, but it is acceptable to PCRE if rewritten to use two top-level
2048 branches:
2049 <pre>
2050 (?&#60;=abc|abde)
2051 </pre>
2052 In some cases, the escape sequence \K
2053 <a href="#resetmatchstart">(see above)</a>
2054 can be used instead of a lookbehind assertion to get round the fixed-length
2055 restriction.
2056 </P>
2057 <P>
2058 The implementation of lookbehind assertions is, for each alternative, to
2059 temporarily move the current position back by the fixed length and then try to
2060 match. If there are insufficient characters before the current position, the
2061 assertion fails.
2062 </P>
2063 <P>
2064 In a UTF mode, PCRE does not allow the \C escape (which matches a single data
2065 unit even in a UTF mode) to appear in lookbehind assertions, because it makes
2066 it impossible to calculate the length of the lookbehind. The \X and \R
2067 escapes, which can match different numbers of data units, are also not
2068 permitted.
2069 </P>
2070 <P>
2071 <a href="#subpatternsassubroutines">"Subroutine"</a>
2072 calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
2073 as the subpattern matches a fixed-length string.
2074 <a href="#recursion">Recursion,</a>
2075 however, is not supported.
2076 </P>
2077 <P>
2078 Possessive quantifiers can be used in conjunction with lookbehind assertions to
2079 specify efficient matching of fixed-length strings at the end of subject
2080 strings. Consider a simple pattern such as
2081 <pre>
2082 abcd$
2083 </pre>
2084 when applied to a long string that does not match. Because matching proceeds
2085 from left to right, PCRE will look for each "a" in the subject and then see if
2086 what follows matches the rest of the pattern. If the pattern is specified as
2087 <pre>
2088 ^.*abcd$
2089 </pre>
2090 the initial .* matches the entire string at first, but when this fails (because
2091 there is no following "a"), it backtracks to match all but the last character,
2092 then all but the last two characters, and so on. Once again the search for "a"
2093 covers the entire string, from right to left, so we are no better off. However,
2094 if the pattern is written as
2095 <pre>
2096 ^.*+(?&#60;=abcd)
2097 </pre>
2098 there can be no backtracking for the .*+ item; it can match only the entire
2099 string. The subsequent lookbehind assertion does a single test on the last four
2100 characters. If it fails, the match fails immediately. For long strings, this
2101 approach makes a significant difference to the processing time.
2102 </P>
2103 <br><b>
2104 Using multiple assertions
2105 </b><br>
2106 <P>
2107 Several assertions (of any sort) may occur in succession. For example,
2108 <pre>
2109 (?&#60;=\d{3})(?&#60;!999)foo
2110 </pre>
2111 matches "foo" preceded by three digits that are not "999". Notice that each of
2112 the assertions is applied independently at the same point in the subject
2113 string. First there is a check that the previous three characters are all
2114 digits, and then there is a check that the same three characters are not "999".
2115 This pattern does <i>not</i> match "foo" preceded by six characters, the first
2116 of which are digits and the last three of which are not "999". For example, it
2117 doesn't match "123abcfoo". A pattern to do that is
2118 <pre>
2119 (?&#60;=\d{3}...)(?&#60;!999)foo
2120 </pre>
2121 This time the first assertion looks at the preceding six characters, checking
2122 that the first three are digits, and then the second assertion checks that the
2123 preceding three characters are not "999".
2124 </P>
2125 <P>
2126 Assertions can be nested in any combination. For example,
2127 <pre>
2128 (?&#60;=(?&#60;!foo)bar)baz
2129 </pre>
2130 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
2131 preceded by "foo", while
2132 <pre>
2133 (?&#60;=\d{3}(?!999)...)foo
2134 </pre>
2135 is another pattern that matches "foo" preceded by three digits and any three
2136 characters that are not "999".
2137 <a name="conditions"></a></P>
2138 <br><a name="SEC20" href="#TOC1">CONDITIONAL SUBPATTERNS</a><br>
2139 <P>
2140 It is possible to cause the matching process to obey a subpattern
2141 conditionally or to choose between two alternative subpatterns, depending on
2142 the result of an assertion, or whether a specific capturing subpattern has
2143 already been matched. The two possible forms of conditional subpattern are:
2144 <pre>
2145 (?(condition)yes-pattern)
2146 (?(condition)yes-pattern|no-pattern)
2147 </pre>
2148 If the condition is satisfied, the yes-pattern is used; otherwise the
2149 no-pattern (if present) is used. If there are more than two alternatives in the
2150 subpattern, a compile-time error occurs. Each of the two alternatives may
2151 itself contain nested subpatterns of any form, including conditional
2152 subpatterns; the restriction to two alternatives applies only at the level of
2153 the condition. This pattern fragment is an example where the alternatives are
2154 complex:
2155 <pre>
2156 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
2158 </PRE>
2159 </P>
2160 <P>
2161 There are four kinds of condition: references to subpatterns, references to
2162 recursion, a pseudo-condition called DEFINE, and assertions.
2163 </P>
2164 <br><b>
2165 Checking for a used subpattern by number
2166 </b><br>
2167 <P>
2168 If the text between the parentheses consists of a sequence of digits, the
2169 condition is true if a capturing subpattern of that number has previously
2170 matched. If there is more than one capturing subpattern with the same number
2171 (see the earlier
2172 <a href="#recursion">section about duplicate subpattern numbers),</a>
2173 the condition is true if any of them have matched. An alternative notation is
2174 to precede the digits with a plus or minus sign. In this case, the subpattern
2175 number is relative rather than absolute. The most recently opened parentheses
2176 can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside
2177 loops it can also make sense to refer to subsequent groups. The next
2178 parentheses to be opened can be referenced as (?(+1), and so on. (The value
2179 zero in any of these forms is not used; it provokes a compile-time error.)
