ViewVC logotype

Contents of /code/trunk/doc/pcrepattern.3

Parent Directory Parent Directory | Revision Log Revision Log

Revision 457 - (hide annotations) (download)
Sat Oct 3 16:24:08 2009 UTC (5 years, 6 months ago) by ph10
File size: 98077 byte(s)
Allow duplicate names for same-numbered groups; forbid different names.

1 nigel 79 .TH PCREPATTERN 3
2 nigel 63 .SH NAME
3     PCRE - Perl-compatible regular expressions
5 nigel 63 .rs
6     .sp
7 ph10 208 The syntax and semantics of the regular expressions that are supported by PCRE
8     are described in detail below. There is a quick-reference syntax summary in the
9     .\" HREF
10     \fBpcresyntax\fP
11     .\"
12 ph10 333 page. PCRE tries to match Perl syntax and semantics as closely as it can. PCRE
13     also supports some alternative regular expression syntax (which does not
14     conflict with the Perl syntax) in order to provide some compatibility with
15     regular expressions in Python, .NET, and Oniguruma.
16     .P
17     Perl's regular expressions are described in its own documentation, and
18 ph10 208 regular expressions in general are covered in a number of books, some of which
19     have copious examples. Jeffrey Friedl's "Mastering Regular Expressions",
20     published by O'Reilly, covers regular expressions in great detail. This
21     description of PCRE's regular expressions is intended as reference material.
22 nigel 75 .P
23     The original operation of PCRE was on strings of one-byte characters. However,
24 ph10 456 there is now also support for UTF-8 character strings. To use this,
25     PCRE must be built to include UTF-8 support, and you must call
26     \fBpcre_compile()\fP or \fBpcre_compile2()\fP with the PCRE_UTF8 option. There
27     is also a special sequence that can be given at the start of a pattern:
28 ph10 412 .sp
29     (*UTF8)
30 ph10 416 .sp
31 ph10 412 Starting a pattern with this sequence is equivalent to setting the PCRE_UTF8
32     option. This feature is not Perl-compatible. How setting UTF-8 mode affects
33     pattern matching is mentioned in several places below. There is also a summary
34     of UTF-8 features in the
35 nigel 63 .\" HTML <a href="pcre.html#utf8support">
36     .\" </a>
37     section on UTF-8 support
38     .\"
39     in the main
40     .\" HREF
41 nigel 75 \fBpcre\fP
42 nigel 63 .\"
43     page.
44 nigel 75 .P
45 nigel 77 The remainder of this document discusses the patterns that are supported by
46     PCRE when its main matching function, \fBpcre_exec()\fP, is used.
47     From release 6.0, PCRE offers a second matching function,
48     \fBpcre_dfa_exec()\fP, which matches using a different algorithm that is not
49 ph10 172 Perl-compatible. Some of the features discussed below are not available when
50 ph10 168 \fBpcre_dfa_exec()\fP is used. The advantages and disadvantages of the
51     alternative function, and how it differs from the normal function, are
52     discussed in the
53 nigel 77 .\" HREF
54     \fBpcrematching\fP
55     .\"
56     page.
57 nigel 93 .
58     .
60     .rs
61     .sp
62     PCRE supports five different conventions for indicating line breaks in
63     strings: a single CR (carriage return) character, a single LF (linefeed)
64     character, the two-character sequence CRLF, any of the three preceding, or any
65     Unicode newline sequence. The
66     .\" HREF
67     \fBpcreapi\fP
68     .\"
69     page has
70     .\" HTML <a href="pcreapi.html#newlines">
71     .\" </a>
72     further discussion
73     .\"
74     about newlines, and shows how to set the newline convention in the
75     \fIoptions\fP arguments for the compiling and matching functions.
76     .P
77     It is also possible to specify a newline convention by starting a pattern
78     string with one of the following five sequences:
79     .sp
80     (*CR) carriage return
81     (*LF) linefeed
82     (*CRLF) carriage return, followed by linefeed
83     (*ANYCRLF) any of the three above
84     (*ANY) all Unicode newline sequences
85     .sp
86 ph10 456 These override the default and the options given to \fBpcre_compile()\fP or
87     \fBpcre_compile2()\fP. For example, on a Unix system where LF is the default
88     newline sequence, the pattern
89 ph10 227 .sp
90     (*CR)a.b
91     .sp
92     changes the convention to CR. That pattern matches "a\enb" because LF is no
93     longer a newline. Note that these special settings, which are not
94     Perl-compatible, are recognized only at the very start of a pattern, and that
95 ph10 231 they must be in upper case. If more than one of them is present, the last one
96     is used.
97     .P
98     The newline convention does not affect what the \eR escape sequence matches. By
99     default, this is any Unicode newline sequence, for Perl compatibility. However,
100     this can be changed; see the description of \eR in the section entitled
101     .\" HTML <a href="#newlineseq">
102     .\" </a>
103     "Newline sequences"
104     .\"
105 ph10 247 below. A change of \eR setting can be combined with a change of newline
106 ph10 246 convention.
107 ph10 227 .
108     .
110     .rs
111     .sp
112 nigel 63 A regular expression is a pattern that is matched against a subject string from
113     left to right. Most characters stand for themselves in a pattern, and match the
114     corresponding characters in the subject. As a trivial example, the pattern
115 nigel 75 .sp
116 nigel 63 The quick brown fox
117 nigel 75 .sp
118 nigel 77 matches a portion of a subject string that is identical to itself. When
119     caseless matching is specified (the PCRE_CASELESS option), letters are matched
120     independently of case. In UTF-8 mode, PCRE always understands the concept of
121     case for characters whose values are less than 128, so caseless matching is
122     always possible. For characters with higher values, the concept of case is
123     supported if PCRE is compiled with Unicode property support, but not otherwise.
124     If you want to use caseless matching for characters 128 and above, you must
125     ensure that PCRE is compiled with Unicode property support as well as with
126     UTF-8 support.
127     .P
128     The power of regular expressions comes from the ability to include alternatives
129     and repetitions in the pattern. These are encoded in the pattern by the use of
130 nigel 75 \fImetacharacters\fP, which do not stand for themselves but instead are
131 nigel 63 interpreted in some special way.
132 nigel 75 .P
133     There are two different sets of metacharacters: those that are recognized
134 nigel 63 anywhere in the pattern except within square brackets, and those that are
135 nigel 93 recognized within square brackets. Outside square brackets, the metacharacters
136     are as follows:
137 nigel 75 .sp
138     \e general escape character with several uses
139 nigel 63 ^ assert start of string (or line, in multiline mode)
140     $ assert end of string (or line, in multiline mode)
141     . match any character except newline (by default)
142     [ start character class definition
143     | start of alternative branch
144     ( start subpattern
145     ) end subpattern
146     ? extends the meaning of (
147     also 0 or 1 quantifier
148     also quantifier minimizer
149     * 0 or more quantifier
150     + 1 or more quantifier
151     also "possessive quantifier"
152     { start min/max quantifier
153 nigel 75 .sp
154 nigel 63 Part of a pattern that is in square brackets is called a "character class". In
155 nigel 75 a character class the only metacharacters are:
156     .sp
157     \e general escape character
158 nigel 63 ^ negate the class, but only if the first character
159     - indicates character range
160 nigel 75 .\" JOIN
161 nigel 63 [ POSIX character class (only if followed by POSIX
162     syntax)
163     ] terminates the character class
164 nigel 75 .sp
165     The following sections describe the use of each of the metacharacters.
166     .
167 nigel 93 .
168 nigel 63 .SH BACKSLASH
169     .rs
170     .sp
171     The backslash character has several uses. Firstly, if it is followed by a
172 nigel 91 non-alphanumeric character, it takes away any special meaning that character
173     may have. This use of backslash as an escape character applies both inside and
174 nigel 63 outside character classes.
175 nigel 75 .P
176     For example, if you want to match a * character, you write \e* in the pattern.
177 nigel 63 This escaping action applies whether or not the following character would
178 nigel 75 otherwise be interpreted as a metacharacter, so it is always safe to precede a
179     non-alphanumeric with backslash to specify that it stands for itself. In
180     particular, if you want to match a backslash, you write \e\e.
181     .P
182 nigel 63 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
183     pattern (other than in a character class) and characters between a # outside
184 nigel 91 a character class and the next newline are ignored. An escaping backslash can
185     be used to include a whitespace or # character as part of the pattern.
186 nigel 75 .P
187 nigel 63 If you want to remove the special meaning from a sequence of characters, you
188 nigel 75 can do so by putting them between \eQ and \eE. This is different from Perl in
189     that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
190 nigel 63 Perl, $ and @ cause variable interpolation. Note the following examples:
191 nigel 75 .sp
192 nigel 63 Pattern PCRE matches Perl matches
193 nigel 75 .sp
194     .\" JOIN
195     \eQabc$xyz\eE abc$xyz abc followed by the
196 nigel 63 contents of $xyz
197 nigel 75 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
198     \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
199     .sp
200     The \eQ...\eE sequence is recognized both inside and outside character classes.
201     .
202     .
203     .\" HTML <a name="digitsafterbackslash"></a>
204     .SS "Non-printing characters"
205     .rs
206     .sp
207 nigel 63 A second use of backslash provides a way of encoding non-printing characters
208     in patterns in a visible manner. There is no restriction on the appearance of
209     non-printing characters, apart from the binary zero that terminates a pattern,
210 ph10 456 but when a pattern is being prepared by text editing, it is often easier to use
211     one of the following escape sequences than the binary character it represents:
212 nigel 75 .sp
213     \ea alarm, that is, the BEL character (hex 07)
214     \ecx "control-x", where x is any character
215     \ee escape (hex 1B)
216     \ef formfeed (hex 0C)
217 ph10 227 \en linefeed (hex 0A)
218 nigel 75 \er carriage return (hex 0D)
219     \et tab (hex 09)
220     \eddd character with octal code ddd, or backreference
221     \exhh character with hex code hh
222 nigel 87 \ex{hhh..} character with hex code hhh..
223 nigel 75 .sp
224     The precise effect of \ecx is as follows: if x is a lower case letter, it
225 nigel 63 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
226 nigel 75 Thus \ecz becomes hex 1A, but \ec{ becomes hex 3B, while \ec; becomes hex
227 nigel 63 7B.
228 nigel 75 .P
229     After \ex, from zero to two hexadecimal digits are read (letters can be in
230 nigel 87 upper or lower case). Any number of hexadecimal digits may appear between \ex{
231     and }, but the value of the character code must be less than 256 in non-UTF-8
232 ph10 211 mode, and less than 2**31 in UTF-8 mode. That is, the maximum value in
233     hexadecimal is 7FFFFFFF. Note that this is bigger than the largest Unicode code
234     point, which is 10FFFF.
235 nigel 75 .P
236 ph10 211 If characters other than hexadecimal digits appear between \ex{ and }, or if
237     there is no terminating }, this form of escape is not recognized. Instead, the
238     initial \ex will be interpreted as a basic hexadecimal escape, with no
239     following digits, giving a character whose value is zero.
240     .P
241 nigel 63 Characters whose value is less than 256 can be defined by either of the two
242 nigel 87 syntaxes for \ex. There is no difference in the way they are handled. For
243     example, \exdc is exactly the same as \ex{dc}.
244 nigel 75 .P
245 nigel 91 After \e0 up to two further octal digits are read. If there are fewer than two
246     digits, just those that are present are used. Thus the sequence \e0\ex\e07
247     specifies two binary zeros followed by a BEL character (code value 7). Make
248     sure you supply two digits after the initial zero if the pattern character that
249     follows is itself an octal digit.
250 nigel 75 .P
251 nigel 63 The handling of a backslash followed by a digit other than 0 is complicated.
252     Outside a character class, PCRE reads it and any following digits as a decimal
253     number. If the number is less than 10, or if there have been at least that many
254     previous capturing left parentheses in the expression, the entire sequence is
255 nigel 75 taken as a \fIback reference\fP. A description of how this works is given
256     .\" HTML <a href="#backreferences">
257     .\" </a>
258     later,
259     .\"
260     following the discussion of
261     .\" HTML <a href="#subpattern">
262     .\" </a>
263     parenthesized subpatterns.
264     .\"
265     .P
266 nigel 63 Inside a character class, or if the decimal number is greater than 9 and there
267     have not been that many capturing subpatterns, PCRE re-reads up to three octal
268 nigel 93 digits following the backslash, and uses them to generate a data character. Any
269 nigel 91 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
270     character specified in octal must be less than \e400. In UTF-8 mode, values up
271     to \e777 are permitted. For example:
272 nigel 75 .sp
273     \e040 is another way of writing a space
274     .\" JOIN
275     \e40 is the same, provided there are fewer than 40
276 nigel 63 previous capturing subpatterns
277 nigel 75 \e7 is always a back reference
278     .\" JOIN
279     \e11 might be a back reference, or another way of
280 nigel 63 writing a tab
281 nigel 75 \e011 is always a tab
282     \e0113 is a tab followed by the character "3"
283     .\" JOIN
284     \e113 might be a back reference, otherwise the
285 nigel 63 character with octal code 113
286 nigel 75 .\" JOIN
287     \e377 might be a back reference, otherwise
288 nigel 63 the byte consisting entirely of 1 bits
289 nigel 75 .\" JOIN
290     \e81 is either a back reference, or a binary zero
291 nigel 63 followed by the two characters "8" and "1"
292 nigel 75 .sp
293 nigel 63 Note that octal values of 100 or greater must not be introduced by a leading
294     zero, because no more than three octal digits are ever read.
