/[pcre]/code/trunk/doc/pcrepattern.3
ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 518 - (show annotations) (download)
Tue May 18 15:47:01 2010 UTC (4 years, 6 months ago) by ph10
File size: 109479 byte(s)
Added PCRE_UCP and related stuff to make \w etc use Unicode properties.

1 .TH PCREPATTERN 3
2 .SH NAME
3 PCRE - Perl-compatible regular expressions
4 .SH "PCRE REGULAR EXPRESSION DETAILS"
5 .rs
6 .sp
7 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 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 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 .P
23 The original operation of PCRE was on strings of one-byte characters. However,
24 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 .sp
29 (*UTF8)
30 .sp
31 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 .\" HTML <a href="pcre.html#utf8support">
36 .\" </a>
37 section on UTF-8 support
38 .\"
39 in the main
40 .\" HREF
41 \fBpcre\fP
42 .\"
43 page.
44 .P
45 Another special sequence that may appear at the start of a pattern or in
46 combination with (*UTF8) is:
47 .sp
48 (*UCP)
49 .sp
50 This has the same effect as setting the PCRE_UCP option: it causes sequences
51 such as \ed and \ew to use Unicode properties to determine character types,
52 instead of recognizing only characters with codes less than 128 via a lookup
53 table.
54 .P
55 The remainder of this document discusses the patterns that are supported by
56 PCRE when its main matching function, \fBpcre_exec()\fP, is used.
57 From release 6.0, PCRE offers a second matching function,
58 \fBpcre_dfa_exec()\fP, which matches using a different algorithm that is not
59 Perl-compatible. Some of the features discussed below are not available when
60 \fBpcre_dfa_exec()\fP is used. The advantages and disadvantages of the
61 alternative function, and how it differs from the normal function, are
62 discussed in the
63 .\" HREF
64 \fBpcrematching\fP
65 .\"
66 page.
67 .
68 .
69 .SH "NEWLINE CONVENTIONS"
70 .rs
71 .sp
72 PCRE supports five different conventions for indicating line breaks in
73 strings: a single CR (carriage return) character, a single LF (linefeed)
74 character, the two-character sequence CRLF, any of the three preceding, or any
75 Unicode newline sequence. The
76 .\" HREF
77 \fBpcreapi\fP
78 .\"
79 page has
80 .\" HTML <a href="pcreapi.html#newlines">
81 .\" </a>
82 further discussion
83 .\"
84 about newlines, and shows how to set the newline convention in the
85 \fIoptions\fP arguments for the compiling and matching functions.
86 .P
87 It is also possible to specify a newline convention by starting a pattern
88 string with one of the following five sequences:
89 .sp
90 (*CR) carriage return
91 (*LF) linefeed
92 (*CRLF) carriage return, followed by linefeed
93 (*ANYCRLF) any of the three above
94 (*ANY) all Unicode newline sequences
95 .sp
96 These override the default and the options given to \fBpcre_compile()\fP or
97 \fBpcre_compile2()\fP. For example, on a Unix system where LF is the default
98 newline sequence, the pattern
99 .sp
100 (*CR)a.b
101 .sp
102 changes the convention to CR. That pattern matches "a\enb" because LF is no
103 longer a newline. Note that these special settings, which are not
104 Perl-compatible, are recognized only at the very start of a pattern, and that
105 they must be in upper case. If more than one of them is present, the last one
106 is used.
107 .P
108 The newline convention affects the interpretation of the dot metacharacter when
109 PCRE_DOTALL is not set, and also the behaviour of \eN. However, it does not
110 affect what the \eR escape sequence matches. By default, this is any Unicode
111 newline sequence, for Perl compatibility. However, this can be changed; see the
112 description of \eR in the section entitled
113 .\" HTML <a href="#newlineseq">
114 .\" </a>
115 "Newline sequences"
116 .\"
117 below. A change of \eR setting can be combined with a change of newline
118 convention.
119 .
120 .
121 .SH "CHARACTERS AND METACHARACTERS"
122 .rs
123 .sp
124 A regular expression is a pattern that is matched against a subject string from
125 left to right. Most characters stand for themselves in a pattern, and match the
126 corresponding characters in the subject. As a trivial example, the pattern
127 .sp
128 The quick brown fox
129 .sp
130 matches a portion of a subject string that is identical to itself. When
131 caseless matching is specified (the PCRE_CASELESS option), letters are matched
132 independently of case. In UTF-8 mode, PCRE always understands the concept of
133 case for characters whose values are less than 128, so caseless matching is
134 always possible. For characters with higher values, the concept of case is
135 supported if PCRE is compiled with Unicode property support, but not otherwise.
136 If you want to use caseless matching for characters 128 and above, you must
137 ensure that PCRE is compiled with Unicode property support as well as with
138 UTF-8 support.
139 .P
140 The power of regular expressions comes from the ability to include alternatives
141 and repetitions in the pattern. These are encoded in the pattern by the use of
142 \fImetacharacters\fP, which do not stand for themselves but instead are
143 interpreted in some special way.