2180 </P>
2181 <P>
2182 Consider the following pattern, which contains non-significant white space to
2183 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
2184 three parts for ease of discussion:
2185 <pre>
2186 ( \( )? [^()]+ (?(1) \) )
2187 </pre>
2188 The first part matches an optional opening parenthesis, and if that
2189 character is present, sets it as the first captured substring. The second part
2190 matches one or more characters that are not parentheses. The third part is a
2191 conditional subpattern that tests whether or not the first set of parentheses
2192 matched. If they did, that is, if subject started with an opening parenthesis,
2193 the condition is true, and so the yes-pattern is executed and a closing
2194 parenthesis is required. Otherwise, since no-pattern is not present, the
2195 subpattern matches nothing. In other words, this pattern matches a sequence of
2196 non-parentheses, optionally enclosed in parentheses.
2197 </P>
2198 <P>
2199 If you were embedding this pattern in a larger one, you could use a relative
2200 reference:
2201 <pre>
2202 ...other stuff... ( \( )? [^()]+ (?(-1) \) ) ...
2203 </pre>
2204 This makes the fragment independent of the parentheses in the larger pattern.
2205 </P>
2206 <br><b>
2207 Checking for a used subpattern by name
2208 </b><br>
2209 <P>
2210 Perl uses the syntax (?(&#60;name&#62;)...) or (?('name')...) to test for a used
2211 subpattern by name. For compatibility with earlier versions of PCRE, which had
2212 this facility before Perl, the syntax (?(name)...) is also recognized. However,
2213 there is a possible ambiguity with this syntax, because subpattern names may
2214 consist entirely of digits. PCRE looks first for a named subpattern; if it
2215 cannot find one and the name consists entirely of digits, PCRE looks for a
2216 subpattern of that number, which must be greater than zero. Using subpattern
2217 names that consist entirely of digits is not recommended.
2218 </P>
2219 <P>
2220 Rewriting the above example to use a named subpattern gives this:
2221 <pre>
2222 (?&#60;OPEN&#62; \( )? [^()]+ (?(&#60;OPEN&#62;) \) )
2223 </pre>
2224 If the name used in a condition of this kind is a duplicate, the test is
2225 applied to all subpatterns of the same name, and is true if any one of them has
2226 matched.
2227 </P>
2228 <br><b>
2229 Checking for pattern recursion
2230 </b><br>
2231 <P>
2232 If the condition is the string (R), and there is no subpattern with the name R,
2233 the condition is true if a recursive call to the whole pattern or any
2234 subpattern has been made. If digits or a name preceded by ampersand follow the
2235 letter R, for example:
2236 <pre>
2237 (?(R3)...) or (?(R&name)...)
2238 </pre>
2239 the condition is true if the most recent recursion is into a subpattern whose
2240 number or name is given. This condition does not check the entire recursion
2241 stack. If the name used in a condition of this kind is a duplicate, the test is
2242 applied to all subpatterns of the same name, and is true if any one of them is
2243 the most recent recursion.
2244 </P>
2245 <P>
2246 At "top level", all these recursion test conditions are false.
2247 <a href="#recursion">The syntax for recursive patterns</a>
2248 is described below.
2249 <a name="subdefine"></a></P>
2250 <br><b>
2251 Defining subpatterns for use by reference only
2252 </b><br>
2253 <P>
2254 If the condition is the string (DEFINE), and there is no subpattern with the
2255 name DEFINE, the condition is always false. In this case, there may be only one
2256 alternative in the subpattern. It is always skipped if control reaches this
2257 point in the pattern; the idea of DEFINE is that it can be used to define
2258 subroutines that can be referenced from elsewhere. (The use of
2259 <a href="#subpatternsassubroutines">subroutines</a>
2260 is described below.) For example, a pattern to match an IPv4 address such as
2261 "" could be written like this (ignore white space and line
2262 breaks):
2263 <pre>
2264 (?(DEFINE) (?&#60;byte&#62; 2[0-4]\d | 25[0-5] | 1\d\d | [1-9]?\d) )
2265 \b (?&byte) (\.(?&byte)){3} \b
2266 </pre>
2267 The first part of the pattern is a DEFINE group inside which a another group
2268 named "byte" is defined. This matches an individual component of an IPv4
2269 address (a number less than 256). When matching takes place, this part of the
2270 pattern is skipped because DEFINE acts like a false condition. The rest of the
2271 pattern uses references to the named group to match the four dot-separated
2272 components of an IPv4 address, insisting on a word boundary at each end.
2273 </P>
2274 <br><b>
2275 Assertion conditions
2276 </b><br>
2277 <P>
2278 If the condition is not in any of the above formats, it must be an assertion.
2279 This may be a positive or negative lookahead or lookbehind assertion. Consider
2280 this pattern, again containing non-significant white space, and with the two
2281 alternatives on the second line:
2282 <pre>
2283 (?(?=[^a-z]*[a-z])
2284 \d{2}-[a-z]{3}-\d{2} | \d{2}-\d{2}-\d{2} )
2285 </pre>
2286 The condition is a positive lookahead assertion that matches an optional
2287 sequence of non-letters followed by a letter. In other words, it tests for the
2288 presence of at least one letter in the subject. If a letter is found, the
2289 subject is matched against the first alternative; otherwise it is matched
2290 against the second. This pattern matches strings in one of the two forms
2291 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2292 <a name="comments"></a></P>
2293 <br><a name="SEC21" href="#TOC1">COMMENTS</a><br>
2294 <P>
2295 There are two ways of including comments in patterns that are processed by
2296 PCRE. In both cases, the start of the comment must not be in a character class,
2297 nor in the middle of any other sequence of related characters such as (?: or a
2298 subpattern name or number. The characters that make up a comment play no part
2299 in the pattern matching.