295 nigel 75 .P
296 nigel 91 All the sequences that define a single character value can be used both inside
297     and outside character classes. In addition, inside a character class, the
298     sequence \eb is interpreted as the backspace character (hex 08), and the
299 nigel 93 sequences \eR and \eX are interpreted as the characters "R" and "X",
300     respectively. Outside a character class, these sequences have different
301     meanings
302 nigel 75 .\" HTML <a href="#uniextseq">
303     .\" </a>
304     (see below).
305     .\"
306     .
307     .
308 nigel 93 .SS "Absolute and relative back references"
309     .rs
310     .sp
311 ph10 208 The sequence \eg followed by an unsigned or a negative number, optionally
312     enclosed in braces, is an absolute or relative back reference. A named back
313     reference can be coded as \eg{name}. Back references are discussed
314 nigel 93 .\" HTML <a href="#backreferences">
315     .\" </a>
316     later,
317     .\"
318     following the discussion of
319     .\" HTML <a href="#subpattern">
320     .\" </a>
321     parenthesized subpatterns.
322     .\"
323     .
324     .
325 ph10 333 .SS "Absolute and relative subroutine calls"
326     .rs
327     .sp
328 ph10 345 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
329     a number enclosed either in angle brackets or single quotes, is an alternative
330     syntax for referencing a subpattern as a "subroutine". Details are discussed
331 ph10 333 .\" HTML <a href="#onigurumasubroutines">
332     .\" </a>
333     later.
334     .\"
335 ph10 345 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
336 ph10 454 synonymous. The former is a back reference; the latter is a
337     .\" HTML <a href="#subpatternsassubroutines">
338     .\" </a>
339     subroutine
340     .\"
341     call.
342 ph10 333 .
343     .
344 nigel 75 .SS "Generic character types"
345     .rs
346     .sp
347 nigel 93 Another use of backslash is for specifying generic character types. The
348 nigel 75 following are always recognized:
349     .sp
350 ph10 182 \ed any decimal digit
351 nigel 75 \eD any character that is not a decimal digit
352 ph10 178 \eh any horizontal whitespace character
353 ph10 182 \eH any character that is not a horizontal whitespace character
354 nigel 75 \es any whitespace character
355     \eS any character that is not a whitespace character
356 ph10 178 \ev any vertical whitespace character
357 ph10 182 \eV any character that is not a vertical whitespace character
358 nigel 75 \ew any "word" character
359     \eW any "non-word" character
360     .sp
361 nigel 63 Each pair of escape sequences partitions the complete set of characters into
362     two disjoint sets. Any given character matches one, and only one, of each pair.
363 nigel 75 .P
364     These character type sequences can appear both inside and outside character
365     classes. They each match one character of the appropriate type. If the current
366     matching point is at the end of the subject string, all of them fail, since
367     there is no character to match.
368     .P
369     For compatibility with Perl, \es does not match the VT character (code 11).
370     This makes it different from the the POSIX "space" class. The \es characters
371 ph10 178 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
372 nigel 91 included in a Perl script, \es may match the VT character. In PCRE, it never
373 ph10 178 does.
374 nigel 75 .P
375 ph10 178 In UTF-8 mode, characters with values greater than 128 never match \ed, \es, or
376     \ew, and always match \eD, \eS, and \eW. This is true even when Unicode
377 ph10 182 character property support is available. These sequences retain their original
378     meanings from before UTF-8 support was available, mainly for efficiency
379 ph10 406 reasons. Note that this also affects \eb, because it is defined in terms of \ew
380 ph10 394 and \eW.
381 ph10 178 .P
382 ph10 182 The sequences \eh, \eH, \ev, and \eV are Perl 5.10 features. In contrast to the
383 ph10 178 other sequences, these do match certain high-valued codepoints in UTF-8 mode.
384     The horizontal space characters are:
385     .sp
386     U+0009 Horizontal tab
387     U+0020 Space
388     U+00A0 Non-break space
389     U+1680 Ogham space mark
390     U+180E Mongolian vowel separator
391     U+2000 En quad
392     U+2001 Em quad
393     U+2002 En space
394     U+2003 Em space
395     U+2004 Three-per-em space
396     U+2005 Four-per-em space
397     U+2006 Six-per-em space
398     U+2007 Figure space
399     U+2008 Punctuation space
400     U+2009 Thin space
401     U+200A Hair space
402     U+202F Narrow no-break space
403     U+205F Medium mathematical space
404     U+3000 Ideographic space
405     .sp
406     The vertical space characters are:
407     .sp
408     U+000A Linefeed
409     U+000B Vertical tab
410     U+000C Formfeed
411     U+000D Carriage return
412     U+0085 Next line
413     U+2028 Line separator
414     U+2029 Paragraph separator
415     .P
416 nigel 75 A "word" character is an underscore or any character less than 256 that is a
417     letter or digit. The definition of letters and digits is controlled by PCRE's
418     low-valued character tables, and may vary if locale-specific matching is taking
419     place (see
420 nigel 63 .\" HTML <a href="pcreapi.html#localesupport">
421     .\" </a>
422     "Locale support"
423     .\"
424     in the
425     .\" HREF
426 nigel 75 \fBpcreapi\fP
427 nigel 63 .\"
428 ph10 139 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
429     or "french" in Windows, some character codes greater than 128 are used for
430 ph10 178 accented letters, and these are matched by \ew. The use of locales with Unicode
431     is discouraged.
432 nigel 75 .
433     .
434 ph10 231 .\" HTML <a name="newlineseq"></a>
435 nigel 93 .SS "Newline sequences"
436     .rs
437     .sp
438 ph10 231 Outside a character class, by default, the escape sequence \eR matches any
439     Unicode newline sequence. This is a Perl 5.10 feature. In non-UTF-8 mode \eR is
440     equivalent to the following:
441 nigel 93 .sp
442     (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
443     .sp
444     This is an example of an "atomic group", details of which are given
445     .\" HTML <a href="#atomicgroup">
446     .\" </a>
447     below.
448     .\"
449     This particular group matches either the two-character sequence CR followed by
450     LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
451     U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
452     line, U+0085). The two-character sequence is treated as a single unit that
453     cannot be split.
454     .P
455     In UTF-8 mode, two additional characters whose codepoints are greater than 255
456     are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
457     Unicode character property support is not needed for these characters to be
458     recognized.
459     .P
460 ph10 231 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
461     complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
462 ph10 247 either at compile time or when the pattern is matched. (BSR is an abbrevation
463 ph10 246 for "backslash R".) This can be made the default when PCRE is built; if this is
464     the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
465     It is also possible to specify these settings by starting a pattern string with
466     one of the following sequences:
467 ph10 231 .sp
468     (*BSR_ANYCRLF) CR, LF, or CRLF only
469     (*BSR_UNICODE) any Unicode newline sequence
470     .sp
471 ph10 456 These override the default and the options given to \fBpcre_compile()\fP or
472     \fBpcre_compile2()\fP, but they can be overridden by options given to
473     \fBpcre_exec()\fP or \fBpcre_dfa_exec()\fP. Note that these special settings,
474     which are not Perl-compatible, are recognized only at the very start of a
475     pattern, and that they must be in upper case. If more than one of them is
476     present, the last one is used. They can be combined with a change of newline
477     convention, for example, a pattern can start with:
478 ph10 246 .sp
479     (*ANY)(*BSR_ANYCRLF)
480     .sp
481 nigel 93 Inside a character class, \eR matches the letter "R".
482     .
483     .
484 nigel 75 .\" HTML <a name="uniextseq"></a>
485     .SS Unicode character properties
486     .rs
487     .sp
488     When PCRE is built with Unicode character property support, three additional
489 ph10 184 escape sequences that match characters with specific properties are available.
490     When not in UTF-8 mode, these sequences are of course limited to testing
491     characters whose codepoints are less than 256, but they do work in this mode.
492     The extra escape sequences are:
493 nigel 75 .sp
494 nigel 87 \ep{\fIxx\fP} a character with the \fIxx\fP property
495     \eP{\fIxx\fP} a character without the \fIxx\fP property
496     \eX an extended Unicode sequence
497 nigel 75 .sp
498 nigel 87 The property names represented by \fIxx\fP above are limited to the Unicode
499     script names, the general category properties, and "Any", which matches any
500     character (including newline). Other properties such as "InMusicalSymbols" are
501     not currently supported by PCRE. Note that \eP{Any} does not match any
502     characters, so always causes a match failure.
503 nigel 75 .P
504 nigel 87 Sets of Unicode characters are defined as belonging to certain scripts. A
505     character from one of these sets can be matched using a script name. For
506     example:
507 nigel 75 .sp
508 nigel 87 \ep{Greek}
509     \eP{Han}
510     .sp
511     Those that are not part of an identified script are lumped together as
512     "Common". The current list of scripts is:
513     .P
514     Arabic,
515     Armenian,
516 nigel 93 Balinese,
517 nigel 87 Bengali,
518     Bopomofo,
519     Braille,
520     Buginese,
521     Buhid,
522     Canadian_Aboriginal,
523     Cherokee,
524     Common,
525     Coptic,
526 nigel 93 Cuneiform,
527 nigel 87 Cypriot,
528     Cyrillic,
529     Deseret,
530     Devanagari,
531     Ethiopic,
532     Georgian,
533     Glagolitic,
534     Gothic,
535     Greek,
536     Gujarati,
537     Gurmukhi,
538     Han,
539     Hangul,
540     Hanunoo,
541     Hebrew,
542     Hiragana,
543     Inherited,
544     Kannada,
545     Katakana,
546     Kharoshthi,
547     Khmer,
548     Lao,
549     Latin,
550     Limbu,
551     Linear_B,
552     Malayalam,
553     Mongolian,
554     Myanmar,
555     New_Tai_Lue,
556 nigel 93 Nko,
557 nigel 87 Ogham,
558     Old_Italic,
559     Old_Persian,
560     Oriya,
561     Osmanya,
562 nigel 93 Phags_Pa,
563     Phoenician,
564 nigel 87 Runic,
565     Shavian,
566     Sinhala,
567     Syloti_Nagri,
568     Syriac,
569     Tagalog,
570     Tagbanwa,
571     Tai_Le,
572     Tamil,
573     Telugu,
574     Thaana,
575     Thai,
576     Tibetan,
577     Tifinagh,
578     Ugaritic,
579     Yi.
580     .P
581     Each character has exactly one general category property, specified by a
582     two-letter abbreviation. For compatibility with Perl, negation can be specified
583     by including a circumflex between the opening brace and the property name. For
584     example, \ep{^Lu} is the same as \eP{Lu}.
585     .P
586     If only one letter is specified with \ep or \eP, it includes all the general
587     category properties that start with that letter. In this case, in the absence
588     of negation, the curly brackets in the escape sequence are optional; these two
589     examples have the same effect:
590     .sp
591 nigel 75 \ep{L}
592     \epL
593     .sp
594 nigel 87 The following general category property codes are supported:
595 nigel 75 .sp
596     C Other
597     Cc Control
598     Cf Format
599     Cn Unassigned
600     Co Private use
601     Cs Surrogate
602     .sp
603     L Letter
604     Ll Lower case letter
605     Lm Modifier letter
606     Lo Other letter
607     Lt Title case letter
608     Lu Upper case letter
609     .sp
610     M Mark
611     Mc Spacing mark
612     Me Enclosing mark
613     Mn Non-spacing mark
614     .sp
615     N Number
616     Nd Decimal number
617     Nl Letter number
618     No Other number
619     .sp
620     P Punctuation
621     Pc Connector punctuation
622     Pd Dash punctuation
623     Pe Close punctuation
624     Pf Final punctuation
625     Pi Initial punctuation
626     Po Other punctuation
627     Ps Open punctuation
628     .sp
629     S Symbol
630     Sc Currency symbol
631     Sk Modifier symbol
632     Sm Mathematical symbol
633     So Other symbol
634     .sp
635     Z Separator
636     Zl Line separator
637     Zp Paragraph separator
638     Zs Space separator
639     .sp
640 nigel 87 The special property L& is also supported: it matches a character that has
641     the Lu, Ll, or Lt property, in other words, a letter that is not classified as
642     a modifier or "other".
643 nigel 75 .P
644 ph10 211 The Cs (Surrogate) property applies only to characters in the range U+D800 to
645     U+DFFF. Such characters are not valid in UTF-8 strings (see RFC 3629) and so
646     cannot be tested by PCRE, unless UTF-8 validity checking has been turned off
647     (see the discussion of PCRE_NO_UTF8_CHECK in the
648     .\" HREF
649     \fBpcreapi\fP
650     .\"
651 ph10 451 page). Perl does not support the Cs property.
652 ph10 211 .P
653 ph10 451 The long synonyms for property names that Perl supports (such as \ep{Letter})
654 nigel 91 are not supported by PCRE, nor is it permitted to prefix any of these
655 nigel 87 properties with "Is".
656     .P
657     No character that is in the Unicode table has the Cn (unassigned) property.
658     Instead, this property is assumed for any code point that is not in the
659     Unicode table.
660     .P
661 nigel 75 Specifying caseless matching does not affect these escape sequences. For
662     example, \ep{Lu} always matches only upper case letters.