144 .P
145 There are two different sets of metacharacters: those that are recognized
146 anywhere in the pattern except within square brackets, and those that are
147 recognized within square brackets. Outside square brackets, the metacharacters
148 are as follows:
149 .sp
150 \e general escape character with several uses
151 ^ assert start of string (or line, in multiline mode)
152 $ assert end of string (or line, in multiline mode)
153 . match any character except newline (by default)
154 [ start character class definition
155 | start of alternative branch
156 ( start subpattern
157 ) end subpattern
158 ? extends the meaning of (
159 also 0 or 1 quantifier
160 also quantifier minimizer
161 * 0 or more quantifier
162 + 1 or more quantifier
163 also "possessive quantifier"
164 { start min/max quantifier
165 .sp
166 Part of a pattern that is in square brackets is called a "character class". In
167 a character class the only metacharacters are:
168 .sp
169 \e general escape character
170 ^ negate the class, but only if the first character
171 - indicates character range
172 .\" JOIN
173 [ POSIX character class (only if followed by POSIX
174 syntax)
175 ] terminates the character class
176 .sp
177 The following sections describe the use of each of the metacharacters.
178 .
179 .
180 .SH BACKSLASH
181 .rs
182 .sp
183 The backslash character has several uses. Firstly, if it is followed by a
184 non-alphanumeric character, it takes away any special meaning that character
185 may have. This use of backslash as an escape character applies both inside and
186 outside character classes.
187 .P
188 For example, if you want to match a * character, you write \e* in the pattern.
189 This escaping action applies whether or not the following character would
190 otherwise be interpreted as a metacharacter, so it is always safe to precede a
191 non-alphanumeric with backslash to specify that it stands for itself. In
192 particular, if you want to match a backslash, you write \e\e.
193 .P
194 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
195 pattern (other than in a character class) and characters between a # outside
196 a character class and the next newline are ignored. An escaping backslash can
197 be used to include a whitespace or # character as part of the pattern.
198 .P
199 If you want to remove the special meaning from a sequence of characters, you
200 can do so by putting them between \eQ and \eE. This is different from Perl in
201 that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
202 Perl, $ and @ cause variable interpolation. Note the following examples:
203 .sp
204 Pattern PCRE matches Perl matches
205 .sp
206 .\" JOIN
207 \eQabc$xyz\eE abc$xyz abc followed by the
208 contents of $xyz
209 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
210 \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
211 .sp
212 The \eQ...\eE sequence is recognized both inside and outside character classes.
213 .
214 .
215 .\" HTML <a name="digitsafterbackslash"></a>
216 .SS "Non-printing characters"
217 .rs
218 .sp
219 A second use of backslash provides a way of encoding non-printing characters
220 in patterns in a visible manner. There is no restriction on the appearance of
221 non-printing characters, apart from the binary zero that terminates a pattern,
222 but when a pattern is being prepared by text editing, it is often easier to use
223 one of the following escape sequences than the binary character it represents:
224 .sp
225 \ea alarm, that is, the BEL character (hex 07)
226 \ecx "control-x", where x is any character
227 \ee escape (hex 1B)
228 \ef formfeed (hex 0C)
229 \en linefeed (hex 0A)
230 \er carriage return (hex 0D)
231 \et tab (hex 09)
232 \eddd character with octal code ddd, or back reference
233 \exhh character with hex code hh
234 \ex{hhh..} character with hex code hhh..
235 .sp
236 The precise effect of \ecx is as follows: if x is a lower case letter, it
237 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
238 Thus \ecz becomes hex 1A, but \ec{ becomes hex 3B, while \ec; becomes hex
239 7B.
240 .P
241 After \ex, from zero to two hexadecimal digits are read (letters can be in
242 upper or lower case). Any number of hexadecimal digits may appear between \ex{
243 and }, but the value of the character code must be less than 256 in non-UTF-8
244 mode, and less than 2**31 in UTF-8 mode. That is, the maximum value in
245 hexadecimal is 7FFFFFFF. Note that this is bigger than the largest Unicode code
246 point, which is 10FFFF.
247 .P
248 If characters other than hexadecimal digits appear between \ex{ and }, or if
249 there is no terminating }, this form of escape is not recognized. Instead, the
250 initial \ex will be interpreted as a basic hexadecimal escape, with no
251 following digits, giving a character whose value is zero.
252 .P
253 Characters whose value is less than 256 can be defined by either of the two
254 syntaxes for \ex. There is no difference in the way they are handled. For
255 example, \exdc is exactly the same as \ex{dc}.
256 .P
257 After \e0 up to two further octal digits are read. If there are fewer than two
258 digits, just those that are present are used. Thus the sequence \e0\ex\e07
259 specifies two binary zeros followed by a BEL character (code value 7). Make
260 sure you supply two digits after the initial zero if the pattern character that
261 follows is itself an octal digit.
262 .P
263 The handling of a backslash followed by a digit other than 0 is complicated.
264 Outside a character class, PCRE reads it and any following digits as a decimal
265 number. If the number is less than 10, or if there have been at least that many
266 previous capturing left parentheses in the expression, the entire sequence is
267 taken as a \fIback reference\fP. A description of how this works is given
268 .\" HTML <a href="#backreferences">
269 .\" </a>
270 later,
271 .\"
272 following the discussion of
273 .\" HTML <a href="#subpattern">
274 .\" </a>
275 parenthesized subpatterns.