2300 </P>
2301 <P>
2302 The sequence (?# marks the start of a comment that continues up to the next
2303 closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED
2304 option is set, an unescaped # character also introduces a comment, which in
2305 this case continues to immediately after the next newline character or
2306 character sequence in the pattern. Which characters are interpreted as newlines
2307 is controlled by the options passed to a compiling function or by a special
2308 sequence at the start of the pattern, as described in the section entitled
2309 <a href="#newlines">"Newline conventions"</a>
2310 above. Note that the end of this type of comment is a literal newline sequence
2311 in the pattern; escape sequences that happen to represent a newline do not
2312 count. For example, consider this pattern when PCRE_EXTENDED is set, and the
2313 default newline convention is in force:
2314 <pre>
2315 abc #comment \n still comment
2316 </pre>
2317 On encountering the # character, <b>pcre_compile()</b> skips along, looking for
2318 a newline in the pattern. The sequence \n is still literal at this stage, so
2319 it does not terminate the comment. Only an actual character with the code value
2320 0x0a (the default newline) does so.
2321 <a name="recursion"></a></P>
2322 <br><a name="SEC22" href="#TOC1">RECURSIVE PATTERNS</a><br>
2323 <P>
2324 Consider the problem of matching a string in parentheses, allowing for
2325 unlimited nested parentheses. Without the use of recursion, the best that can
2326 be done is to use a pattern that matches up to some fixed depth of nesting. It
2327 is not possible to handle an arbitrary nesting depth.
2328 </P>
2329 <P>
2330 For some time, Perl has provided a facility that allows regular expressions to
2331 recurse (amongst other things). It does this by interpolating Perl code in the
2332 expression at run time, and the code can refer to the expression itself. A Perl
2333 pattern using code interpolation to solve the parentheses problem can be
2334 created like this:
2335 <pre>
2336 $re = qr{\( (?: (?&#62;[^()]+) | (?p{$re}) )* \)}x;
2337 </pre>
2338 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2339 recursively to the pattern in which it appears.
2340 </P>
2341 <P>
2342 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2343 supports special syntax for recursion of the entire pattern, and also for
2344 individual subpattern recursion. After its introduction in PCRE and Python,
2345 this kind of recursion was subsequently introduced into Perl at release 5.10.
2346 </P>
2347 <P>
2348 A special item that consists of (? followed by a number greater than zero and a
2349 closing parenthesis is a recursive subroutine call of the subpattern of the
2350 given number, provided that it occurs inside that subpattern. (If not, it is a
2351 <a href="#subpatternsassubroutines">non-recursive subroutine</a>
2352 call, which is described in the next section.) The special item (?R) or (?0) is
2353 a recursive call of the entire regular expression.
2354 </P>
2355 <P>
2356 This PCRE pattern solves the nested parentheses problem (assume the
2357 PCRE_EXTENDED option is set so that white space is ignored):
2358 <pre>
2359 \( ( [^()]++ | (?R) )* \)
2360 </pre>
2361 First it matches an opening parenthesis. Then it matches any number of
2362 substrings which can either be a sequence of non-parentheses, or a recursive
2363 match of the pattern itself (that is, a correctly parenthesized substring).
2364 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2365 to avoid backtracking into sequences of non-parentheses.
2366 </P>
2367 <P>
2368 If this were part of a larger pattern, you would not want to recurse the entire
2369 pattern, so instead you could use this:
2370 <pre>
2371 ( \( ( [^()]++ | (?1) )* \) )
2372 </pre>
2373 We have put the pattern into parentheses, and caused the recursion to refer to
2374 them instead of the whole pattern.
2375 </P>
2376 <P>
2377 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2378 is made easier by the use of relative references. Instead of (?1) in the
2379 pattern above you can write (?-2) to refer to the second most recently opened
2380 parentheses preceding the recursion. In other words, a negative number counts
2381 capturing parentheses leftwards from the point at which it is encountered.
2382 </P>
2383 <P>
2384 It is also possible to refer to subsequently opened parentheses, by writing
2385 references such as (?+2). However, these cannot be recursive because the
2386 reference is not inside the parentheses that are referenced. They are always
2387 <a href="#subpatternsassubroutines">non-recursive subroutine</a>
2388 calls, as described in the next section.
2389 </P>
2390 <P>
2391 An alternative approach is to use named parentheses instead. The Perl syntax
2392 for this is (?&name); PCRE's earlier syntax (?P&#62;name) is also supported. We
2393 could rewrite the above example as follows:
2394 <pre>
2395 (?&#60;pn&#62; \( ( [^()]++ | (?&pn) )* \) )
2396 </pre>
2397 If there is more than one subpattern with the same name, the earliest one is
2398 used.
2399 </P>
2400 <P>
2401 This particular example pattern that we have been looking at contains nested
2402 unlimited repeats, and so the use of a possessive quantifier for matching
2403 strings of non-parentheses is important when applying the pattern to strings
2404 that do not match. For example, when this pattern is applied to
2405 <pre>
2406 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2407 </pre>
2408 it yields "no match" quickly. However, if a possessive quantifier is not used,
2409 the match runs for a very long time indeed because there are so many different
2410 ways the + and * repeats can carve up the subject, and all have to be tested
2411 before failure can be reported.
2412 </P>
2413 <P>
2414 At the end of a match, the values of capturing parentheses are those from
2415 the outermost level. If you want to obtain intermediate values, a callout
2416 function can be used (see below and the
2417 <a href="pcrecallout.html"><b>pcrecallout</b></a>
2418 documentation). If the pattern above is matched against
2419 <pre>
2420 (ab(cd)ef)
2421 </pre>
2422 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2423 the last value taken on at the top level. If a capturing subpattern is not
2424 matched at the top level, its final captured value is unset, even if it was
2425 (temporarily) set at a deeper level during the matching process.