663     .P
664     The \eX escape matches any number of Unicode characters that form an extended
665     Unicode sequence. \eX is equivalent to
666     .sp
667     (?>\ePM\epM*)
668     .sp
669     That is, it matches a character without the "mark" property, followed by zero
670     or more characters with the "mark" property, and treats the sequence as an
671     atomic group
672     .\" HTML <a href="#atomicgroup">
673     .\" </a>
674     (see below).
675     .\"
676     Characters with the "mark" property are typically accents that affect the
677 ph10 185 preceding character. None of them have codepoints less than 256, so in
678 ph10 184 non-UTF-8 mode \eX matches any one character.
679 nigel 75 .P
680     Matching characters by Unicode property is not fast, because PCRE has to search
681     a structure that contains data for over fifteen thousand characters. That is
682     why the traditional escape sequences such as \ed and \ew do not use Unicode
683     properties in PCRE.
684     .
685     .
686 ph10 168 .\" HTML <a name="resetmatchstart"></a>
687     .SS "Resetting the match start"
688     .rs
689     .sp
690 ph10 172 The escape sequence \eK, which is a Perl 5.10 feature, causes any previously
691     matched characters not to be included in the final matched sequence. For
692 ph10 168 example, the pattern:
693     .sp
694     foo\eKbar
695     .sp
696 ph10 172 matches "foobar", but reports that it has matched "bar". This feature is
697 ph10 168 similar to a lookbehind assertion
698     .\" HTML <a href="#lookbehind">
699     .\" </a>
700     (described below).
701     .\"
702 ph10 172 However, in this case, the part of the subject before the real match does not
703     have to be of fixed length, as lookbehind assertions do. The use of \eK does
704 ph10 168 not interfere with the setting of
705     .\" HTML <a href="#subpattern">
706     .\" </a>
707     captured substrings.
708 ph10 172 .\"
709 ph10 168 For example, when the pattern
710     .sp
711     (foo)\eKbar
712     .sp
713 ph10 172 matches "foobar", the first substring is still set to "foo".
714 ph10 168 .
715     .
716 nigel 75 .\" HTML <a name="smallassertions"></a>
717     .SS "Simple assertions"
718     .rs
719     .sp
720 nigel 93 The final use of backslash is for certain simple assertions. An assertion
721 nigel 63 specifies a condition that has to be met at a particular point in a match,
722     without consuming any characters from the subject string. The use of
723 nigel 75 subpatterns for more complicated assertions is described
724     .\" HTML <a href="#bigassertions">
725     .\" </a>
726     below.
727     .\"
728 nigel 91 The backslashed assertions are:
729 nigel 75 .sp
730     \eb matches at a word boundary
731     \eB matches when not at a word boundary
732 nigel 93 \eA matches at the start of the subject
733     \eZ matches at the end of the subject
734     also matches before a newline at the end of the subject
735     \ez matches only at the end of the subject
736     \eG matches at the first matching position in the subject
737 nigel 75 .sp
738     These assertions may not appear in character classes (but note that \eb has a
739 nigel 63 different meaning, namely the backspace character, inside a character class).
740 nigel 75 .P
741 nigel 63 A word boundary is a position in the subject string where the current character
742 nigel 75 and the previous character do not both match \ew or \eW (i.e. one matches
743     \ew and the other matches \eW), or the start or end of the string if the
744 ph10 456 first or last character matches \ew, respectively. Neither PCRE nor Perl has a
745     separte "start of word" or "end of word" metasequence. However, whatever
746     follows \eb normally determines which it is. For example, the fragment
747     \eba matches "a" at the start of a word.
748 nigel 75 .P
749     The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
750     dollar (described in the next section) in that they only ever match at the very
751     start and end of the subject string, whatever options are set. Thus, they are
752     independent of multiline mode. These three assertions are not affected by the
753     PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
754     circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
755     argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
756     at a point other than the beginning of the subject, \eA can never match. The
757 nigel 91 difference between \eZ and \ez is that \eZ matches before a newline at the end
758     of the string as well as at the very end, whereas \ez matches only at the end.
759 nigel 75 .P
760     The \eG assertion is true only when the current matching position is at the
761     start point of the match, as specified by the \fIstartoffset\fP argument of
762     \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
763     non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
764 nigel 63 arguments, you can mimic Perl's /g option, and it is in this kind of
765 nigel 75 implementation where \eG can be useful.
766     .P
767     Note, however, that PCRE's interpretation of \eG, as the start of the current
768 nigel 63 match, is subtly different from Perl's, which defines it as the end of the
769     previous match. In Perl, these can be different when the previously matched
770     string was empty. Because PCRE does just one match at a time, it cannot
771     reproduce this behaviour.
772 nigel 75 .P
773     If all the alternatives of a pattern begin with \eG, the expression is anchored
774 nigel 63 to the starting match position, and the "anchored" flag is set in the compiled
775     regular expression.
776 nigel 75 .
777     .
779 nigel 63 .rs
780     .sp
781     Outside a character class, in the default matching mode, the circumflex
782 nigel 75 character is an assertion that is true only if the current matching point is
783     at the start of the subject string. If the \fIstartoffset\fP argument of
784     \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
785 nigel 63 option is unset. Inside a character class, circumflex has an entirely different
786 nigel 75 meaning
787     .\" HTML <a href="#characterclass">
788     .\" </a>
789     (see below).
790     .\"
791     .P
792 nigel 63 Circumflex need not be the first character of the pattern if a number of
793     alternatives are involved, but it should be the first thing in each alternative
794     in which it appears if the pattern is ever to match that branch. If all
795     possible alternatives start with a circumflex, that is, if the pattern is
796     constrained to match only at the start of the subject, it is said to be an
797     "anchored" pattern. (There are also other constructs that can cause a pattern
798     to be anchored.)
799 nigel 75 .P
800     A dollar character is an assertion that is true only if the current matching
801 nigel 63 point is at the end of the subject string, or immediately before a newline
802 nigel 91 at the end of the string (by default). Dollar need not be the last character of
803     the pattern if a number of alternatives are involved, but it should be the last
804     item in any branch in which it appears. Dollar has no special meaning in a
805     character class.
806 nigel 75 .P
807 nigel 63 The meaning of dollar can be changed so that it matches only at the very end of
808     the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
809 nigel 75 does not affect the \eZ assertion.
810     .P
811 nigel 63 The meanings of the circumflex and dollar characters are changed if the
812 nigel 91 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
813     immediately after internal newlines as well as at the start of the subject
814     string. It does not match after a newline that ends the string. A dollar
815     matches before any newlines in the string, as well as at the very end, when
816     PCRE_MULTILINE is set. When newline is specified as the two-character
817     sequence CRLF, isolated CR and LF characters do not indicate newlines.
818 nigel 75 .P
819 nigel 91 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
820     \en represents a newline) in multiline mode, but not otherwise. Consequently,
821     patterns that are anchored in single line mode because all branches start with
822     ^ are not anchored in multiline mode, and a match for circumflex is possible
823     when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
824     PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
825     .P
826 nigel 75 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
827 nigel 63 end of the subject in both modes, and if all branches of a pattern start with
828 nigel 91 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
829 nigel 75 .
830     .
832 nigel 63 .rs
833     .sp
834     Outside a character class, a dot in the pattern matches any one character in
835 nigel 91 the subject string except (by default) a character that signifies the end of a
836 nigel 93 line. In UTF-8 mode, the matched character may be more than one byte long.
837 nigel 91 .P
838 nigel 93 When a line ending is defined as a single character, dot never matches that
839     character; when the two-character sequence CRLF is used, dot does not match CR
840     if it is immediately followed by LF, but otherwise it matches all characters
841     (including isolated CRs and LFs). When any Unicode line endings are being
842     recognized, dot does not match CR or LF or any of the other line ending
843     characters.
844     .P
845 nigel 91 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
846 nigel 93 option is set, a dot matches any one character, without exception. If the
847     two-character sequence CRLF is present in the subject string, it takes two dots
848     to match it.
849 nigel 91 .P
850     The handling of dot is entirely independent of the handling of circumflex and
851     dollar, the only relationship being that they both involve newlines. Dot has no
852     special meaning in a character class.
853 nigel 75 .
854     .
856 nigel 63 .rs
857     .sp
858 nigel 75 Outside a character class, the escape sequence \eC matches any one byte, both
859 nigel 93 in and out of UTF-8 mode. Unlike a dot, it always matches any line-ending
860     characters. The feature is provided in Perl in order to match individual bytes
861     in UTF-8 mode. Because it breaks up UTF-8 characters into individual bytes,
862     what remains in the string may be a malformed UTF-8 string. For this reason,
863     the \eC escape sequence is best avoided.
864 nigel 75 .P
865     PCRE does not allow \eC to appear in lookbehind assertions
866     .\" HTML <a href="#lookbehind">
867     .\" </a>
868     (described below),
869     .\"
870     because in UTF-8 mode this would make it impossible to calculate the length of
871     the lookbehind.
872     .
873     .
874     .\" HTML <a name="characterclass"></a>
876 nigel 63 .rs
877     .sp
878     An opening square bracket introduces a character class, terminated by a closing
879 ph10 456 square bracket. A closing square bracket on its own is not special by default.
880     However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
881     bracket causes a compile-time error. If a closing square bracket is required as
882     a member of the class, it should be the first data character in the class
883     (after an initial circumflex, if present) or escaped with a backslash.
884 nigel 75 .P
885 nigel 63 A character class matches a single character in the subject. In UTF-8 mode, the
886 ph10 456 character may be more than one byte long. A matched character must be in the
887     set of characters defined by the class, unless the first character in the class
888 nigel 63 definition is a circumflex, in which case the subject character must not be in
889     the set defined by the class. If a circumflex is actually required as a member
890     of the class, ensure it is not the first character, or escape it with a
891     backslash.
892 nigel 75 .P
893 nigel 63 For example, the character class [aeiou] matches any lower case vowel, while
894     [^aeiou] matches any character that is not a lower case vowel. Note that a
895 nigel 75 circumflex is just a convenient notation for specifying the characters that
896     are in the class by enumerating those that are not. A class that starts with a
897 ph10 456 circumflex is not an assertion; it still consumes a character from the subject
898 nigel 75 string, and therefore it fails if the current pointer is at the end of the
899     string.
900     .P
901 nigel 63 In UTF-8 mode, characters with values greater than 255 can be included in a
902 nigel 75 class as a literal string of bytes, or by using the \ex{ escaping mechanism.
903     .P
904 nigel 63 When caseless matching is set, any letters in a class represent both their
905     upper case and lower case versions, so for example, a caseless [aeiou] matches
906     "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
907 nigel 77 caseful version would. In UTF-8 mode, PCRE always understands the concept of
908     case for characters whose values are less than 128, so caseless matching is
909     always possible. For characters with higher values, the concept of case is
910     supported if PCRE is compiled with Unicode property support, but not otherwise.
911 ph10 456 If you want to use caseless matching in UTF8-mode for characters 128 and above,
912     you must ensure that PCRE is compiled with Unicode property support as well as
913     with UTF-8 support.
914 nigel 75 .P
915 nigel 93 Characters that might indicate line breaks are never treated in any special way
916     when matching character classes, whatever line-ending sequence is in use, and
917     whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
918     such as [^a] always matches one of these characters.
919 nigel 75 .P
920 nigel 63 The minus (hyphen) character can be used to specify a range of characters in a
921     character class. For example, [d-m] matches any letter between d and m,
922     inclusive. If a minus character is required in a class, it must be escaped with
923     a backslash or appear in a position where it cannot be interpreted as
924     indicating a range, typically as the first or last character in the class.
925 nigel 75 .P
926 nigel 63 It is not possible to have the literal character "]" as the end character of a
927     range. A pattern such as [W-]46] is interpreted as a class of two characters
928     ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
929     "-46]". However, if the "]" is escaped with a backslash it is interpreted as
930 nigel 75 the end of range, so [W-\e]46] is interpreted as a class containing a range
931     followed by two other characters. The octal or hexadecimal representation of
932     "]" can also be used to end a range.
933     .P
934 nigel 63 Ranges operate in the collating sequence of character values. They can also be
935 nigel 75 used for characters specified numerically, for example [\e000-\e037]. In UTF-8
936 nigel 63 mode, ranges can include characters whose values are greater than 255, for
937 nigel 75 example [\ex{100}-\ex{2ff}].
938     .P
939 nigel 63 If a range that includes letters is used when caseless matching is set, it
940     matches the letters in either case. For example, [W-c] is equivalent to
941 nigel 75 [][\e\e^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
942 ph10 139 tables for a French locale are in use, [\exc8-\excb] matches accented E
943 nigel 75 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
944     characters with values greater than 128 only when it is compiled with Unicode
945     property support.
946     .P
947     The character types \ed, \eD, \ep, \eP, \es, \eS, \ew, and \eW may also appear
948     in a character class, and add the characters that they match to the class. For
949     example, [\edABCDEF] matches any hexadecimal digit. A circumflex can
950 nigel 63 conveniently be used with the upper case character types to specify a more
951     restricted set of characters than the matching lower case type. For example,
952 nigel 75 the class [^\eW_] matches any letter or digit, but not underscore.
953     .P
954     The only metacharacters that are recognized in character classes are backslash,
955     hyphen (only where it can be interpreted as specifying a range), circumflex
956     (only at the start), opening square bracket (only when it can be interpreted as
957     introducing a POSIX class name - see the next section), and the terminating
958     closing square bracket. However, escaping other non-alphanumeric characters
959     does no harm.