276 .\"
277 .P
278 Inside a character class, or if the decimal number is greater than 9 and there
279 have not been that many capturing subpatterns, PCRE re-reads up to three octal
280 digits following the backslash, and uses them to generate a data character. Any
281 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
282 character specified in octal must be less than \e400. In UTF-8 mode, values up
283 to \e777 are permitted. For example:
284 .sp
285 \e040 is another way of writing a space
286 .\" JOIN
287 \e40 is the same, provided there are fewer than 40
288 previous capturing subpatterns
289 \e7 is always a back reference
290 .\" JOIN
291 \e11 might be a back reference, or another way of
292 writing a tab
293 \e011 is always a tab
294 \e0113 is a tab followed by the character "3"
295 .\" JOIN
296 \e113 might be a back reference, otherwise the
297 character with octal code 113
298 .\" JOIN
299 \e377 might be a back reference, otherwise
300 the byte consisting entirely of 1 bits
301 .\" JOIN
302 \e81 is either a back reference, or a binary zero
303 followed by the two characters "8" and "1"
304 .sp
305 Note that octal values of 100 or greater must not be introduced by a leading
306 zero, because no more than three octal digits are ever read.
307 .P
308 All the sequences that define a single character value can be used both inside
309 and outside character classes. In addition, inside a character class, the
310 sequence \eb is interpreted as the backspace character (hex 08). The sequences
311 \eB, \eN, \eR, and \eX are not special inside a character class. Like any other
312 unrecognized escape sequences, they are treated as the literal characters "B",
313 "N", "R", and "X" by default, but cause an error if the PCRE_EXTRA option is
314 set. Outside a character class, these sequences have different meanings.
315 .
316 .
317 .SS "Absolute and relative back references"
318 .rs
319 .sp
320 The sequence \eg followed by an unsigned or a negative number, optionally
321 enclosed in braces, is an absolute or relative back reference. A named back
322 reference can be coded as \eg{name}. Back references are discussed
323 .\" HTML <a href="#backreferences">
324 .\" </a>
325 later,
326 .\"
327 following the discussion of
328 .\" HTML <a href="#subpattern">
329 .\" </a>
330 parenthesized subpatterns.
331 .\"
332 .
333 .
334 .SS "Absolute and relative subroutine calls"
335 .rs
336 .sp
337 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
338 a number enclosed either in angle brackets or single quotes, is an alternative
339 syntax for referencing a subpattern as a "subroutine". Details are discussed
340 .\" HTML <a href="#onigurumasubroutines">
341 .\" </a>
342 later.
343 .\"
344 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
345 synonymous. The former is a back reference; the latter is a
346 .\" HTML <a href="#subpatternsassubroutines">
347 .\" </a>
348 subroutine
349 .\"
350 call.
351 .
352 .
353 .\" HTML <a name="genericchartypes"></a>
354 .SS "Generic character types"
355 .rs
356 .sp
357 Another use of backslash is for specifying generic character types:
358 .sp
359 \ed any decimal digit
360 \eD any character that is not a decimal digit
361 \eh any horizontal whitespace character
362 \eH any character that is not a horizontal whitespace character
363 \es any whitespace character
364 \eS any character that is not a whitespace character
365 \ev any vertical whitespace character
366 \eV any character that is not a vertical whitespace character
367 \ew any "word" character
368 \eW any "non-word" character
369 .sp
370 There is also the single sequence \eN, which matches a non-newline character.
371 This is the same as
372 .\" HTML <a href="#fullstopdot">
373 .\" </a>
374 the "." metacharacter
375 .\"
376 when PCRE_DOTALL is not set.
377 .P
378 Each pair of lower and upper case escape sequences partitions the complete set
379 of characters into two disjoint sets. Any given character matches one, and only
380 one, of each pair. The sequences can appear both inside and outside character
381 classes. They each match one character of the appropriate type. If the current
382 matching point is at the end of the subject string, all of them fail, because
383 there is no character to match.
384 .P
385 For compatibility with Perl, \es does not match the VT character (code 11).
386 This makes it different from the the POSIX "space" class. The \es characters
387 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
388 included in a Perl script, \es may match the VT character. In PCRE, it never
389 does.
390 .P
391 A "word" character is an underscore or any character that is a letter or digit.
392 By default, the definition of letters and digits is controlled by PCRE's
393 low-valued character tables, and may vary if locale-specific matching is taking
394 place (see
395 .\" HTML <a href="pcreapi.html#localesupport">
396 .\" </a>
397 "Locale support"
398 .\"
399 in the
400 .\" HREF
401 \fBpcreapi\fP
402 .\"
403 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
404 or "french" in Windows, some character codes greater than 128 are used for
405 accented letters, and these are then matched by \ew. The use of locales with
406 Unicode is discouraged.
407 .P
408 By default, in UTF-8 mode, characters with values greater than 128 never match
409 \ed, \es, or \ew, and always match \eD, \eS, and \eW. These sequences retain
410 their original meanings from before UTF-8 support was available, mainly for
411 efficiency reasons. However, if PCRE is compiled with Unicode property support,
412 and the PCRE_UCP option is set, the behaviour is changed so that Unicode
413 properties are used to determine character types, as follows:
414 .sp
415 \ed any character that \ep{Nd} matches (decimal digit)
416 \es any character that \ep{Z} matches, plus HT, LF, FF, CR
417 \ew any character that \ep{L} or \ep{N} matches, plus underscore
418 .sp
419 The upper case escapes match the inverse sets of characters. Note that \ed
420 matches only decimal digits, whereas \ew matches any Unicode digit, as well as
421 any Unicode letter, and underscore. Note also that PCRE_UCP affects \eb, and
422 \eB because they are defined in terms of \ew and \eW. Matching these sequences
423 is noticeably slower when PCRE_UCP is set.