2426 </P>
2427 <P>
2428 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2429 obtain extra memory to store data during a recursion, which it does by using
2430 <b>pcre_malloc</b>, freeing it via <b>pcre_free</b> afterwards. If no memory can
2431 be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2432 </P>
2433 <P>
2434 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2435 Consider this pattern, which matches text in angle brackets, allowing for
2436 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2437 recursing), whereas any characters are permitted at the outer level.
2438 <pre>
2439 &#60; (?: (?(R) \d++ | [^&#60;&#62;]*+) | (?R)) * &#62;
2440 </pre>
2441 In this pattern, (?(R) is the start of a conditional subpattern, with two
2442 different alternatives for the recursive and non-recursive cases. The (?R) item
2443 is the actual recursive call.
2444 <a name="recursiondifference"></a></P>
2445 <br><b>
2446 Differences in recursion processing between PCRE and Perl
2447 </b><br>
2448 <P>
2449 Recursion processing in PCRE differs from Perl in two important ways. In PCRE
2450 (like Python, but unlike Perl), a recursive subpattern call is always treated
2451 as an atomic group. That is, once it has matched some of the subject string, it
2452 is never re-entered, even if it contains untried alternatives and there is a
2453 subsequent matching failure. This can be illustrated by the following pattern,
2454 which purports to match a palindromic string that contains an odd number of
2455 characters (for example, "a", "aba", "abcba", "abcdcba"):
2456 <pre>
2457 ^(.|(.)(?1)\2)$
2458 </pre>
2459 The idea is that it either matches a single character, or two identical
2460 characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2461 it does not if the pattern is longer than three characters. Consider the
2462 subject string "abcba":
2463 </P>
2464 <P>
2465 At the top level, the first character is matched, but as it is not at the end
2466 of the string, the first alternative fails; the second alternative is taken
2467 and the recursion kicks in. The recursive call to subpattern 1 successfully
2468 matches the next character ("b"). (Note that the beginning and end of line
2469 tests are not part of the recursion).
2470 </P>
2471 <P>
2472 Back at the top level, the next character ("c") is compared with what
2473 subpattern 2 matched, which was "a". This fails. Because the recursion is
2474 treated as an atomic group, there are now no backtracking points, and so the
2475 entire match fails. (Perl is able, at this point, to re-enter the recursion and
2476 try the second alternative.) However, if the pattern is written with the
2477 alternatives in the other order, things are different:
2478 <pre>
2479 ^((.)(?1)\2|.)$
2480 </pre>
2481 This time, the recursing alternative is tried first, and continues to recurse
2482 until it runs out of characters, at which point the recursion fails. But this
2483 time we do have another alternative to try at the higher level. That is the big
2484 difference: in the previous case the remaining alternative is at a deeper
2485 recursion level, which PCRE cannot use.
2486 </P>
2487 <P>
2488 To change the pattern so that it matches all palindromic strings, not just
2489 those with an odd number of characters, it is tempting to change the pattern to
2490 this:
2491 <pre>
2492 ^((.)(?1)\2|.?)$
2493 </pre>
2494 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2495 deeper recursion has matched a single character, it cannot be entered again in
2496 order to match an empty string. The solution is to separate the two cases, and
2497 write out the odd and even cases as alternatives at the higher level:
2498 <pre>
2499 ^(?:((.)(?1)\2|)|((.)(?3)\4|.))
2500 </pre>
2501 If you want to match typical palindromic phrases, the pattern has to ignore all
2502 non-word characters, which can be done like this:
2503 <pre>
2504 ^\W*+(?:((.)\W*+(?1)\W*+\2|)|((.)\W*+(?3)\W*+\4|\W*+.\W*+))\W*+$
2505 </pre>
2506 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2507 man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2508 the use of the possessive quantifier *+ to avoid backtracking into sequences of
2509 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2510 more) to match typical phrases, and Perl takes so long that you think it has
2511 gone into a loop.
2512 </P>
2513 <P>
2514 <b>WARNING</b>: The palindrome-matching patterns above work only if the subject
2515 string does not start with a palindrome that is shorter than the entire string.
2516 For example, although "abcba" is correctly matched, if the subject is "ababa",
2517 PCRE finds the palindrome "aba" at the start, then fails at top level because
2518 the end of the string does not follow. Once again, it cannot jump back into the
2519 recursion to try other alternatives, so the entire match fails.
2520 </P>
2521 <P>
2522 The second way in which PCRE and Perl differ in their recursion processing is
2523 in the handling of captured values. In Perl, when a subpattern is called
2524 recursively or as a subpattern (see the next section), it has no access to any
2525 values that were captured outside the recursion, whereas in PCRE these values
2526 can be referenced. Consider this pattern:
2527 <pre>
2528 ^(.)(\1|a(?2))
2529 </pre>
2530 In PCRE, this pattern matches "bab". The first capturing parentheses match "b",
2531 then in the second group, when the back reference \1 fails to match "b", the
2532 second alternative matches "a" and then recurses. In the recursion, \1 does
2533 now match "b" and so the whole match succeeds. In Perl, the pattern fails to
2534 match because inside the recursive call \1 cannot access the externally set
2535 value.
2536 <a name="subpatternsassubroutines"></a></P>
2537 <br><a name="SEC23" href="#TOC1">SUBPATTERNS AS SUBROUTINES</a><br>
2538 <P>
2539 If the syntax for a recursive subpattern call (either by number or by
2540 name) is used outside the parentheses to which it refers, it operates like a
2541 subroutine in a programming language. The called subpattern may be defined
2542 before or after the reference. A numbered reference can be absolute or
2543 relative, as in these examples:
2544 <pre>
2545 (...(absolute)...)...(?2)...
2546 (...(relative)...)...(?-1)...
2547 (...(?+1)...(relative)...