960     .
961     .
963 nigel 63 .rs
964     .sp
965 nigel 75 Perl supports the POSIX notation for character classes. This uses names
966 nigel 63 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
967     this notation. For example,
968 nigel 75 .sp
969 nigel 63 [01[:alpha:]%]
970 nigel 75 .sp
971 nigel 63 matches "0", "1", any alphabetic character, or "%". The supported class names
972     are
973 nigel 75 .sp
974 nigel 63 alnum letters and digits
975     alpha letters
976     ascii character codes 0 - 127
977     blank space or tab only
978     cntrl control characters
979 nigel 75 digit decimal digits (same as \ed)
980 nigel 63 graph printing characters, excluding space
981     lower lower case letters
982     print printing characters, including space
983     punct printing characters, excluding letters and digits
984 nigel 75 space white space (not quite the same as \es)
985 nigel 63 upper upper case letters
986 nigel 75 word "word" characters (same as \ew)
987 nigel 63 xdigit hexadecimal digits
988 nigel 75 .sp
989 nigel 63 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
990     space (32). Notice that this list includes the VT character (code 11). This
991 nigel 75 makes "space" different to \es, which does not include VT (for Perl
992 nigel 63 compatibility).
993 nigel 75 .P
994 nigel 63 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
995     5.8. Another Perl extension is negation, which is indicated by a ^ character
996     after the colon. For example,
997 nigel 75 .sp
998 nigel 63 [12[:^digit:]]
999 nigel 75 .sp
1000 nigel 63 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1001     syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1002     supported, and an error is given if they are encountered.
1003 nigel 75 .P
1004     In UTF-8 mode, characters with values greater than 128 do not match any of
1005 nigel 63 the POSIX character classes.
1006 nigel 75 .
1007     .
1008     .SH "VERTICAL BAR"
1009 nigel 63 .rs
1010     .sp
1011     Vertical bar characters are used to separate alternative patterns. For example,
1012     the pattern
1013 nigel 75 .sp
1014 nigel 63 gilbert|sullivan
1015 nigel 75 .sp
1016 nigel 63 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1017 nigel 91 and an empty alternative is permitted (matching the empty string). The matching
1018     process tries each alternative in turn, from left to right, and the first one
1019     that succeeds is used. If the alternatives are within a subpattern
1020 nigel 75 .\" HTML <a href="#subpattern">
1021     .\" </a>
1022     (defined below),
1023     .\"
1024     "succeeds" means matching the rest of the main pattern as well as the
1025     alternative in the subpattern.
1026     .
1027     .
1029 nigel 63 .rs
1030     .sp
1031     The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1032 ph10 231 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1033     the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1034     The option letters are
1035 nigel 75 .sp
1036 nigel 63 i for PCRE_CASELESS
1037     m for PCRE_MULTILINE
1038     s for PCRE_DOTALL
1039     x for PCRE_EXTENDED
1040 nigel 75 .sp
1041 nigel 63 For example, (?im) sets caseless, multiline matching. It is also possible to
1042     unset these options by preceding the letter with a hyphen, and a combined
1043     setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1044     PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1045     permitted. If a letter appears both before and after the hyphen, the option is
1046     unset.
1047 nigel 75 .P
1048 ph10 231 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1049     changed in the same way as the Perl-compatible options by using the characters
1050     J, U and X respectively.
1051     .P
1052 ph10 412 When one of these option changes occurs at top level (that is, not inside
1053     subpattern parentheses), the change applies to the remainder of the pattern
1054     that follows. If the change is placed right at the start of a pattern, PCRE
1055     extracts it into the global options (and it will therefore show up in data
1056     extracted by the \fBpcre_fullinfo()\fP function).
1057 nigel 75 .P
1058 nigel 93 An option change within a subpattern (see below for a description of
1059     subpatterns) affects only that part of the current pattern that follows it, so
1060 nigel 75 .sp
1061 nigel 63 (a(?i)b)c
1062 nigel 75 .sp
1063 nigel 63 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1064     By this means, options can be made to have different settings in different
1065     parts of the pattern. Any changes made in one alternative do carry on
1066     into subsequent branches within the same subpattern. For example,
1067 nigel 75 .sp
1068 nigel 63 (a(?i)b|c)
1069 nigel 75 .sp
1070 nigel 63 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1071     branch is abandoned before the option setting. This is because the effects of
1072     option settings happen at compile time. There would be some very weird
1073     behaviour otherwise.
1074 ph10 251 .P
1075     \fBNote:\fP There are other PCRE-specific options that can be set by the
1076     application when the compile or match functions are called. In some cases the
1077 ph10 412 pattern can contain special leading sequences such as (*CRLF) to override what
1078     the application has set or what has been defaulted. Details are given in the
1079     section entitled
1080 ph10 251 .\" HTML <a href="#newlineseq">
1081     .\" </a>
1082     "Newline sequences"
1083     .\"
1084 ph10 416 above. There is also the (*UTF8) leading sequence that can be used to set UTF-8
1085 ph10 412 mode; this is equivalent to setting the PCRE_UTF8 option.
1086 nigel 75 .
1087     .
1088     .\" HTML <a name="subpattern"></a>
1089 nigel 63 .SH SUBPATTERNS
1090     .rs
1091     .sp
1092     Subpatterns are delimited by parentheses (round brackets), which can be nested.
1093 nigel 75 Turning part of a pattern into a subpattern does two things:
1094     .sp
1095 nigel 63 1. It localizes a set of alternatives. For example, the pattern
1096 nigel 75 .sp
1097 nigel 63 cat(aract|erpillar|)
1098 nigel 75 .sp
1099 nigel 63 matches one of the words "cat", "cataract", or "caterpillar". Without the
1100 nigel 93 parentheses, it would match "cataract", "erpillar" or an empty string.
1101 nigel 75 .sp
1102     2. It sets up the subpattern as a capturing subpattern. This means that, when
1103     the whole pattern matches, that portion of the subject string that matched the
1104     subpattern is passed back to the caller via the \fIovector\fP argument of
1105     \fBpcre_exec()\fP. Opening parentheses are counted from left to right (starting
1106     from 1) to obtain numbers for the capturing subpatterns.
1107     .P
1108 nigel 63 For example, if the string "the red king" is matched against the pattern
1109 nigel 75 .sp
1110 nigel 63 the ((red|white) (king|queen))
1111 nigel 75 .sp
1112 nigel 63 the captured substrings are "red king", "red", and "king", and are numbered 1,
1113     2, and 3, respectively.
1114 nigel 75 .P
1115 nigel 63 The fact that plain parentheses fulfil two functions is not always helpful.
1116     There are often times when a grouping subpattern is required without a
1117     capturing requirement. If an opening parenthesis is followed by a question mark
1118     and a colon, the subpattern does not do any capturing, and is not counted when
1119     computing the number of any subsequent capturing subpatterns. For example, if
1120     the string "the white queen" is matched against the pattern
1121 nigel 75 .sp
1122 nigel 63 the ((?:red|white) (king|queen))
1123 nigel 75 .sp
1124 nigel 63 the captured substrings are "white queen" and "queen", and are numbered 1 and
1125 nigel 93 2. The maximum number of capturing subpatterns is 65535.
1126 nigel 75 .P
1127 nigel 63 As a convenient shorthand, if any option settings are required at the start of
1128     a non-capturing subpattern, the option letters may appear between the "?" and
1129     the ":". Thus the two patterns
1130 nigel 75 .sp
1131 nigel 63 (?i:saturday|sunday)
1132     (?:(?i)saturday|sunday)
1133 nigel 75 .sp
1134 nigel 63 match exactly the same set of strings. Because alternative branches are tried
1135     from left to right, and options are not reset until the end of the subpattern
1136     is reached, an option setting in one branch does affect subsequent branches, so
1137     the above patterns match "SUNDAY" as well as "Saturday".
1138 nigel 75 .
1139     .
1140 ph10 456 .\" HTML <a name="dupsubpatternnumber"></a>
1142     .rs
1143     .sp
1144 ph10 182 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1145     the same numbers for its capturing parentheses. Such a subpattern starts with
1146     (?| and is itself a non-capturing subpattern. For example, consider this
1147 ph10 175 pattern:
1148     .sp
1149     (?|(Sat)ur|(Sun))day
1150 ph10 182 .sp
1151     Because the two alternatives are inside a (?| group, both sets of capturing
1152     parentheses are numbered one. Thus, when the pattern matches, you can look
1153     at captured substring number one, whichever alternative matched. This construct
1154     is useful when you want to capture part, but not all, of one of a number of
1155     alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1156 ph10 175 number is reset at the start of each branch. The numbers of any capturing
1157 ph10 182 buffers that follow the subpattern start after the highest number used in any
1158     branch. The following example is taken from the Perl documentation.
1159 ph10 175 The numbers underneath show in which buffer the captured content will be
1160     stored.
1161     .sp
1162     # before ---------------branch-reset----------- after
1163     / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1164     # 1 2 2 3 2 3 4
1165 ph10 182 .sp
1166 ph10 456 A backreference to a numbered subpattern uses the most recent value that is set
1167     for that number by any subpattern. The following pattern matches "abcabc" or
1168     "defdef":
1169     .sp
1170     /(?|(abc)|(def))\1/
1171     .sp
1172     In contrast, a recursive or "subroutine" call to a numbered subpattern always
1173     refers to the first one in the pattern with the given number. The following
1174     pattern matches "abcabc" or "defabc":
1175     .sp
1176     /(?|(abc)|(def))(?1)/
1177     .sp
1178     An alternative approach to using the "branch reset" feature is to use
1179 ph10 175 duplicate named subpatterns, as described in the next section.
1180     .
1181     .
1182 nigel 75 .SH "NAMED SUBPATTERNS"
1183 nigel 63 .rs
1184     .sp
1185     Identifying capturing parentheses by number is simple, but it can be very hard
1186     to keep track of the numbers in complicated regular expressions. Furthermore,
1187 nigel 75 if an expression is modified, the numbers may change. To help with this
1188 nigel 93 difficulty, PCRE supports the naming of subpatterns. This feature was not
1189     added to Perl until release 5.10. Python had the feature earlier, and PCRE
1190     introduced it at release 4.0, using the Python syntax. PCRE now supports both
1191     the Perl and the Python syntax.
1192     .P
1193     In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1194     (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1195 nigel 91 parentheses from other parts of the pattern, such as
1196     .\" HTML <a href="#backreferences">
1197     .\" </a>
1198     backreferences,
1199     .\"
1200     .\" HTML <a href="#recursion">
1201     .\" </a>
1202     recursion,
1203     .\"
1204     and
1205     .\" HTML <a href="#conditions">
1206     .\" </a>
1207     conditions,
1208     .\"
1209     can be made by name as well as by number.
1210 nigel 75 .P
1211 nigel 91 Names consist of up to 32 alphanumeric characters and underscores. Named
1212 nigel 93 capturing parentheses are still allocated numbers as well as names, exactly as
1213     if the names were not present. The PCRE API provides function calls for
1214     extracting the name-to-number translation table from a compiled pattern. There
1215     is also a convenience function for extracting a captured substring by name.
1216 nigel 91 .P
1217     By default, a name must be unique within a pattern, but it is possible to relax
1218 ph10 457 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1219     names are also always permitted for subpatterns with the same number, set up as
1220     described in the previous section.) Duplicate names can be useful for patterns
1221     where only one instance of the named parentheses can match. Suppose you want to
1222     match the name of a weekday, either as a 3-letter abbreviation or as the full
1223     name, and in both cases you want to extract the abbreviation. This pattern
1224     (ignoring the line breaks) does the job:
1225 nigel 91 .sp
1226 nigel 93 (?<DN>Mon|Fri|Sun)(?:day)?|
1227     (?<DN>Tue)(?:sday)?|
1228     (?<DN>Wed)(?:nesday)?|
1229     (?<DN>Thu)(?:rsday)?|
1230     (?<DN>Sat)(?:urday)?
1231 nigel 91 .sp
1232     There are five capturing substrings, but only one is ever set after a match.
1233 ph10 182 (An alternative way of solving this problem is to use a "branch reset"
1234 ph10 175 subpattern, as described in the previous section.)
1235     .P
1236 nigel 91 The convenience function for extracting the data by name returns the substring
1237 nigel 93 for the first (and in this example, the only) subpattern of that name that
1238 nigel 91 matched. This saves searching to find which numbered subpattern it was. If you
1239     make a reference to a non-unique named subpattern from elsewhere in the
1240     pattern, the one that corresponds to the lowest number is used. For further
1241     details of the interfaces for handling named subpatterns, see the
1242 nigel 63 .\" HREF
1243 nigel 75 \fBpcreapi\fP
1244 nigel 63 .\"
1245     documentation.
1246 ph10 385 .P
1247     \fBWarning:\fP You cannot use different names to distinguish between two
1248 ph10 457 subpatterns with the same number because PCRE uses only the numbers when
1249     matching. For this reason, an error is given at compile time if different names
1250     are given to subpatterns with the same number. However, you can give the same
1251     name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
1252 nigel 75 .
1253     .