424 .P
425 The sequences \eh, \eH, \ev, and \eV are Perl 5.10 features. In contrast to the
426 other sequences, which match only ASCII characters by default, these always
427 match certain high-valued codepoints in UTF-8 mode, whether or not PCRE_UCP is
428 set. The horizontal space characters are:
429 .sp
430 U+0009 Horizontal tab
431 U+0020 Space
432 U+00A0 Non-break space
433 U+1680 Ogham space mark
434 U+180E Mongolian vowel separator
435 U+2000 En quad
436 U+2001 Em quad
437 U+2002 En space
438 U+2003 Em space
439 U+2004 Three-per-em space
440 U+2005 Four-per-em space
441 U+2006 Six-per-em space
442 U+2007 Figure space
443 U+2008 Punctuation space
444 U+2009 Thin space
445 U+200A Hair space
446 U+202F Narrow no-break space
447 U+205F Medium mathematical space
448 U+3000 Ideographic space
449 .sp
450 The vertical space characters are:
451 .sp
452 U+000A Linefeed
453 U+000B Vertical tab
454 U+000C Formfeed
455 U+000D Carriage return
456 U+0085 Next line
457 U+2028 Line separator
458 U+2029 Paragraph separator
459 .
460 .
461 .\" HTML <a name="newlineseq"></a>
462 .SS "Newline sequences"
463 .rs
464 .sp
465 Outside a character class, by default, the escape sequence \eR matches any
466 Unicode newline sequence. This is a Perl 5.10 feature. In non-UTF-8 mode \eR is
467 equivalent to the following:
468 .sp
469 (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
470 .sp
471 This is an example of an "atomic group", details of which are given
472 .\" HTML <a href="#atomicgroup">
473 .\" </a>
474 below.
475 .\"
476 This particular group matches either the two-character sequence CR followed by
477 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
478 U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
479 line, U+0085). The two-character sequence is treated as a single unit that
480 cannot be split.
481 .P
482 In UTF-8 mode, two additional characters whose codepoints are greater than 255
483 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
484 Unicode character property support is not needed for these characters to be
485 recognized.
486 .P
487 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
488 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
489 either at compile time or when the pattern is matched. (BSR is an abbrevation
490 for "backslash R".) This can be made the default when PCRE is built; if this is
491 the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
492 It is also possible to specify these settings by starting a pattern string with
493 one of the following sequences:
494 .sp
495 (*BSR_ANYCRLF) CR, LF, or CRLF only
496 (*BSR_UNICODE) any Unicode newline sequence
497 .sp
498 These override the default and the options given to \fBpcre_compile()\fP or
499 \fBpcre_compile2()\fP, but they can be overridden by options given to
500 \fBpcre_exec()\fP or \fBpcre_dfa_exec()\fP. Note that these special settings,
501 which are not Perl-compatible, are recognized only at the very start of a
502 pattern, and that they must be in upper case. If more than one of them is
503 present, the last one is used. They can be combined with a change of newline
504 convention; for example, a pattern can start with:
505 .sp
506 (*ANY)(*BSR_ANYCRLF)
507 .sp
508 They can also be combined with the (*UTF8) or (*UCP) special sequences. Inside
509 a character class, \eR is treated as an unrecognized escape sequence, and so
510 matches the letter "R" by default, but causes an error if PCRE_EXTRA is set.
511 .
512 .
513 .\" HTML <a name="uniextseq"></a>
514 .SS Unicode character properties
515 .rs
516 .sp
517 When PCRE is built with Unicode character property support, three additional
518 escape sequences that match characters with specific properties are available.
519 When not in UTF-8 mode, these sequences are of course limited to testing
520 characters whose codepoints are less than 256, but they do work in this mode.
521 The extra escape sequences are:
522 .sp
523 \ep{\fIxx\fP} a character with the \fIxx\fP property
524 \eP{\fIxx\fP} a character without the \fIxx\fP property
525 \eX an extended Unicode sequence
526 .sp
527 The property names represented by \fIxx\fP above are limited to the Unicode
528 script names, the general category properties, "Any", which matches any
529 character (including newline), and some special PCRE properties (described
530 in the
531 .\" HTML <a href="#extraprops">
532 .\" </a>
533 next section).
534 .\"
535 Other Perl properties such as "InMusicalSymbols" are not currently supported by
536 PCRE. Note that \eP{Any} does not match any characters, so always causes a
537 match failure.