2548 </pre>
2549 An earlier example pointed out that the pattern
2550 <pre>
2551 (sens|respons)e and \1ibility
2552 </pre>
2553 matches "sense and sensibility" and "response and responsibility", but not
2554 "sense and responsibility". If instead the pattern
2555 <pre>
2556 (sens|respons)e and (?1)ibility
2557 </pre>
2558 is used, it does match "sense and responsibility" as well as the other two
2559 strings. Another example is given in the discussion of DEFINE above.
2560 </P>
2561 <P>
2562 All subroutine calls, whether recursive or not, are always treated as atomic
2563 groups. That is, once a subroutine has matched some of the subject string, it
2564 is never re-entered, even if it contains untried alternatives and there is a
2565 subsequent matching failure. Any capturing parentheses that are set during the
2566 subroutine call revert to their previous values afterwards.
2567 </P>
2568 <P>
2569 Processing options such as case-independence are fixed when a subpattern is
2570 defined, so if it is used as a subroutine, such options cannot be changed for
2571 different calls. For example, consider this pattern:
2572 <pre>
2573 (abc)(?i:(?-1))
2574 </pre>
2575 It matches "abcabc". It does not match "abcABC" because the change of
2576 processing option does not affect the called subpattern.
2577 <a name="onigurumasubroutines"></a></P>
2578 <br><a name="SEC24" href="#TOC1">ONIGURUMA SUBROUTINE SYNTAX</a><br>
2579 <P>
2580 For compatibility with Oniguruma, the non-Perl syntax \g followed by a name or
2581 a number enclosed either in angle brackets or single quotes, is an alternative
2582 syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2583 are two of the examples used above, rewritten using this syntax:
2584 <pre>
2585 (?&#60;pn&#62; \( ( (?&#62;[^()]+) | \g&#60;pn&#62; )* \) )
2586 (sens|respons)e and \g'1'ibility
2587 </pre>
2588 PCRE supports an extension to Oniguruma: if a number is preceded by a
2589 plus or a minus sign it is taken as a relative reference. For example:
2590 <pre>
2591 (abc)(?i:\g&#60;-1&#62;)
2592 </pre>
2593 Note that \g{...} (Perl syntax) and \g&#60;...&#62; (Oniguruma syntax) are <i>not</i>
2594 synonymous. The former is a back reference; the latter is a subroutine call.
2595 </P>
2596 <br><a name="SEC25" href="#TOC1">CALLOUTS</a><br>
2597 <P>
2598 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2599 code to be obeyed in the middle of matching a regular expression. This makes it
2600 possible, amongst other things, to extract different substrings that match the
2601 same pair of parentheses when there is a repetition.
2602 </P>
2603 <P>
2604 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2605 code. The feature is called "callout". The caller of PCRE provides an external
2606 function by putting its entry point in the global variable <i>pcre_callout</i>
2607 (8-bit library) or <i>pcre[16|32]_callout</i> (16-bit or 32-bit library).
2608 By default, this variable contains NULL, which disables all calling out.
2609 </P>
2610 <P>
2611 Within a regular expression, (?C) indicates the points at which the external
2612 function is to be called. If you want to identify different callout points, you
2613 can put a number less than 256 after the letter C. The default value is zero.
2614 For example, this pattern has two callout points:
2615 <pre>
2616 (?C1)abc(?C2)def
2617 </pre>
2618 If the PCRE_AUTO_CALLOUT flag is passed to a compiling function, callouts are
2619 automatically installed before each item in the pattern. They are all numbered
2620 255. If there is a conditional group in the pattern whose condition is an
2621 assertion, an additional callout is inserted just before the condition. An
2622 explicit callout may also be set at this position, as in this example:
2623 <pre>
2624 (?(?C9)(?=a)abc|def)
2625 </pre>
2626 Note that this applies only to assertion conditions, not to other types of
2627 condition.
2628 </P>
2629 <P>
2630 During matching, when PCRE reaches a callout point, the external function is
2631 called. It is provided with the number of the callout, the position in the
2632 pattern, and, optionally, one item of data originally supplied by the caller of
2633 the matching function. The callout function may cause matching to proceed, to
2634 backtrack, or to fail altogether. A complete description of the interface to
2635 the callout function is given in the
2636 <a href="pcrecallout.html"><b>pcrecallout</b></a>
2637 documentation.
2638 <a name="backtrackcontrol"></a></P>
2639 <br><a name="SEC26" href="#TOC1">BACKTRACKING CONTROL</a><br>
2640 <P>
2641 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2642 are still described in the Perl documentation as "experimental and subject to
2643 change or removal in a future version of Perl". It goes on to say: "Their usage
2644 in production code should be noted to avoid problems during upgrades." The same
2645 remarks apply to the PCRE features described in this section.
2646 </P>
2647 <P>
2648 The new verbs make use of what was previously invalid syntax: an opening
2649 parenthesis followed by an asterisk. They are generally of the form
2650 (*VERB) or (*VERB:NAME). Some may take either form, with differing behaviour,
2651 depending on whether or not a name is present. A name is any sequence of
2652 characters that does not include a closing parenthesis. The maximum length of
2653 name is 255 in the 8-bit library and 65535 in the 16-bit and 32-bit libraries.
2654 If the name is empty, that is, if the closing parenthesis immediately follows
2655 the colon, the effect is as if the colon were not there. Any number of these
2656 verbs may occur in a pattern.
2657 </P>
2658 <P>
2659 Since these verbs are specifically related to backtracking, most of them can be
2660 used only when the pattern is to be matched using one of the traditional
2661 matching functions, because these use a backtracking algorithm. With the
2662 exception of (*FAIL), which behaves like a failing negative assertion, the
2663 backtracking control verbs cause an error if encountered by a DFA matching
2664 function.
2665 </P>
2666 <P>
2667 The behaviour of these verbs in
2668 <a href="#btrepeat">repeated groups,</a>
2669 <a href="#btassert">assertions,</a>
2670 and in
2671 <a href="#btsub">subpatterns called as subroutines</a>
2672 (whether or not recursively) is documented below.