1254 nigel 63 .SH REPETITION
1255     .rs
1256     .sp
1257     Repetition is specified by quantifiers, which can follow any of the following
1258     items:
1259 nigel 75 .sp
1260 nigel 63 a literal data character
1261 nigel 93 the dot metacharacter
1262 nigel 75 the \eC escape sequence
1263     the \eX escape sequence (in UTF-8 mode with Unicode properties)
1264 nigel 93 the \eR escape sequence
1265 nigel 75 an escape such as \ed that matches a single character
1266 nigel 63 a character class
1267     a back reference (see next section)
1268     a parenthesized subpattern (unless it is an assertion)
1269 ph10 456 a recursive or "subroutine" call to a subpattern
1270 nigel 75 .sp
1271 nigel 63 The general repetition quantifier specifies a minimum and maximum number of
1272     permitted matches, by giving the two numbers in curly brackets (braces),
1273     separated by a comma. The numbers must be less than 65536, and the first must
1274     be less than or equal to the second. For example:
1275 nigel 75 .sp
1276 nigel 63 z{2,4}
1277 nigel 75 .sp
1278 nigel 63 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1279     character. If the second number is omitted, but the comma is present, there is
1280     no upper limit; if the second number and the comma are both omitted, the
1281     quantifier specifies an exact number of required matches. Thus
1282 nigel 75 .sp
1283 nigel 63 [aeiou]{3,}
1284 nigel 75 .sp
1285 nigel 63 matches at least 3 successive vowels, but may match many more, while
1286 nigel 75 .sp
1287     \ed{8}
1288     .sp
1289 nigel 63 matches exactly 8 digits. An opening curly bracket that appears in a position
1290     where a quantifier is not allowed, or one that does not match the syntax of a
1291     quantifier, is taken as a literal character. For example, {,6} is not a
1292     quantifier, but a literal string of four characters.
1293 nigel 75 .P
1294 nigel 63 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1295 nigel 75 bytes. Thus, for example, \ex{100}{2} matches two UTF-8 characters, each of
1296     which is represented by a two-byte sequence. Similarly, when Unicode property
1297     support is available, \eX{3} matches three Unicode extended sequences, each of
1298     which may be several bytes long (and they may be of different lengths).
1299     .P
1300 nigel 63 The quantifier {0} is permitted, causing the expression to behave as if the
1301 ph10 345 previous item and the quantifier were not present. This may be useful for
1302     subpatterns that are referenced as
1303 ph10 335 .\" HTML <a href="#subpatternsassubroutines">
1304     .\" </a>
1305     subroutines
1306     .\"
1307 ph10 345 from elsewhere in the pattern. Items other than subpatterns that have a {0}
1308 ph10 335 quantifier are omitted from the compiled pattern.
1309 nigel 75 .P
1310 nigel 93 For convenience, the three most common quantifiers have single-character
1311     abbreviations:
1312 nigel 75 .sp
1313 nigel 63 * is equivalent to {0,}
1314     + is equivalent to {1,}
1315     ? is equivalent to {0,1}
1316 nigel 75 .sp
1317 nigel 63 It is possible to construct infinite loops by following a subpattern that can
1318     match no characters with a quantifier that has no upper limit, for example:
1319 nigel 75 .sp
1320 nigel 63 (a?)*
1321 nigel 75 .sp
1322 nigel 63 Earlier versions of Perl and PCRE used to give an error at compile time for
1323     such patterns. However, because there are cases where this can be useful, such
1324     patterns are now accepted, but if any repetition of the subpattern does in fact
1325     match no characters, the loop is forcibly broken.
1326 nigel 75 .P
1327 nigel 63 By default, the quantifiers are "greedy", that is, they match as much as
1328     possible (up to the maximum number of permitted times), without causing the
1329     rest of the pattern to fail. The classic example of where this gives problems
1330 nigel 75 is in trying to match comments in C programs. These appear between /* and */
1331     and within the comment, individual * and / characters may appear. An attempt to
1332     match C comments by applying the pattern
1333     .sp
1334     /\e*.*\e*/
1335     .sp
1336 nigel 63 to the string
1337 nigel 75 .sp
1338     /* first comment */ not comment /* second comment */
1339     .sp
1340 nigel 63 fails, because it matches the entire string owing to the greediness of the .*
1341     item.
1342 nigel 75 .P
1343 nigel 63 However, if a quantifier is followed by a question mark, it ceases to be
1344     greedy, and instead matches the minimum number of times possible, so the
1345     pattern
1346 nigel 75 .sp
1347     /\e*.*?\e*/
1348     .sp
1349 nigel 63 does the right thing with the C comments. The meaning of the various
1350     quantifiers is not otherwise changed, just the preferred number of matches.
1351     Do not confuse this use of question mark with its use as a quantifier in its
1352     own right. Because it has two uses, it can sometimes appear doubled, as in
1353 nigel 75 .sp
1354     \ed??\ed
1355     .sp
1356 nigel 63 which matches one digit by preference, but can match two if that is the only
1357     way the rest of the pattern matches.
1358 nigel 75 .P
1359 nigel 93 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1360 nigel 63 the quantifiers are not greedy by default, but individual ones can be made
1361     greedy by following them with a question mark. In other words, it inverts the
1362     default behaviour.
1363 nigel 75 .P
1364 nigel 63 When a parenthesized subpattern is quantified with a minimum repeat count that
1365 nigel 75 is greater than 1 or with a limited maximum, more memory is required for the
1366 nigel 63 compiled pattern, in proportion to the size of the minimum or maximum.
1367 nigel 75 .P
1368 nigel 63 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1369 nigel 93 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1370 nigel 63 implicitly anchored, because whatever follows will be tried against every
1371     character position in the subject string, so there is no point in retrying the
1372     overall match at any position after the first. PCRE normally treats such a
1373 nigel 75 pattern as though it were preceded by \eA.
1374     .P
1375 nigel 63 In cases where it is known that the subject string contains no newlines, it is
1376     worth setting PCRE_DOTALL in order to obtain this optimization, or
1377     alternatively using ^ to indicate anchoring explicitly.
1378 nigel 75 .P
1379 nigel 63 However, there is one situation where the optimization cannot be used. When .*
1380     is inside capturing parentheses that are the subject of a backreference
1381 nigel 93 elsewhere in the pattern, a match at the start may fail where a later one
1382     succeeds. Consider, for example:
1383 nigel 75 .sp
1384     (.*)abc\e1
1385     .sp
1386 nigel 63 If the subject is "xyz123abc123" the match point is the fourth character. For
1387     this reason, such a pattern is not implicitly anchored.
1388 nigel 75 .P
1389 nigel 63 When a capturing subpattern is repeated, the value captured is the substring
1390     that matched the final iteration. For example, after
1391 nigel 75 .sp
1392     (tweedle[dume]{3}\es*)+
1393     .sp
1394 nigel 63 has matched "tweedledum tweedledee" the value of the captured substring is
1395     "tweedledee". However, if there are nested capturing subpatterns, the
1396     corresponding captured values may have been set in previous iterations. For
1397     example, after
1398 nigel 75 .sp
1399 nigel 63 /(a|(b))+/
1400 nigel 75 .sp
1401 nigel 63 matches "aba" the value of the second captured substring is "b".
1402 nigel 75 .
1403     .
1404     .\" HTML <a name="atomicgroup"></a>
1406 nigel 63 .rs
1407     .sp
1408 nigel 93 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1409     repetition, failure of what follows normally causes the repeated item to be
1410     re-evaluated to see if a different number of repeats allows the rest of the
1411     pattern to match. Sometimes it is useful to prevent this, either to change the
1412     nature of the match, or to cause it fail earlier than it otherwise might, when
1413     the author of the pattern knows there is no point in carrying on.
1414 nigel 75 .P
1415     Consider, for example, the pattern \ed+foo when applied to the subject line
1416     .sp
1417 nigel 63 123456bar
1418 nigel 75 .sp
1419 nigel 63 After matching all 6 digits and then failing to match "foo", the normal
1420 nigel 75 action of the matcher is to try again with only 5 digits matching the \ed+
1421 nigel 63 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1422     (a term taken from Jeffrey Friedl's book) provides the means for specifying
1423     that once a subpattern has matched, it is not to be re-evaluated in this way.
1424 nigel 75 .P
1425 nigel 93 If we use atomic grouping for the previous example, the matcher gives up
1426 nigel 63 immediately on failing to match "foo" the first time. The notation is a kind of
1427     special parenthesis, starting with (?> as in this example:
1428 nigel 75 .sp
1429     (?>\ed+)foo
1430     .sp
1431 nigel 63 This kind of parenthesis "locks up" the part of the pattern it contains once
1432     it has matched, and a failure further into the pattern is prevented from
1433     backtracking into it. Backtracking past it to previous items, however, works as
1434     normal.
1435 nigel 75 .P
1436 nigel 63 An alternative description is that a subpattern of this type matches the string
1437     of characters that an identical standalone pattern would match, if anchored at
1438     the current point in the subject string.
1439 nigel 75 .P
1440 nigel 63 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1441     the above example can be thought of as a maximizing repeat that must swallow
1442 nigel 75 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1443 nigel 63 number of digits they match in order to make the rest of the pattern match,
1444 nigel 75 (?>\ed+) can only match an entire sequence of digits.
1445     .P
1446 nigel 63 Atomic groups in general can of course contain arbitrarily complicated
1447     subpatterns, and can be nested. However, when the subpattern for an atomic
1448     group is just a single repeated item, as in the example above, a simpler
1449     notation, called a "possessive quantifier" can be used. This consists of an
1450     additional + character following a quantifier. Using this notation, the
1451     previous example can be rewritten as
1452 nigel 75 .sp
1453     \ed++foo
1454     .sp
1455 ph10 208 Note that a possessive quantifier can be used with an entire group, for
1456     example:
1457     .sp
1458     (abc|xyz){2,3}+
1459     .sp
1460 nigel 63 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1461     option is ignored. They are a convenient notation for the simpler forms of
1462 nigel 93 atomic group. However, there is no difference in the meaning of a possessive
1463     quantifier and the equivalent atomic group, though there may be a performance
1464     difference; possessive quantifiers should be slightly faster.
1465 nigel 75 .P
1466 nigel 93 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1467     Jeffrey Friedl originated the idea (and the name) in the first edition of his
1468     book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1469     package, and PCRE copied it from there. It ultimately found its way into Perl
1470     at release 5.10.
1471 nigel 75 .P
1472 nigel 93 PCRE has an optimization that automatically "possessifies" certain simple
1473     pattern constructs. For example, the sequence A+B is treated as A++B because
1474     there is no point in backtracking into a sequence of A's when B must follow.
1475     .P
1476 nigel 63 When a pattern contains an unlimited repeat inside a subpattern that can itself
1477     be repeated an unlimited number of times, the use of an atomic group is the
1478     only way to avoid some failing matches taking a very long time indeed. The
1479     pattern
1480 nigel 75 .sp
1481     (\eD+|<\ed+>)*[!?]
1482     .sp
1483 nigel 63 matches an unlimited number of substrings that either consist of non-digits, or
1484     digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1485     quickly. However, if it is applied to
1486 nigel 75 .sp
1487 nigel 63 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1488 nigel 75 .sp
1489 nigel 63 it takes a long time before reporting failure. This is because the string can
1490 nigel 75 be divided between the internal \eD+ repeat and the external * repeat in a
1491     large number of ways, and all have to be tried. (The example uses [!?] rather
1492     than a single character at the end, because both PCRE and Perl have an
1493     optimization that allows for fast failure when a single character is used. They
1494     remember the last single character that is required for a match, and fail early
1495     if it is not present in the string.) If the pattern is changed so that it uses
1496     an atomic group, like this:
1497     .sp
1498     ((?>\eD+)|<\ed+>)*[!?]
1499     .sp
1500 nigel 63 sequences of non-digits cannot be broken, and failure happens quickly.
1501 nigel 75 .
1502     .
1503     .\" HTML <a name="backreferences"></a>
1505 nigel 63 .rs
1506     .sp
1507     Outside a character class, a backslash followed by a digit greater than 0 (and
1508     possibly further digits) is a back reference to a capturing subpattern earlier
1509     (that is, to its left) in the pattern, provided there have been that many
1510     previous capturing left parentheses.
1511 nigel 75 .P
1512 nigel 63 However, if the decimal number following the backslash is less than 10, it is
1513     always taken as a back reference, and causes an error only if there are not
1514     that many capturing left parentheses in the entire pattern. In other words, the
1515     parentheses that are referenced need not be to the left of the reference for
1516 nigel 91 numbers less than 10. A "forward back reference" of this type can make sense
1517     when a repetition is involved and the subpattern to the right has participated
1518     in an earlier iteration.
1519     .P
1520 nigel 93 It is not possible to have a numerical "forward back reference" to a subpattern
1521     whose number is 10 or more using this syntax because a sequence such as \e50 is
1522     interpreted as a character defined in octal. See the subsection entitled
1523 nigel 91 "Non-printing characters"
1524 nigel 75 .\" HTML <a href="#digitsafterbackslash">
1525     .\" </a>
1526     above
1527     .\"
1528 nigel 93 for further details of the handling of digits following a backslash. There is
1529     no such problem when named parentheses are used. A back reference to any
1530     subpattern is possible using named parentheses (see below).