538 .P
539 Sets of Unicode characters are defined as belonging to certain scripts. A
540 character from one of these sets can be matched using a script name. For
541 example:
542 .sp
543 \ep{Greek}
544 \eP{Han}
545 .sp
546 Those that are not part of an identified script are lumped together as
547 "Common". The current list of scripts is:
548 .P
549 Arabic,
550 Armenian,
551 Avestan,
552 Balinese,
553 Bamum,
554 Bengali,
555 Bopomofo,
556 Braille,
557 Buginese,
558 Buhid,
559 Canadian_Aboriginal,
560 Carian,
561 Cham,
562 Cherokee,
563 Common,
564 Coptic,
565 Cuneiform,
566 Cypriot,
567 Cyrillic,
568 Deseret,
569 Devanagari,
570 Egyptian_Hieroglyphs,
571 Ethiopic,
572 Georgian,
573 Glagolitic,
574 Gothic,
575 Greek,
576 Gujarati,
577 Gurmukhi,
578 Han,
579 Hangul,
580 Hanunoo,
581 Hebrew,
582 Hiragana,
583 Imperial_Aramaic,
584 Inherited,
585 Inscriptional_Pahlavi,
586 Inscriptional_Parthian,
587 Javanese,
588 Kaithi,
589 Kannada,
590 Katakana,
591 Kayah_Li,
592 Kharoshthi,
593 Khmer,
594 Lao,
595 Latin,
596 Lepcha,
597 Limbu,
598 Linear_B,
599 Lisu,
600 Lycian,
601 Lydian,
602 Malayalam,
603 Meetei_Mayek,
604 Mongolian,
605 Myanmar,
606 New_Tai_Lue,
607 Nko,
608 Ogham,
609 Old_Italic,
610 Old_Persian,
611 Old_South_Arabian,
612 Old_Turkic,
613 Ol_Chiki,
614 Oriya,
615 Osmanya,
616 Phags_Pa,
617 Phoenician,
618 Rejang,
619 Runic,
620 Samaritan,
621 Saurashtra,
622 Shavian,
623 Sinhala,
624 Sundanese,
625 Syloti_Nagri,
626 Syriac,
627 Tagalog,
628 Tagbanwa,
629 Tai_Le,
630 Tai_Tham,
631 Tai_Viet,
632 Tamil,
633 Telugu,
634 Thaana,
635 Thai,
636 Tibetan,
637 Tifinagh,
638 Ugaritic,
639 Vai,
640 Yi.
641 .P
642 Each character has exactly one Unicode general category property, specified by
643 a two-letter abbreviation. For compatibility with Perl, negation can be
644 specified by including a circumflex between the opening brace and the property
645 name. For example, \ep{^Lu} is the same as \eP{Lu}.
646 .P
647 If only one letter is specified with \ep or \eP, it includes all the general
648 category properties that start with that letter. In this case, in the absence
649 of negation, the curly brackets in the escape sequence are optional; these two
650 examples have the same effect:
651 .sp
652 \ep{L}
653 \epL
654 .sp
655 The following general category property codes are supported:
656 .sp
657 C Other
658 Cc Control
659 Cf Format
660 Cn Unassigned
661 Co Private use
662 Cs Surrogate
663 .sp
664 L Letter
665 Ll Lower case letter
666 Lm Modifier letter
667 Lo Other letter
668 Lt Title case letter
669 Lu Upper case letter
670 .sp
671 M Mark
672 Mc Spacing mark
673 Me Enclosing mark
674 Mn Non-spacing mark
675 .sp
676 N Number
677 Nd Decimal number
678 Nl Letter number
679 No Other number
680 .sp
681 P Punctuation
682 Pc Connector punctuation
683 Pd Dash punctuation
684 Pe Close punctuation
685 Pf Final punctuation
686 Pi Initial punctuation
687 Po Other punctuation
688 Ps Open punctuation
689 .sp
690 S Symbol
691 Sc Currency symbol
692 Sk Modifier symbol
693 Sm Mathematical symbol
694 So Other symbol
695 .sp
696 Z Separator
697 Zl Line separator
698 Zp Paragraph separator
699 Zs Space separator
700 .sp
701 The special property L& is also supported: it matches a character that has
702 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
703 a modifier or "other".
704 .P
705 The Cs (Surrogate) property applies only to characters in the range U+D800 to
706 U+DFFF. Such characters are not valid in UTF-8 strings (see RFC 3629) and so
707 cannot be tested by PCRE, unless UTF-8 validity checking has been turned off
708 (see the discussion of PCRE_NO_UTF8_CHECK in the
709 .\" HREF
710 \fBpcreapi\fP
711 .\"
712 page). Perl does not support the Cs property.
713 .P
714 The long synonyms for property names that Perl supports (such as \ep{Letter})
715 are not supported by PCRE, nor is it permitted to prefix any of these
716 properties with "Is".
717 .P
718 No character that is in the Unicode table has the Cn (unassigned) property.
719 Instead, this property is assumed for any code point that is not in the
720 Unicode table.
721 .P
722 Specifying caseless matching does not affect these escape sequences. For
723 example, \ep{Lu} always matches only upper case letters.
724 .P
725 The \eX escape matches any number of Unicode characters that form an extended
726 Unicode sequence. \eX is equivalent to
727 .sp
728 (?>\ePM\epM*)
729 .sp
730 That is, it matches a character without the "mark" property, followed by zero
731 or more characters with the "mark" property, and treats the sequence as an
732 atomic group
733 .\" HTML <a href="#atomicgroup">
734 .\" </a>
735 (see below).