2673 <a name="nooptimize"></a></P>
2674 <br><b>
2675 Optimizations that affect backtracking verbs
2676 </b><br>
2677 <P>
2678 PCRE contains some optimizations that are used to speed up matching by running
2679 some checks at the start of each match attempt. For example, it may know the
2680 minimum length of matching subject, or that a particular character must be
2681 present. When one of these optimizations bypasses the running of a match, any
2682 included backtracking verbs will not, of course, be processed. You can suppress
2683 the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2684 when calling <b>pcre_compile()</b> or <b>pcre_exec()</b>, or by starting the
2685 pattern with (*NO_START_OPT). There is more discussion of this option in the
2686 section entitled
2687 <a href="pcreapi.html#execoptions">"Option bits for <b>pcre_exec()</b>"</a>
2688 in the
2689 <a href="pcreapi.html"><b>pcreapi</b></a>
2690 documentation.
2691 </P>
2692 <P>
2693 Experiments with Perl suggest that it too has similar optimizations, sometimes
2694 leading to anomalous results.
2695 </P>
2696 <br><b>
2697 Verbs that act immediately
2698 </b><br>
2699 <P>
2700 The following verbs act as soon as they are encountered. They may not be
2701 followed by a name.
2702 <pre>
2703 (*ACCEPT)
2704 </pre>
2705 This verb causes the match to end successfully, skipping the remainder of the
2706 pattern. However, when it is inside a subpattern that is called as a
2707 subroutine, only that subpattern is ended successfully. Matching then continues
2708 at the outer level. If (*ACCEPT) in triggered in a positive assertion, the
2709 assertion succeeds; in a negative assertion, the assertion fails.
2710 </P>
2711 <P>
2712 If (*ACCEPT) is inside capturing parentheses, the data so far is captured. For
2713 example:
2714 <pre>
2715 A((?:A|B(*ACCEPT)|C)D)
2716 </pre>
2717 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2718 the outer parentheses.
2719 <pre>
2720 (*FAIL) or (*F)
2721 </pre>
2722 This verb causes a matching failure, forcing backtracking to occur. It is
2723 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2724 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2725 Perl features that are not present in PCRE. The nearest equivalent is the
2726 callout feature, as for example in this pattern:
2727 <pre>
2728 a+(?C)(*FAIL)
2729 </pre>
2730 A match with the string "aaaa" always fails, but the callout is taken before
2731 each backtrack happens (in this example, 10 times).
2732 </P>
2733 <br><b>
2734 Recording which path was taken
2735 </b><br>
2736 <P>
2737 There is one verb whose main purpose is to track how a match was arrived at,
2738 though it also has a secondary use in conjunction with advancing the match
2739 starting point (see (*SKIP) below).
2740 <pre>
2741 (*MARK:NAME) or (*:NAME)
2742 </pre>
2743 A name is always required with this verb. There may be as many instances of
2744 (*MARK) as you like in a pattern, and their names do not have to be unique.
2745 </P>
2746 <P>
2747 When a match succeeds, the name of the last-encountered (*MARK:NAME),
2748 (*PRUNE:NAME), or (*THEN:NAME) on the matching path is passed back to the
2749 caller as described in the section entitled
2750 <a href="pcreapi.html#extradata">"Extra data for <b>pcre_exec()</b>"</a>
2751 in the
2752 <a href="pcreapi.html"><b>pcreapi</b></a>
2753 documentation. Here is an example of <b>pcretest</b> output, where the /K
2754 modifier requests the retrieval and outputting of (*MARK) data:
2755 <pre>
2756 re&#62; /X(*MARK:A)Y|X(*MARK:B)Z/K
2757 data&#62; XY
2758 0: XY
2759 MK: A
2760 XZ
2761 0: XZ
2762 MK: B
2763 </pre>
2764 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2765 indicates which of the two alternatives matched. This is a more efficient way
2766 of obtaining this information than putting each alternative in its own
2767 capturing parentheses.
2768 </P>
2769 <P>
2770 If a verb with a name is encountered in a positive assertion, its name is
2771 recorded and passed back if it is the last-encountered. This does not happen
2772 for negative assertions.
2773 </P>
2774 <P>
2775 After a partial match or a failed match, the last encountered name in the
2776 entire match process is returned. For example:
2777 <pre>
2778 re&#62; /X(*MARK:A)Y|X(*MARK:B)Z/K
2779 data&#62; XP
2780 No match, mark = B
2781 </pre>
2782 Note that in this unanchored example the mark is retained from the match
2783 attempt that started at the letter "X" in the subject. Subsequent match
2784 attempts starting at "P" and then with an empty string do not get as far as the
2785 (*MARK) item, but nevertheless do not reset it.
2786 </P>
2787 <P>
2788 If you are interested in (*MARK) values after failed matches, you should
2789 probably set the PCRE_NO_START_OPTIMIZE option
2790 <a href="#nooptimize">(see above)</a>
2791 to ensure that the match is always attempted.
2792 </P>
2793 <br><b>
2794 Verbs that act after backtracking
2795 </b><br>
2796 <P>
2797 The following verbs do nothing when they are encountered. Matching continues
2798 with what follows, but if there is no subsequent match, causing a backtrack to
2799 the verb, a failure is forced. That is, backtracking cannot pass to the left of
2800 the verb. However, when one of these verbs appears inside an atomic group or an
2801 assertion, its effect is confined to that group, because once the group has
2802 been matched, there is never any backtracking into it. In this situation,
2803 backtracking can "jump back" to the left of the entire atomic group or
2804 assertion. (Remember also, as stated above, that this localization also applies
2805 in subroutine calls.)
2806 </P>
2807 <P>
2808 These verbs differ in exactly what kind of failure occurs when backtracking
2809 reaches them.