1531 nigel 75 .P
1532 nigel 93 Another way of avoiding the ambiguity inherent in the use of digits following a
1533     backslash is to use the \eg escape sequence, which is a feature introduced in
1534 ph10 208 Perl 5.10. This escape must be followed by an unsigned number or a negative
1535     number, optionally enclosed in braces. These examples are all identical:
1536 nigel 93 .sp
1537     (ring), \e1
1538     (ring), \eg1
1539     (ring), \eg{1}
1540     .sp
1541 ph10 208 An unsigned number specifies an absolute reference without the ambiguity that
1542     is present in the older syntax. It is also useful when literal digits follow
1543     the reference. A negative number is a relative reference. Consider this
1544     example:
1545 nigel 93 .sp
1546     (abc(def)ghi)\eg{-1}
1547     .sp
1548     The sequence \eg{-1} is a reference to the most recently started capturing
1549     subpattern before \eg, that is, is it equivalent to \e2. Similarly, \eg{-2}
1550     would be equivalent to \e1. The use of relative references can be helpful in
1551     long patterns, and also in patterns that are created by joining together
1552     fragments that contain references within themselves.
1553     .P
1554 nigel 63 A back reference matches whatever actually matched the capturing subpattern in
1555     the current subject string, rather than anything matching the subpattern
1556     itself (see
1557     .\" HTML <a href="#subpatternsassubroutines">
1558     .\" </a>
1559     "Subpatterns as subroutines"
1560     .\"
1561     below for a way of doing that). So the pattern
1562 nigel 75 .sp
1563     (sens|respons)e and \e1ibility
1564     .sp
1565 nigel 63 matches "sense and sensibility" and "response and responsibility", but not
1566     "sense and responsibility". If caseful matching is in force at the time of the
1567     back reference, the case of letters is relevant. For example,
1568 nigel 75 .sp
1569     ((?i)rah)\es+\e1
1570     .sp
1571 nigel 63 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1572     capturing subpattern is matched caselessly.
1573 nigel 75 .P
1574 ph10 171 There are several different ways of writing back references to named
1575     subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1576     \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1577     back reference syntax, in which \eg can be used for both numeric and named
1578     references, is also supported. We could rewrite the above example in any of
1579 nigel 93 the following ways:
1580 nigel 75 .sp
1581 nigel 93 (?<p1>(?i)rah)\es+\ek<p1>
1582 ph10 171 (?'p1'(?i)rah)\es+\ek{p1}
1583 nigel 91 (?P<p1>(?i)rah)\es+(?P=p1)
1584 ph10 171 (?<p1>(?i)rah)\es+\eg{p1}
1585 nigel 75 .sp
1586 nigel 91 A subpattern that is referenced by name may appear in the pattern before or
1587     after the reference.
1588     .P
1589 nigel 63 There may be more than one back reference to the same subpattern. If a
1590     subpattern has not actually been used in a particular match, any back
1591 ph10 456 references to it always fail by default. For example, the pattern
1592 nigel 75 .sp
1593     (a|(bc))\e2
1594     .sp
1595 ph10 456 always fails if it starts to match "a" rather than "bc". However, if the
1596     PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1597     unset value matches an empty string.
1598     .P
1599     Because there may be many capturing parentheses in a pattern, all digits
1600     following a backslash are taken as part of a potential back reference number.
1601     If the pattern continues with a digit character, some delimiter must be used to
1602     terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1603     whitespace. Otherwise, the \eg{ syntax or an empty comment (see
1604 nigel 75 .\" HTML <a href="#comments">
1605     .\" </a>
1606     "Comments"
1607     .\"
1608     below) can be used.
1609     .P
1610 nigel 63 A back reference that occurs inside the parentheses to which it refers fails
1611 nigel 75 when the subpattern is first used, so, for example, (a\e1) never matches.
1612 nigel 63 However, such references can be useful inside repeated subpatterns. For
1613     example, the pattern
1614 nigel 75 .sp
1615     (a|b\e1)+
1616     .sp
1617 nigel 63 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1618     the subpattern, the back reference matches the character string corresponding
1619     to the previous iteration. In order for this to work, the pattern must be such
1620     that the first iteration does not need to match the back reference. This can be
1621     done using alternation, as in the example above, or by a quantifier with a
1622     minimum of zero.
1623 nigel 75 .
1624     .
1625     .\" HTML <a name="bigassertions"></a>
1626 nigel 63 .SH ASSERTIONS
1627     .rs
1628     .sp
1629     An assertion is a test on the characters following or preceding the current
1630     matching point that does not actually consume any characters. The simple
1631 nigel 75 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1632     .\" HTML <a href="#smallassertions">
1633     .\" </a>
1634     above.
1635     .\"
1636     .P
1637 nigel 63 More complicated assertions are coded as subpatterns. There are two kinds:
1638     those that look ahead of the current position in the subject string, and those
1639 nigel 75 that look behind it. An assertion subpattern is matched in the normal way,
1640     except that it does not cause the current matching position to be changed.
1641     .P
1642     Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1643     because it makes no sense to assert the same thing several times. If any kind
1644     of assertion contains capturing subpatterns within it, these are counted for
1645     the purposes of numbering the capturing subpatterns in the whole pattern.
1646     However, substring capturing is carried out only for positive assertions,
1647     because it does not make sense for negative assertions.
1648     .
1649     .
1650     .SS "Lookahead assertions"
1651     .rs
1652     .sp
1653 nigel 91 Lookahead assertions start with (?= for positive assertions and (?! for
1654     negative assertions. For example,
1655 nigel 75 .sp
1656     \ew+(?=;)
1657     .sp
1658 nigel 63 matches a word followed by a semicolon, but does not include the semicolon in
1659     the match, and
1660 nigel 75 .sp
1661 nigel 63 foo(?!bar)
1662 nigel 75 .sp
1663 nigel 63 matches any occurrence of "foo" that is not followed by "bar". Note that the
1664     apparently similar pattern
1665 nigel 75 .sp
1666 nigel 63 (?!foo)bar
1667 nigel 75 .sp
1668 nigel 63 does not find an occurrence of "bar" that is preceded by something other than
1669     "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1670     (?!foo) is always true when the next three characters are "bar". A
1671 nigel 75 lookbehind assertion is needed to achieve the other effect.
1672     .P
1673 nigel 63 If you want to force a matching failure at some point in a pattern, the most
1674     convenient way to do it is with (?!) because an empty string always matches, so
1675     an assertion that requires there not to be an empty string must always fail.
1676 ph10 456 The Perl 5.10 backtracking control verb (*FAIL) or (*F) is essentially a
1677     synonym for (?!).
1678 nigel 75 .
1679     .
1680     .\" HTML <a name="lookbehind"></a>
1681     .SS "Lookbehind assertions"
1682     .rs
1683     .sp
1684 nigel 63 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1685     negative assertions. For example,
1686 nigel 75 .sp
1687 nigel 63 (?<!foo)bar
1688 nigel 75 .sp
1689 nigel 63 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1690     a lookbehind assertion are restricted such that all the strings it matches must
1691 nigel 91 have a fixed length. However, if there are several top-level alternatives, they
1692     do not all have to have the same fixed length. Thus
1693 nigel 75 .sp
1694 nigel 63 (?<=bullock|donkey)
1695 nigel 75 .sp
1696 nigel 63 is permitted, but
1697 nigel 75 .sp
1698 nigel 63 (?<!dogs?|cats?)
1699 nigel 75 .sp
1700 nigel 63 causes an error at compile time. Branches that match different length strings
1701     are permitted only at the top level of a lookbehind assertion. This is an
1702 ph10 454 extension compared with Perl (5.8 and 5.10), which requires all branches to
1703 nigel 63 match the same length of string. An assertion such as
1704 nigel 75 .sp
1705 nigel 63 (?<=ab(c|de))
1706 nigel 75 .sp
1707 nigel 63 is not permitted, because its single top-level branch can match two different
1708 ph10 454 lengths, but it is acceptable to PCRE if rewritten to use two top-level
1709     branches:
1710 nigel 75 .sp
1711 nigel 63 (?<=abc|abde)
1712 nigel 75 .sp
1713 ph10 172 In some cases, the Perl 5.10 escape sequence \eK
1714 ph10 168 .\" HTML <a href="#resetmatchstart">
1715     .\" </a>
1716     (see above)
1717     .\"
1718 ph10 454 can be used instead of a lookbehind assertion to get round the fixed-length
1719     restriction.
1720 ph10 168 .P
1721 nigel 63 The implementation of lookbehind assertions is, for each alternative, to
1722 nigel 93 temporarily move the current position back by the fixed length and then try to
1723 nigel 63 match. If there are insufficient characters before the current position, the
1724 nigel 93 assertion fails.
1725 nigel 75 .P
1726     PCRE does not allow the \eC escape (which matches a single byte in UTF-8 mode)
1727 nigel 63 to appear in lookbehind assertions, because it makes it impossible to calculate
1728 nigel 93 the length of the lookbehind. The \eX and \eR escapes, which can match
1729     different numbers of bytes, are also not permitted.
1730 nigel 75 .P
1731 ph10 454 .\" HTML <a href="#subpatternsassubroutines">
1732     .\" </a>
1733     "Subroutine"
1734     .\"
1735     calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
1736     as the subpattern matches a fixed-length string.
1737     .\" HTML <a href="#recursion">
1738     .\" </a>
1739     Recursion,
1740     .\"
1741     however, is not supported.
1742     .P
1743 nigel 93 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1744 ph10 456 specify efficient matching of fixed-length strings at the end of subject
1745     strings. Consider a simple pattern such as
1746 nigel 75 .sp
1747 nigel 63 abcd$
1748 nigel 75 .sp
1749 nigel 63 when applied to a long string that does not match. Because matching proceeds
1750     from left to right, PCRE will look for each "a" in the subject and then see if
1751     what follows matches the rest of the pattern. If the pattern is specified as
1752 nigel 75 .sp
1753 nigel 63 ^.*abcd$
1754 nigel 75 .sp
1755 nigel 63 the initial .* matches the entire string at first, but when this fails (because
1756     there is no following "a"), it backtracks to match all but the last character,
1757     then all but the last two characters, and so on. Once again the search for "a"
1758     covers the entire string, from right to left, so we are no better off. However,
1759     if the pattern is written as
1760 nigel 75 .sp
1761 nigel 63 ^.*+(?<=abcd)
1762 nigel 75 .sp
1763 nigel 93 there can be no backtracking for the .*+ item; it can match only the entire
1764 nigel 63 string. The subsequent lookbehind assertion does a single test on the last four
1765     characters. If it fails, the match fails immediately. For long strings, this
1766     approach makes a significant difference to the processing time.
1767 nigel 75 .
1768     .
1769     .SS "Using multiple assertions"
1770     .rs
1771     .sp
1772 nigel 63 Several assertions (of any sort) may occur in succession. For example,
1773 nigel 75 .sp
1774     (?<=\ed{3})(?<!999)foo
1775     .sp
1776 nigel 63 matches "foo" preceded by three digits that are not "999". Notice that each of
1777     the assertions is applied independently at the same point in the subject
1778     string. First there is a check that the previous three characters are all
1779     digits, and then there is a check that the same three characters are not "999".
1780 nigel 75 This pattern does \fInot\fP match "foo" preceded by six characters, the first
1781 nigel 63 of which are digits and the last three of which are not "999". For example, it
1782     doesn't match "123abcfoo". A pattern to do that is
1783 nigel 75 .sp
1784     (?<=\ed{3}...)(?<!999)foo
1785     .sp
1786 nigel 63 This time the first assertion looks at the preceding six characters, checking
1787     that the first three are digits, and then the second assertion checks that the
1788     preceding three characters are not "999".
1789 nigel 75 .P
1790 nigel 63 Assertions can be nested in any combination. For example,
1791 nigel 75 .sp
1792 nigel 63 (?<=(?<!foo)bar)baz
1793 nigel 75 .sp
1794 nigel 63 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1795     preceded by "foo", while
1796 nigel 75 .sp
1797     (?<=\ed{3}(?!999)...)foo
1798     .sp
1799     is another pattern that matches "foo" preceded by three digits and any three
1800 nigel 63 characters that are not "999".
1801 nigel 75 .
1802     .
1803 nigel 91 .\" HTML <a name="conditions"></a>
1805 nigel 63 .rs
1806     .sp
1807     It is possible to cause the matching process to obey a subpattern
1808     conditionally or to choose between two alternative subpatterns, depending on
1809 ph10 456 the result of an assertion, or whether a specific capturing subpattern has
1810     already been matched. The two possible forms of conditional subpattern are:
1811 nigel 75 .sp
1812 nigel 63 (?(condition)yes-pattern)
1813     (?(condition)yes-pattern|no-pattern)
1814 nigel 75 .sp
1815 nigel 63 If the condition is satisfied, the yes-pattern is used; otherwise the
1816     no-pattern (if present) is used. If there are more than two alternatives in the
1817     subpattern, a compile-time error occurs.
1818 nigel 75 .P
1819 nigel 93 There are four kinds of condition: references to subpatterns, references to
1820     recursion, a pseudo-condition called DEFINE, and assertions.
1821     .
1822     .SS "Checking for a used subpattern by number"
1823     .rs
1824     .sp
1825     If the text between the parentheses consists of a sequence of digits, the
1826 ph10 456 condition is true if a capturing subpattern of that number has previously
1827     matched. If there is more than one capturing subpattern with the same number
1828     (see the earlier
1829     .\"
1830     .\" HTML <a href="#recursion">
1831     .\" </a>
1832     section about duplicate subpattern numbers),
1833     .\"
1834     the condition is true if any of them have been set. An alternative notation is
1835     to precede the digits with a plus or minus sign. In this case, the subpattern
1836     number is relative rather than absolute. The most recently opened parentheses
1837     can be referenced by (?(-1), the next most recent by (?(-2), and so on. In
1838     looping constructs it can also make sense to refer to subsequent groups with
1839     constructs such as (?(+2).