736 .\"
737 Characters with the "mark" property are typically accents that affect the
738 preceding character. None of them have codepoints less than 256, so in
739 non-UTF-8 mode \eX matches any one character.
740 .P
741 Matching characters by Unicode property is not fast, because PCRE has to search
742 a structure that contains data for over fifteen thousand characters. That is
743 why the traditional escape sequences such as \ed and \ew do not use Unicode
744 properties in PCRE by default, though you can make them do so by setting the
745 PCRE_UCP option for \fBpcre_compile()\fP or by starting the pattern with
746 (*UCP).
747 .
748 .
749 .\" HTML <a name="extraprops"></a>
750 .SS PCRE's additional properties
751 .rs
752 .sp
753 As well as the standard Unicode properties described in the previous
754 section, PCRE supports four more that make it possible to convert traditional
755 escape sequences such as \ew and \es and POSIX character classes to use Unicode
756 properties. PCRE uses these non-standard, non-Perl properties internally when
757 PCRE_UCP is set. They are:
758 .sp
759 Xan Any alphanumeric character
760 Xps Any POSIX space character
761 Xsp Any Perl space character
762 Xwd Any Perl "word" character
763 .sp
764 Xan matches characters that have either the L (letter) or the N (number)
765 property. Xps matches the characters tab, linefeed, vertical tab, formfeed, or
766 carriage return, and any other character that has the Z (separator) property.
767 Xsp is the same as Xps, except that vertical tab is excluded. Xwd matches the
768 same characters as Xan, plus underscore.
769 .
770 .
771 .\" HTML <a name="resetmatchstart"></a>
772 .SS "Resetting the match start"
773 .rs
774 .sp
775 The escape sequence \eK, which is a Perl 5.10 feature, causes any previously
776 matched characters not to be included in the final matched sequence. For
777 example, the pattern:
778 .sp
779 foo\eKbar
780 .sp
781 matches "foobar", but reports that it has matched "bar". This feature is
782 similar to a lookbehind assertion
783 .\" HTML <a href="#lookbehind">
784 .\" </a>
785 (described below).
786 .\"
787 However, in this case, the part of the subject before the real match does not
788 have to be of fixed length, as lookbehind assertions do. The use of \eK does
789 not interfere with the setting of
790 .\" HTML <a href="#subpattern">
791 .\" </a>
792 captured substrings.
793 .\"
794 For example, when the pattern
795 .sp
796 (foo)\eKbar
797 .sp
798 matches "foobar", the first substring is still set to "foo".
799 .P
800 Perl documents that the use of \eK within assertions is "not well defined". In
801 PCRE, \eK is acted upon when it occurs inside positive assertions, but is
802 ignored in negative assertions.
803 .
804 .
805 .\" HTML <a name="smallassertions"></a>
806 .SS "Simple assertions"
807 .rs
808 .sp
809 The final use of backslash is for certain simple assertions. An assertion
810 specifies a condition that has to be met at a particular point in a match,
811 without consuming any characters from the subject string. The use of
812 subpatterns for more complicated assertions is described
813 .\" HTML <a href="#bigassertions">
814 .\" </a>
815 below.
816 .\"
817 The backslashed assertions are:
818 .sp
819 \eb matches at a word boundary
820 \eB matches when not at a word boundary
821 \eA matches at the start of the subject
822 \eZ matches at the end of the subject
823 also matches before a newline at the end of the subject
824 \ez matches only at the end of the subject
825 \eG matches at the first matching position in the subject
826 .sp
827 Inside a character class, \eb has a different meaning; it matches the backspace
828 character. If any other of these assertions appears in a character class, by
829 default it matches the corresponding literal character (for example, \eB
830 matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
831 escape sequence" error is generated instead.
832 .P
833 A word boundary is a position in the subject string where the current character
834 and the previous character do not both match \ew or \eW (i.e. one matches
835 \ew and the other matches \eW), or the start or end of the string if the
836 first or last character matches \ew, respectively. In UTF-8 mode, the meanings
837 of \ew and \eW can be changed by setting the PCRE_UCP option. When this is
838 done, it also affects \eb and \eB. Neither PCRE nor Perl has a separate "start
839 of word" or "end of word" metasequence. However, whatever follows \eb normally
840 determines which it is. For example, the fragment \eba matches "a" at the start
841 of a word.
842 .P
843 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
844 dollar (described in the next section) in that they only ever match at the very
845 start and end of the subject string, whatever options are set. Thus, they are
846 independent of multiline mode. These three assertions are not affected by the
847 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
848 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
849 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
850 at a point other than the beginning of the subject, \eA can never match. The
851 difference between \eZ and \ez is that \eZ matches before a newline at the end
852 of the string as well as at the very end, whereas \ez matches only at the end.
853 .P
854 The \eG assertion is true only when the current matching position is at the
855 start point of the match, as specified by the \fIstartoffset\fP argument of
856 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
857 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
858 arguments, you can mimic Perl's /g option, and it is in this kind of
859 implementation where \eG can be useful.
860 .P
861 Note, however, that PCRE's interpretation of \eG, as the start of the current
862 match, is subtly different from Perl's, which defines it as the end of the
863 previous match. In Perl, these can be different when the previously matched
864 string was empty. Because PCRE does just one match at a time, it cannot
865 reproduce this behaviour.