2810 <pre>
2811 (*COMMIT)
2812 </pre>
2813 This verb, which may not be followed by a name, causes the whole match to fail
2814 outright if there is a later matching failure that causes backtracking to reach
2815 it. Even if the pattern is unanchored, no further attempts to find a match by
2816 advancing the starting point take place. If (*COMMIT) is the only backtracking
2817 verb that is encountered, once it has been passed <b>pcre_exec()</b> is
2818 committed to finding a match at the current starting point, or not at all. For
2819 example:
2820 <pre>
2821 a+(*COMMIT)b
2822 </pre>
2823 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2824 dynamic anchor, or "I've started, so I must finish." The name of the most
2825 recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2826 match failure.
2827 </P>
2828 <P>
2829 If there is more than one backtracking verb in a pattern, a different one that
2830 follows (*COMMIT) may be triggered first, so merely passing (*COMMIT) during a
2831 match does not always guarantee that a match must be at this starting point.
2832 </P>
2833 <P>
2834 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2835 unless PCRE's start-of-match optimizations are turned off, as shown in this
2836 <b>pcretest</b> example:
2837 <pre>
2838 re&#62; /(*COMMIT)abc/
2839 data&#62; xyzabc
2840 0: abc
2841 xyzabc\Y
2842 No match
2843 </pre>
2844 PCRE knows that any match must start with "a", so the optimization skips along
2845 the subject to "a" before running the first match attempt, which succeeds. When
2846 the optimization is disabled by the \Y escape in the second subject, the match
2847 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2848 starting points.
2849 <pre>
2850 (*PRUNE) or (*PRUNE:NAME)
2851 </pre>
2852 This verb causes the match to fail at the current starting position in the
2853 subject if there is a later matching failure that causes backtracking to reach
2854 it. If the pattern is unanchored, the normal "bumpalong" advance to the next
2855 starting character then happens. Backtracking can occur as usual to the left of
2856 (*PRUNE), before it is reached, or when matching to the right of (*PRUNE), but
2857 if there is no match to the right, backtracking cannot cross (*PRUNE). In
2858 simple cases, the use of (*PRUNE) is just an alternative to an atomic group or
2859 possessive quantifier, but there are some uses of (*PRUNE) that cannot be
2860 expressed in any other way. In an anchored pattern (*PRUNE) has the same effect
2861 as (*COMMIT).
2862 </P>
2863 <P>
2864 The behaviour of (*PRUNE:NAME) is the not the same as (*MARK:NAME)(*PRUNE).
2865 It is like (*MARK:NAME) in that the name is remembered for passing back to the
2866 caller. However, (*SKIP:NAME) searches only for names set with (*MARK).
2867 <pre>
2868 (*SKIP)
2869 </pre>
2870 This verb, when given without a name, is like (*PRUNE), except that if the
2871 pattern is unanchored, the "bumpalong" advance is not to the next character,
2872 but to the position in the subject where (*SKIP) was encountered. (*SKIP)
2873 signifies that whatever text was matched leading up to it cannot be part of a
2874 successful match. Consider:
2875 <pre>
2876 a+(*SKIP)b
2877 </pre>
2878 If the subject is "aaaac...", after the first match attempt fails (starting at
2879 the first character in the string), the starting point skips on to start the
2880 next attempt at "c". Note that a possessive quantifer does not have the same
2881 effect as this example; although it would suppress backtracking during the
2882 first match attempt, the second attempt would start at the second character
2883 instead of skipping on to "c".
2884 <pre>
2885 (*SKIP:NAME)
2886 </pre>
2887 When (*SKIP) has an associated name, its behaviour is modified. When it is
2888 triggered, the previous path through the pattern is searched for the most
2889 recent (*MARK) that has the same name. If one is found, the "bumpalong" advance
2890 is to the subject position that corresponds to that (*MARK) instead of to where
2891 (*SKIP) was encountered. If no (*MARK) with a matching name is found, the
2892 (*SKIP) is ignored.
2893 </P>
2894 <P>
2895 Note that (*SKIP:NAME) searches only for names set by (*MARK:NAME). It ignores
2896 names that are set by (*PRUNE:NAME) or (*THEN:NAME).
2897 <pre>
2898 (*THEN) or (*THEN:NAME)
2899 </pre>
2900 This verb causes a skip to the next innermost alternative when backtracking
2901 reaches it. That is, it cancels any further backtracking within the current
2902 alternative. Its name comes from the observation that it can be used for a
2903 pattern-based if-then-else block:
2904 <pre>
2905 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2906 </pre>
2907 If the COND1 pattern matches, FOO is tried (and possibly further items after
2908 the end of the group if FOO succeeds); on failure, the matcher skips to the
2909 second alternative and tries COND2, without backtracking into COND1.
2910 If (*THEN) is not inside an alternation, it acts like (*PRUNE).
2911 </P>
2912 <P>
2913 The behaviour of (*THEN:NAME) is the not the same as (*MARK:NAME)(*THEN).
2914 It is like (*MARK:NAME) in that the name is remembered for passing back to the
2915 caller. However, (*SKIP:NAME) searches only for names set with (*MARK).
2916 </P>
2917 <P>
2918 A subpattern that does not contain a | character is just a part of the
2919 enclosing alternative; it is not a nested alternation with only one
2920 alternative. The effect of (*THEN) extends beyond such a subpattern to the
2921 enclosing alternative. Consider this pattern, where A, B, etc. are complex
2922 pattern fragments that do not contain any | characters at this level:
2923 <pre>
2924 A (B(*THEN)C) | D
2925 </pre>
2926 If A and B are matched, but there is a failure in C, matching does not
2927 backtrack into A; instead it moves to the next alternative, that is, D.
2928 However, if the subpattern containing (*THEN) is given an alternative, it
2929 behaves differently:
2930 <pre>
2931 A (B(*THEN)C | (*FAIL)) | D
2932 </pre>
2933 The effect of (*THEN) is now confined to the inner subpattern. After a failure
2934 in C, matching moves to (*FAIL), which causes the whole subpattern to fail
2935 because there are no more alternatives to try. In this case, matching does now
2936 backtrack into A.