1840 nigel 91 .P
1841     Consider the following pattern, which contains non-significant white space to
1842     make it more readable (assume the PCRE_EXTENDED option) and to divide it into
1843     three parts for ease of discussion:
1844 nigel 75 .sp
1845     ( \e( )? [^()]+ (?(1) \e) )
1846     .sp
1847 nigel 63 The first part matches an optional opening parenthesis, and if that
1848     character is present, sets it as the first captured substring. The second part
1849     matches one or more characters that are not parentheses. The third part is a
1850     conditional subpattern that tests whether the first set of parentheses matched
1851     or not. If they did, that is, if subject started with an opening parenthesis,
1852     the condition is true, and so the yes-pattern is executed and a closing
1853     parenthesis is required. Otherwise, since no-pattern is not present, the
1854     subpattern matches nothing. In other words, this pattern matches a sequence of
1855 nigel 93 non-parentheses, optionally enclosed in parentheses.
1856 ph10 167 .P
1857 ph10 172 If you were embedding this pattern in a larger one, you could use a relative
1858 ph10 167 reference:
1859     .sp
1860     ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
1861     .sp
1862     This makes the fragment independent of the parentheses in the larger pattern.
1863 nigel 93 .
1864     .SS "Checking for a used subpattern by name"
1865     .rs
1866 nigel 91 .sp
1867 nigel 93 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
1868     subpattern by name. For compatibility with earlier versions of PCRE, which had
1869     this facility before Perl, the syntax (?(name)...) is also recognized. However,
1870     there is a possible ambiguity with this syntax, because subpattern names may
1871     consist entirely of digits. PCRE looks first for a named subpattern; if it
1872     cannot find one and the name consists entirely of digits, PCRE looks for a
1873     subpattern of that number, which must be greater than zero. Using subpattern
1874     names that consist entirely of digits is not recommended.
1875     .P
1876     Rewriting the above example to use a named subpattern gives this:
1877 nigel 91 .sp
1878 nigel 93 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
1879     .sp
1880     .
1881     .SS "Checking for pattern recursion"
1882     .rs
1883     .sp
1884 nigel 91 If the condition is the string (R), and there is no subpattern with the name R,
1885 nigel 93 the condition is true if a recursive call to the whole pattern or any
1886     subpattern has been made. If digits or a name preceded by ampersand follow the
1887     letter R, for example:
1888     .sp
1889     (?(R3)...) or (?(R&name)...)
1890     .sp
1891 ph10 456 the condition is true if the most recent recursion is into a subpattern whose
1892 nigel 93 number or name is given. This condition does not check the entire recursion
1893     stack.
1894 nigel 75 .P
1895 ph10 454 At "top level", all these recursion test conditions are false.
1896     .\" HTML <a href="#recursion">
1897     .\" </a>
1898     Recursive patterns
1899     .\"
1900     are described below.
1901 nigel 93 .
1902     .SS "Defining subpatterns for use by reference only"
1903     .rs
1904     .sp
1905     If the condition is the string (DEFINE), and there is no subpattern with the
1906     name DEFINE, the condition is always false. In this case, there may be only one
1907     alternative in the subpattern. It is always skipped if control reaches this
1908     point in the pattern; the idea of DEFINE is that it can be used to define
1909 ph10 454 "subroutines" that can be referenced from elsewhere. (The use of
1910     .\" HTML <a href="#subpatternsassubroutines">
1911     .\" </a>
1912     "subroutines"
1913     .\"
1914 nigel 93 is described below.) For example, a pattern to match an IPv4 address could be
1915     written like this (ignore whitespace and line breaks):
1916     .sp
1917     (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
1918     \eb (?&byte) (\e.(?&byte)){3} \eb
1919     .sp
1920     The first part of the pattern is a DEFINE group inside which a another group
1921     named "byte" is defined. This matches an individual component of an IPv4
1922     address (a number less than 256). When matching takes place, this part of the
1923 ph10 456 pattern is skipped because DEFINE acts like a false condition. The rest of the
1924     pattern uses references to the named group to match the four dot-separated
1925     components of an IPv4 address, insisting on a word boundary at each end.
1926 nigel 93 .
1927     .SS "Assertion conditions"
1928     .rs
1929     .sp
1930     If the condition is not in any of the above formats, it must be an assertion.
1931 nigel 63 This may be a positive or negative lookahead or lookbehind assertion. Consider
1932     this pattern, again containing non-significant white space, and with the two
1933     alternatives on the second line:
1934 nigel 75 .sp
1935 nigel 63 (?(?=[^a-z]*[a-z])
1936 nigel 75 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
1937     .sp
1938 nigel 63 The condition is a positive lookahead assertion that matches an optional
1939     sequence of non-letters followed by a letter. In other words, it tests for the
1940     presence of at least one letter in the subject. If a letter is found, the
1941     subject is matched against the first alternative; otherwise it is matched
1942     against the second. This pattern matches strings in one of the two forms
1943     dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
1944 nigel 75 .
1945     .
1946     .\" HTML <a name="comments"></a>
1947 nigel 63 .SH COMMENTS
1948     .rs
1949     .sp
1950 nigel 75 The sequence (?# marks the start of a comment that continues up to the next
1951 nigel 63 closing parenthesis. Nested parentheses are not permitted. The characters
1952     that make up a comment play no part in the pattern matching at all.
1953 nigel 75 .P
1954 nigel 63 If the PCRE_EXTENDED option is set, an unescaped # character outside a
1955 nigel 91 character class introduces a comment that continues to immediately after the
1956     next newline in the pattern.
1957 nigel 75 .
1958     .
1959 nigel 91 .\" HTML <a name="recursion"></a>
1960 nigel 75 .SH "RECURSIVE PATTERNS"
1961 nigel 63 .rs
1962     .sp
1963     Consider the problem of matching a string in parentheses, allowing for
1964     unlimited nested parentheses. Without the use of recursion, the best that can
1965     be done is to use a pattern that matches up to some fixed depth of nesting. It
1966 nigel 93 is not possible to handle an arbitrary nesting depth.
1967     .P
1968     For some time, Perl has provided a facility that allows regular expressions to
1969     recurse (amongst other things). It does this by interpolating Perl code in the
1970     expression at run time, and the code can refer to the expression itself. A Perl
1971     pattern using code interpolation to solve the parentheses problem can be
1972     created like this:
1973 nigel 75 .sp
1974     $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
1975     .sp
1976 nigel 63 The (?p{...}) item interpolates Perl code at run time, and in this case refers
1977 nigel 93 recursively to the pattern in which it appears.
1978 nigel 75 .P
1979 nigel 93 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
1980     supports special syntax for recursion of the entire pattern, and also for
1981     individual subpattern recursion. After its introduction in PCRE and Python,
1982 ph10 453 this kind of recursion was subsequently introduced into Perl at release 5.10.
1983 nigel 75 .P
1984 nigel 93 A special item that consists of (? followed by a number greater than zero and a
1985     closing parenthesis is a recursive call of the subpattern of the given number,
1986 ph10 454 provided that it occurs inside that subpattern. (If not, it is a
1987     .\" HTML <a href="#subpatternsassubroutines">
1988     .\" </a>
1989     "subroutine"
1990     .\"
1991 nigel 93 call, which is described in the next section.) The special item (?R) or (?0) is
1992     a recursive call of the entire regular expression.
1993 nigel 87 .P
1994     This PCRE pattern solves the nested parentheses problem (assume the
1995     PCRE_EXTENDED option is set so that white space is ignored):
1996 nigel 75 .sp
1997 ph10 456 \e( ( [^()]++ | (?R) )* \e)
1998 nigel 75 .sp
1999 nigel 63 First it matches an opening parenthesis. Then it matches any number of
2000     substrings which can either be a sequence of non-parentheses, or a recursive
2001 nigel 87 match of the pattern itself (that is, a correctly parenthesized substring).
2002 ph10 456 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2003     to avoid backtracking into sequences of non-parentheses.
2004 nigel 75 .P
2005 nigel 63 If this were part of a larger pattern, you would not want to recurse the entire
2006     pattern, so instead you could use this:
2007 nigel 75 .sp
2008 ph10 456 ( \e( ( [^()]++ | (?1) )* \e) )
2009 nigel 75 .sp
2010 nigel 63 We have put the pattern into parentheses, and caused the recursion to refer to
2011 ph10 172 them instead of the whole pattern.
2012 ph10 166 .P
2013     In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2014 ph10 456 is made easier by the use of relative references (a Perl 5.10 feature).
2015 ph10 166 Instead of (?1) in the pattern above you can write (?-2) to refer to the second
2016     most recently opened parentheses preceding the recursion. In other words, a
2017     negative number counts capturing parentheses leftwards from the point at which
2018     it is encountered.
2019     .P
2020     It is also possible to refer to subsequently opened parentheses, by writing
2021     references such as (?+2). However, these cannot be recursive because the
2022     reference is not inside the parentheses that are referenced. They are always
2023 ph10 454 .\" HTML <a href="#subpatternsassubroutines">
2024     .\" </a>
2025     "subroutine"
2026     .\"
2027     calls, as described in the next section.
2028 ph10 166 .P
2029     An alternative approach is to use named parentheses instead. The Perl syntax
2030     for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
2031     could rewrite the above example as follows:
2032 nigel 75 .sp
2033 ph10 456 (?<pn> \e( ( [^()]++ | (?&pn) )* \e) )
2034 nigel 75 .sp
2035 nigel 93 If there is more than one subpattern with the same name, the earliest one is
2036 ph10 172 used.
2037 ph10 166 .P
2038     This particular example pattern that we have been looking at contains nested
2039 ph10 456 unlimited repeats, and so the use of a possessive quantifier for matching
2040     strings of non-parentheses is important when applying the pattern to strings
2041     that do not match. For example, when this pattern is applied to
2042 nigel 75 .sp
2043 nigel 63 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2044 nigel 75 .sp
2045 ph10 456 it yields "no match" quickly. However, if a possessive quantifier is not used,
2046 nigel 63 the match runs for a very long time indeed because there are so many different
2047     ways the + and * repeats can carve up the subject, and all have to be tested
2048     before failure can be reported.
2049 nigel 75 .P
2050 nigel 63 At the end of a match, the values set for any capturing subpatterns are those
2051     from the outermost level of the recursion at which the subpattern value is set.
2052     If you want to obtain intermediate values, a callout function can be used (see
2053 nigel 93 below and the
2054 nigel 63 .\" HREF
2055 nigel 75 \fBpcrecallout\fP
2056 nigel 63 .\"
2057     documentation). If the pattern above is matched against
2058 nigel 75 .sp
2059 nigel 63 (ab(cd)ef)
2060 nigel 75 .sp
2061 nigel 63 the value for the capturing parentheses is "ef", which is the last value taken
2062     on at the top level. If additional parentheses are added, giving
2063 nigel 75 .sp
2064 ph10 456 \e( ( ( [^()]++ | (?R) )* ) \e)
2065 nigel 63 ^ ^
2066     ^ ^
2067 nigel 75 .sp
2068 nigel 63 the string they capture is "ab(cd)ef", the contents of the top level
2069     parentheses. If there are more than 15 capturing parentheses in a pattern, PCRE
2070     has to obtain extra memory to store data during a recursion, which it does by
2071 nigel 75 using \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no
2072 nigel 63 memory can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2073 nigel 75 .P
2074 nigel 63 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2075     Consider this pattern, which matches text in angle brackets, allowing for
2076     arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2077     recursing), whereas any characters are permitted at the outer level.
2078 nigel 75 .sp
2079     < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
2080     .sp
2081 nigel 63 In this pattern, (?(R) is the start of a conditional subpattern, with two
2082     different alternatives for the recursive and non-recursive cases. The (?R) item
2083     is the actual recursive call.
2084 nigel 75 .
2085     .
2086 ph10 453 .\" HTML <a name="recursiondifference"></a>
2087     .SS "Recursion difference from Perl"
2088     .rs
2089     .sp
2090     In PCRE (like Python, but unlike Perl), a recursive subpattern call is always
2091     treated as an atomic group. That is, once it has matched some of the subject
2092     string, it is never re-entered, even if it contains untried alternatives and
2093     there is a subsequent matching failure. This can be illustrated by the
2094     following pattern, which purports to match a palindromic string that contains
2095     an odd number of characters (for example, "a", "aba", "abcba", "abcdcba"):
2096     .sp
2097     ^(.|(.)(?1)\e2)$
2098     .sp
2099     The idea is that it either matches a single character, or two identical
2100     characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2101     it does not if the pattern is longer than three characters. Consider the
2102     subject string "abcba":
2103     .P
2104     At the top level, the first character is matched, but as it is not at the end
2105     of the string, the first alternative fails; the second alternative is taken
2106     and the recursion kicks in. The recursive call to subpattern 1 successfully
2107     matches the next character ("b"). (Note that the beginning and end of line
2108     tests are not part of the recursion).