866 .P
867 If all the alternatives of a pattern begin with \eG, the expression is anchored
868 to the starting match position, and the "anchored" flag is set in the compiled
869 regular expression.
870 .
871 .
872 .SH "CIRCUMFLEX AND DOLLAR"
873 .rs
874 .sp
875 Outside a character class, in the default matching mode, the circumflex
876 character is an assertion that is true only if the current matching point is
877 at the start of the subject string. If the \fIstartoffset\fP argument of
878 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
879 option is unset. Inside a character class, circumflex has an entirely different
880 meaning
881 .\" HTML <a href="#characterclass">
882 .\" </a>
883 (see below).
884 .\"
885 .P
886 Circumflex need not be the first character of the pattern if a number of
887 alternatives are involved, but it should be the first thing in each alternative
888 in which it appears if the pattern is ever to match that branch. If all
889 possible alternatives start with a circumflex, that is, if the pattern is
890 constrained to match only at the start of the subject, it is said to be an
891 "anchored" pattern. (There are also other constructs that can cause a pattern
892 to be anchored.)
893 .P
894 A dollar character is an assertion that is true only if the current matching
895 point is at the end of the subject string, or immediately before a newline
896 at the end of the string (by default). Dollar need not be the last character of
897 the pattern if a number of alternatives are involved, but it should be the last
898 item in any branch in which it appears. Dollar has no special meaning in a
899 character class.
900 .P
901 The meaning of dollar can be changed so that it matches only at the very end of
902 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
903 does not affect the \eZ assertion.
904 .P
905 The meanings of the circumflex and dollar characters are changed if the
906 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
907 immediately after internal newlines as well as at the start of the subject
908 string. It does not match after a newline that ends the string. A dollar
909 matches before any newlines in the string, as well as at the very end, when
910 PCRE_MULTILINE is set. When newline is specified as the two-character
911 sequence CRLF, isolated CR and LF characters do not indicate newlines.
912 .P
913 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
914 \en represents a newline) in multiline mode, but not otherwise. Consequently,
915 patterns that are anchored in single line mode because all branches start with
916 ^ are not anchored in multiline mode, and a match for circumflex is possible
917 when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
918 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
919 .P
920 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
921 end of the subject in both modes, and if all branches of a pattern start with
922 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
923 .
924 .
925 .\" HTML <a name="fullstopdot"></a>
926 .SH "FULL STOP (PERIOD, DOT) AND \eN"
927 .rs
928 .sp
929 Outside a character class, a dot in the pattern matches any one character in
930 the subject string except (by default) a character that signifies the end of a
931 line. In UTF-8 mode, the matched character may be more than one byte long.
932 .P
933 When a line ending is defined as a single character, dot never matches that
934 character; when the two-character sequence CRLF is used, dot does not match CR
935 if it is immediately followed by LF, but otherwise it matches all characters
936 (including isolated CRs and LFs). When any Unicode line endings are being
937 recognized, dot does not match CR or LF or any of the other line ending
938 characters.
939 .P
940 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
941 option is set, a dot matches any one character, without exception. If the
942 two-character sequence CRLF is present in the subject string, it takes two dots
943 to match it.
944 .P
945 The handling of dot is entirely independent of the handling of circumflex and
946 dollar, the only relationship being that they both involve newlines. Dot has no
947 special meaning in a character class.
948 .P
949 The escape sequence \eN always behaves as a dot does when PCRE_DOTALL is not
950 set. In other words, it matches any one character except one that signifies the
951 end of a line.
952 .
953 .
954 .SH "MATCHING A SINGLE BYTE"
955 .rs
956 .sp
957 Outside a character class, the escape sequence \eC matches any one byte, both
958 in and out of UTF-8 mode. Unlike a dot, it always matches any line-ending
959 characters. The feature is provided in Perl in order to match individual bytes
960 in UTF-8 mode. Because it breaks up UTF-8 characters into individual bytes,
961 what remains in the string may be a malformed UTF-8 string. For this reason,
962 the \eC escape sequence is best avoided.
963 .P
964 PCRE does not allow \eC to appear in lookbehind assertions
965 .\" HTML <a href="#lookbehind">
966 .\" </a>
967 (described below),
968 .\"
969 because in UTF-8 mode this would make it impossible to calculate the length of
970 the lookbehind.
971 .
972 .
973 .\" HTML <a name="characterclass"></a>
974 .SH "SQUARE BRACKETS AND CHARACTER CLASSES"
975 .rs
976 .sp
977 An opening square bracket introduces a character class, terminated by a closing
978 square bracket. A closing square bracket on its own is not special by default.
979 However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
980 bracket causes a compile-time error. If a closing square bracket is required as
981 a member of the class, it should be the first data character in the class
982 (after an initial circumflex, if present) or escaped with a backslash.
983 .P
984 A character class matches a single character in the subject. In UTF-8 mode, the
985 character may be more than one byte long. A matched character must be in the
986 set of characters defined by the class, unless the first character in the class
987 definition is a circumflex, in which case the subject character must not be in
988 the set defined by the class. If a circumflex is actually required as a member
989 of the class, ensure it is not the first character, or escape it with a
990 backslash.