2937 </P>
2938 <P>
2939 Note that a conditional subpattern is not considered as having two
2940 alternatives, because only one is ever used. In other words, the | character in
2941 a conditional subpattern has a different meaning. Ignoring white space,
2942 consider:
2943 <pre>
2944 ^.*? (?(?=a) a | b(*THEN)c )
2945 </pre>
2946 If the subject is "ba", this pattern does not match. Because .*? is ungreedy,
2947 it initially matches zero characters. The condition (?=a) then fails, the
2948 character "b" is matched, but "c" is not. At this point, matching does not
2949 backtrack to .*? as might perhaps be expected from the presence of the |
2950 character. The conditional subpattern is part of the single alternative that
2951 comprises the whole pattern, and so the match fails. (If there was a backtrack
2952 into .*?, allowing it to match "b", the match would succeed.)
2953 </P>
2954 <P>
2955 The verbs just described provide four different "strengths" of control when
2956 subsequent matching fails. (*THEN) is the weakest, carrying on the match at the
2957 next alternative. (*PRUNE) comes next, failing the match at the current
2958 starting position, but allowing an advance to the next character (for an
2959 unanchored pattern). (*SKIP) is similar, except that the advance may be more
2960 than one character. (*COMMIT) is the strongest, causing the entire match to
2961 fail.
2962 </P>
2963 <br><b>
2964 More than one backtracking verb
2965 </b><br>
2966 <P>
2967 If more than one backtracking verb is present in a pattern, the one that is
2968 backtracked onto first acts. For example, consider this pattern, where A, B,
2969 etc. are complex pattern fragments:
2970 <pre>
2972 </pre>
2973 If A matches but B fails, the backtrack to (*COMMIT) causes the entire match to
2974 fail. However, if A and B match, but C fails, the backtrack to (*THEN) causes
2975 the next alternative (ABD) to be tried. This behaviour is consistent, but is
2976 not always the same as Perl's. It means that if two or more backtracking verbs
2977 appear in succession, all the the last of them has no effect. Consider this
2978 example:
2979 <pre>
2980 ...(*COMMIT)(*PRUNE)...
2981 </pre>
2982 If there is a matching failure to the right, backtracking onto (*PRUNE) cases
2983 it to be triggered, and its action is taken. There can never be a backtrack
2984 onto (*COMMIT).
2985 <a name="btrepeat"></a></P>
2986 <br><b>
2987 Backtracking verbs in repeated groups
2988 </b><br>
2989 <P>
2990 PCRE differs from Perl in its handling of backtracking verbs in repeated
2991 groups. For example, consider:
2992 <pre>
2993 /(a(*COMMIT)b)+ac/
2994 </pre>
2995 If the subject is "abac", Perl matches, but PCRE fails because the (*COMMIT) in
2996 the second repeat of the group acts.
2997 <a name="btassert"></a></P>
2998 <br><b>
2999 Backtracking verbs in assertions
3000 </b><br>
3001 <P>
3002 (*FAIL) in an assertion has its normal effect: it forces an immediate backtrack.
3003 </P>
3004 <P>
3005 (*ACCEPT) in a positive assertion causes the assertion to succeed without any
3006 further processing. In a negative assertion, (*ACCEPT) causes the assertion to
3007 fail without any further processing.
3008 </P>
3009 <P>
3010 The other backtracking verbs are not treated specially if they appear in an
3011 assertion. In particular, (*THEN) skips to the next alternative in the
3012 innermost enclosing group that has alternations, whether or not this is within
3013 the assertion.
3014 <a name="btsub"></a></P>
3015 <br><b>
3016 Backtracking verbs in subroutines
3017 </b><br>
3018 <P>
3019 These behaviours occur whether or not the subpattern is called recursively.
3020 Perl's treatment of subroutines is different in some cases.
3021 </P>
3022 <P>
3023 (*FAIL) in a subpattern called as a subroutine has its normal effect: it forces
3024 an immediate backtrack.
3025 </P>
3026 <P>
3027 (*ACCEPT) in a subpattern called as a subroutine causes the subroutine match to
3028 succeed without any further processing. Matching then continues after the
3029 subroutine call.
3030 </P>
3031 <P>
3032 (*COMMIT), (*SKIP), and (*PRUNE) in a subpattern called as a subroutine cause
3033 the subroutine match to fail.
3034 </P>
3035 <P>
3036 (*THEN) skips to the next alternative in the innermost enclosing group within
3037 the subpattern that has alternatives. If there is no such group within the
3038 subpattern, (*THEN) causes the subroutine match to fail.
3039 </P>
3040 <br><a name="SEC27" href="#TOC1">SEE ALSO</a><br>
3041 <P>
3042 <b>pcreapi</b>(3), <b>pcrecallout</b>(3), <b>pcrematching</b>(3),
3043 <b>pcresyntax</b>(3), <b>pcre</b>(3), <b>pcre16(3)</b>, <b>pcre32(3)</b>.
3044 </P>
3045 <br><a name="SEC28" href="#TOC1">AUTHOR</a><br>
3046 <P>
3047 Philip Hazel
3048 <br>
3049 University Computing Service
3050 <br>
3051 Cambridge CB2 3QH, England.
3052 <br>
3053 </P>
3054 <br><a name="SEC29" href="#TOC1">REVISION</a><br>
3055 <P>
3056 Last updated: 22 March 2013
3057 <br>
3058 Copyright &copy; 1997-2013 University of Cambridge.
3059 <br>
3060 <p>
3061 Return to the <a href="index.html">PCRE index page</a>.
3062 </p>


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

  ViewVC Help
Powered by ViewVC 1.1.5