2109     .P
2110     Back at the top level, the next character ("c") is compared with what
2111     subpattern 2 matched, which was "a". This fails. Because the recursion is
2112     treated as an atomic group, there are now no backtracking points, and so the
2113     entire match fails. (Perl is able, at this point, to re-enter the recursion and
2114     try the second alternative.) However, if the pattern is written with the
2115     alternatives in the other order, things are different:
2116     .sp
2117     ^((.)(?1)\e2|.)$
2118     .sp
2119     This time, the recursing alternative is tried first, and continues to recurse
2120     until it runs out of characters, at which point the recursion fails. But this
2121     time we do have another alternative to try at the higher level. That is the big
2122     difference: in the previous case the remaining alternative is at a deeper
2123     recursion level, which PCRE cannot use.
2124     .P
2125     To change the pattern so that matches all palindromic strings, not just those
2126     with an odd number of characters, it is tempting to change the pattern to this:
2127     .sp
2128     ^((.)(?1)\e2|.?)$
2129     .sp
2130     Again, this works in Perl, but not in PCRE, and for the same reason. When a
2131     deeper recursion has matched a single character, it cannot be entered again in
2132     order to match an empty string. The solution is to separate the two cases, and
2133     write out the odd and even cases as alternatives at the higher level:
2134     .sp
2135     ^(?:((.)(?1)\e2|)|((.)(?3)\e4|.))
2136     .sp
2137     If you want to match typical palindromic phrases, the pattern has to ignore all
2138     non-word characters, which can be done like this:
2139     .sp
2140     ^\eW*+(?:((.)\eW*+(?1)\eW*+\e2|)|((.)\eW*+(?3)\eW*+\4|\eW*+.\eW*+))\eW*+$
2141     .sp
2142     If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2143     man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2144     the use of the possessive quantifier *+ to avoid backtracking into sequences of
2145     non-word characters. Without this, PCRE takes a great deal longer (ten times or
2146     more) to match typical phrases, and Perl takes so long that you think it has
2147     gone into a loop.
2148 ph10 456 .P
2149     \fBWARNING\fP: The palindrome-matching patterns above work only if the subject
2150     string does not start with a palindrome that is shorter than the entire string.
2151     For example, although "abcba" is correctly matched, if the subject is "ababa",
2152     PCRE finds the palindrome "aba" at the start, then fails at top level because
2153     the end of the string does not follow. Once again, it cannot jump back into the
2154     recursion to try other alternatives, so the entire match fails.
2155 ph10 453 .
2156     .
2157 nigel 63 .\" HTML <a name="subpatternsassubroutines"></a>
2159 nigel 63 .rs
2160     .sp
2161     If the syntax for a recursive subpattern reference (either by number or by
2162     name) is used outside the parentheses to which it refers, it operates like a
2163 nigel 93 subroutine in a programming language. The "called" subpattern may be defined
2164 ph10 166 before or after the reference. A numbered reference can be absolute or
2165     relative, as in these examples:
2166 nigel 75 .sp
2167 ph10 166 (...(absolute)...)...(?2)...
2168     (...(relative)...)...(?-1)...
2169 ph10 172 (...(?+1)...(relative)...
2170 ph10 166 .sp
2171     An earlier example pointed out that the pattern
2172     .sp
2173 nigel 75 (sens|respons)e and \e1ibility
2174     .sp
2175 nigel 63 matches "sense and sensibility" and "response and responsibility", but not
2176     "sense and responsibility". If instead the pattern
2177 nigel 75 .sp
2178 nigel 63 (sens|respons)e and (?1)ibility
2179 nigel 75 .sp
2180 nigel 63 is used, it does match "sense and responsibility" as well as the other two
2181 nigel 93 strings. Another example is given in the discussion of DEFINE above.
2182 nigel 87 .P
2183     Like recursive subpatterns, a "subroutine" call is always treated as an atomic
2184     group. That is, once it has matched some of the subject string, it is never
2185     re-entered, even if it contains untried alternatives and there is a subsequent
2186     matching failure.
2187 nigel 93 .P
2188     When a subpattern is used as a subroutine, processing options such as
2189     case-independence are fixed when the subpattern is defined. They cannot be
2190     changed for different calls. For example, consider this pattern:
2191     .sp
2192 ph10 166 (abc)(?i:(?-1))
2193 nigel 93 .sp
2194     It matches "abcabc". It does not match "abcABC" because the change of
2195     processing option does not affect the called subpattern.
2196 nigel 75 .
2197     .
2198 ph10 333 .\" HTML <a name="onigurumasubroutines"></a>
2200     .rs
2201     .sp
2202 ph10 345 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
2203     a number enclosed either in angle brackets or single quotes, is an alternative
2204     syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2205 ph10 333 are two of the examples used above, rewritten using this syntax:
2206     .sp
2207     (?<pn> \e( ( (?>[^()]+) | \eg<pn> )* \e) )
2208     (sens|respons)e and \eg'1'ibility
2209     .sp
2210 ph10 345 PCRE supports an extension to Oniguruma: if a number is preceded by a
2211 ph10 333 plus or a minus sign it is taken as a relative reference. For example:
2212     .sp
2213     (abc)(?i:\eg<-1>)
2214     .sp
2215 ph10 345 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
2216 ph10 333 synonymous. The former is a back reference; the latter is a subroutine call.
2217     .
2218     .
2219 nigel 63 .SH CALLOUTS
2220     .rs
2221     .sp
2222     Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2223     code to be obeyed in the middle of matching a regular expression. This makes it
2224     possible, amongst other things, to extract different substrings that match the
2225     same pair of parentheses when there is a repetition.
2226 nigel 75 .P
2227 nigel 63 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2228     code. The feature is called "callout". The caller of PCRE provides an external
2229 nigel 75 function by putting its entry point in the global variable \fIpcre_callout\fP.
2230 nigel 63 By default, this variable contains NULL, which disables all calling out.
2231 nigel 75 .P
2232 nigel 63 Within a regular expression, (?C) indicates the points at which the external
2233     function is to be called. If you want to identify different callout points, you
2234     can put a number less than 256 after the letter C. The default value is zero.
2235     For example, this pattern has two callout points:
2236 nigel 75 .sp
2237 ph10 155 (?C1)abc(?C2)def
2238 nigel 75 .sp
2239     If the PCRE_AUTO_CALLOUT flag is passed to \fBpcre_compile()\fP, callouts are
2240     automatically installed before each item in the pattern. They are all numbered
2241     255.
2242     .P
2243     During matching, when PCRE reaches a callout point (and \fIpcre_callout\fP is
2244 nigel 63 set), the external function is called. It is provided with the number of the
2245 nigel 75 callout, the position in the pattern, and, optionally, one item of data
2246     originally supplied by the caller of \fBpcre_exec()\fP. The callout function
2247     may cause matching to proceed, to backtrack, or to fail altogether. A complete
2248     description of the interface to the callout function is given in the
2249 nigel 63 .\" HREF
2250 nigel 75 \fBpcrecallout\fP
2251 nigel 63 .\"
2252     documentation.
2253 nigel 93 .
2254     .
2256 ph10 210 .rs
2257     .sp
2258 ph10 211 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2259 ph10 210 are described in the Perl documentation as "experimental and subject to change
2260 ph10 211 or removal in a future version of Perl". It goes on to say: "Their usage in
2261     production code should be noted to avoid problems during upgrades." The same
2262 ph10 210 remarks apply to the PCRE features described in this section.
2263     .P
2264 ph10 341 Since these verbs are specifically related to backtracking, most of them can be
2265     used only when the pattern is to be matched using \fBpcre_exec()\fP, which uses
2266 ph10 345 a backtracking algorithm. With the exception of (*FAIL), which behaves like a
2267 ph10 341 failing negative assertion, they cause an error if encountered by
2268 ph10 210 \fBpcre_dfa_exec()\fP.
2269     .P
2270 ph10 445 If any of these verbs are used in an assertion subpattern, their effect is
2271     confined to that subpattern; it does not extend to the surrounding pattern.
2272     Note that assertion subpatterns are processed as anchored at the point where
2273     they are tested.
2274     .P
2275 ph10 211 The new verbs make use of what was previously invalid syntax: an opening
2276 ph10 210 parenthesis followed by an asterisk. In Perl, they are generally of the form
2277     (*VERB:ARG) but PCRE does not support the use of arguments, so its general
2278     form is just (*VERB). Any number of these verbs may occur in a pattern. There
2279     are two kinds:
2280     .
2281     .SS "Verbs that act immediately"
2282     .rs
2283     .sp
2284     The following verbs act as soon as they are encountered:
2285     .sp
2286     (*ACCEPT)
2287     .sp
2288     This verb causes the match to end successfully, skipping the remainder of the
2289 ph10 211 pattern. When inside a recursion, only the innermost pattern is ended
2290 ph10 456 immediately. If (*ACCEPT) is inside capturing parentheses, the data so far is
2291     captured. (This feature was added to PCRE at release 8.00.) For example:
2292 ph10 210 .sp
2293 ph10 447 A((?:A|B(*ACCEPT)|C)D)
2294 ph10 210 .sp
2295 ph10 447 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2296     the outer parentheses.
2297 ph10 210 .sp
2298     (*FAIL) or (*F)
2299     .sp
2300 ph10 211 This verb causes the match to fail, forcing backtracking to occur. It is
2301 ph10 210 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2302     probably useful only when combined with (?{}) or (??{}). Those are, of course,
2303     Perl features that are not present in PCRE. The nearest equivalent is the
2304     callout feature, as for example in this pattern:
2305     .sp
2306     a+(?C)(*FAIL)
2307     .sp
2308 ph10 211 A match with the string "aaaa" always fails, but the callout is taken before
2309     each backtrack happens (in this example, 10 times).
2310 ph10 210 .
2311     .SS "Verbs that act after backtracking"
2312     .rs
2313     .sp
2314 ph10 211 The following verbs do nothing when they are encountered. Matching continues
2315     with what follows, but if there is no subsequent match, a failure is forced.
2316 ph10 210 The verbs differ in exactly what kind of failure occurs.
2317     .sp
2318     (*COMMIT)
2319     .sp
2320 ph10 211 This verb causes the whole match to fail outright if the rest of the pattern
2321 ph10 210 does not match. Even if the pattern is unanchored, no further attempts to find
2322 ph10 456 a match by advancing the starting point take place. Once (*COMMIT) has been
2323 ph10 211 passed, \fBpcre_exec()\fP is committed to finding a match at the current
2324 ph10 210 starting point, or not at all. For example:
2325     .sp
2326     a+(*COMMIT)b
2327     .sp
2328 ph10 211 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2329 ph10 210 dynamic anchor, or "I've started, so I must finish."
2330     .sp
2331     (*PRUNE)
2332     .sp
2333 ph10 211 This verb causes the match to fail at the current position if the rest of the
2334     pattern does not match. If the pattern is unanchored, the normal "bumpalong"
2335 ph10 210 advance to the next starting character then happens. Backtracking can occur as
2336     usual to the left of (*PRUNE), or when matching to the right of (*PRUNE), but
2337     if there is no match to the right, backtracking cannot cross (*PRUNE).
2338 ph10 211 In simple cases, the use of (*PRUNE) is just an alternative to an atomic
2339 ph10 210 group or possessive quantifier, but there are some uses of (*PRUNE) that cannot
2340     be expressed in any other way.
2341     .sp
2342     (*SKIP)
2343     .sp
2344 ph10 211 This verb is like (*PRUNE), except that if the pattern is unanchored, the
2345 ph10 210 "bumpalong" advance is not to the next character, but to the position in the
2346     subject where (*SKIP) was encountered. (*SKIP) signifies that whatever text
2347     was matched leading up to it cannot be part of a successful match. Consider:
2348     .sp
2349     a+(*SKIP)b
2350     .sp
2351 ph10 211 If the subject is "aaaac...", after the first match attempt fails (starting at
2352 ph10 210 the first character in the string), the starting point skips on to start the
2353 ph10 211 next attempt at "c". Note that a possessive quantifer does not have the same
2354 ph10 456 effect as this example; although it would suppress backtracking during the
2355 ph10 210 first match attempt, the second attempt would start at the second character
2356     instead of skipping on to "c".
2357     .sp
2358     (*THEN)
2359 ph10 211 .sp
2360 ph10 210 This verb causes a skip to the next alternation if the rest of the pattern does
2361     not match. That is, it cancels pending backtracking, but only within the
2362     current alternation. Its name comes from the observation that it can be used
2363     for a pattern-based if-then-else block:
2364     .sp
2365     ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2366     .sp
2367 ph10 211 If the COND1 pattern matches, FOO is tried (and possibly further items after
2368 ph10 210 the end of the group if FOO succeeds); on failure the matcher skips to the
2369 ph10 211 second alternative and tries COND2, without backtracking into COND1. If (*THEN)
2370 ph10 210 is used outside of any alternation, it acts exactly like (*PRUNE).
2371     .
2372     .
2373 nigel 93 .SH "SEE ALSO"
2374     .rs
2375     .sp
2376 ph10 456 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3),
2377     \fBpcresyntax\fP(3), \fBpcre\fP(3).
2378 ph10 99 .
2379     .
2380     .SH AUTHOR
2381     .rs
2382     .sp
2383     .nf
2384     Philip Hazel
2385     University Computing Service
2386     Cambridge CB2 3QH, England.
2387     .fi
2388     .
2389     .
2390     .SH REVISION
2391     .rs
2392     .sp
2393     .nf
2394 ph10 457 Last updated: 03 October 2009
2395 ph10 385 Copyright (c) 1997-2009 University of Cambridge.
2396 ph10 99 .fi


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

ViewVC Help
Powered by ViewVC 1.1.12