991 .P
992 For example, the character class [aeiou] matches any lower case vowel, while
993 [^aeiou] matches any character that is not a lower case vowel. Note that a
994 circumflex is just a convenient notation for specifying the characters that
995 are in the class by enumerating those that are not. A class that starts with a
996 circumflex is not an assertion; it still consumes a character from the subject
997 string, and therefore it fails if the current pointer is at the end of the
998 string.
999 .P
1000 In UTF-8 mode, characters with values greater than 255 can be included in a
1001 class as a literal string of bytes, or by using the \ex{ escaping mechanism.
1002 .P
1003 When caseless matching is set, any letters in a class represent both their
1004 upper case and lower case versions, so for example, a caseless [aeiou] matches
1005 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1006 caseful version would. In UTF-8 mode, PCRE always understands the concept of
1007 case for characters whose values are less than 128, so caseless matching is
1008 always possible. For characters with higher values, the concept of case is
1009 supported if PCRE is compiled with Unicode property support, but not otherwise.
1010 If you want to use caseless matching in UTF8-mode for characters 128 and above,
1011 you must ensure that PCRE is compiled with Unicode property support as well as
1012 with UTF-8 support.
1013 .P
1014 Characters that might indicate line breaks are never treated in any special way
1015 when matching character classes, whatever line-ending sequence is in use, and
1016 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
1017 such as [^a] always matches one of these characters.
1018 .P
1019 The minus (hyphen) character can be used to specify a range of characters in a
1020 character class. For example, [d-m] matches any letter between d and m,
1021 inclusive. If a minus character is required in a class, it must be escaped with
1022 a backslash or appear in a position where it cannot be interpreted as
1023 indicating a range, typically as the first or last character in the class.
1024 .P
1025 It is not possible to have the literal character "]" as the end character of a
1026 range. A pattern such as [W-]46] is interpreted as a class of two characters
1027 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1028 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1029 the end of range, so [W-\e]46] is interpreted as a class containing a range
1030 followed by two other characters. The octal or hexadecimal representation of
1031 "]" can also be used to end a range.
1032 .P
1033 Ranges operate in the collating sequence of character values. They can also be
1034 used for characters specified numerically, for example [\e000-\e037]. In UTF-8
1035 mode, ranges can include characters whose values are greater than 255, for
1036 example [\ex{100}-\ex{2ff}].
1037 .P
1038 If a range that includes letters is used when caseless matching is set, it
1039 matches the letters in either case. For example, [W-c] is equivalent to
1040 [][\e\e^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
1041 tables for a French locale are in use, [\exc8-\excb] matches accented E
1042 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
1043 characters with values greater than 128 only when it is compiled with Unicode
1044 property support.
1045 .P
1046 The character types \ed, \eD, \eh, \eH, \ep, \eP, \es, \eS, \ev, \eV, \ew, and
1047 \eW may also appear in a character class, and add the characters that they
1048 match to the class. For example, [\edABCDEF] matches any hexadecimal digit. A
1049 circumflex can conveniently be used with the upper case character types to
1050 specify a more restricted set of characters than the matching lower case type.
1051 For example, the class [^\eW_] matches any letter or digit, but not underscore.
1052 .P
1053 The only metacharacters that are recognized in character classes are backslash,
1054 hyphen (only where it can be interpreted as specifying a range), circumflex
1055 (only at the start), opening square bracket (only when it can be interpreted as
1056 introducing a POSIX class name - see the next section), and the terminating
1057 closing square bracket. However, escaping other non-alphanumeric characters
1058 does no harm.
1059 .
1060 .
1061 .SH "POSIX CHARACTER CLASSES"
1062 .rs
1063 .sp
1064 Perl supports the POSIX notation for character classes. This uses names
1065 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1066 this notation. For example,
1067 .sp
1068 [01[:alpha:]%]
1069 .sp
1070 matches "0", "1", any alphabetic character, or "%". The supported class names
1071 are:
1072 .sp
1073 alnum letters and digits
1074 alpha letters
1075 ascii character codes 0 - 127
1076 blank space or tab only
1077 cntrl control characters
1078 digit decimal digits (same as \ed)
1079 graph printing characters, excluding space
1080 lower lower case letters
1081 print printing characters, including space
1082 punct printing characters, excluding letters and digits and space
1083 space white space (not quite the same as \es)
1084 upper upper case letters
1085 word "word" characters (same as \ew)
1086 xdigit hexadecimal digits
1087 .sp
1088 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1089 space (32). Notice that this list includes the VT character (code 11). This
1090 makes "space" different to \es, which does not include VT (for Perl
1091 compatibility).
1092 .P
1093 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1094 5.8. Another Perl extension is negation, which is indicated by a ^ character
1095 after the colon. For example,
1096 .sp
1097 [12[:^digit:]]
1098 .sp
1099 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1100 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1101 supported, and an error is given if they are encountered.
1102 .P
1103 By default, in UTF-8 mode, characters with values greater than 128 do not match
1104 any of the POSIX character classes. However, if the PCRE_UCP option is passed
1105 to \fBpcre_compile()\fP, some of the classes are changed so that Unicode
1106 character properties are used. This is achieved by replacing the POSIX classes
1107 by other sequences, as follows:
1108 .sp
1109 [:alnum:] becomes \ep{Xan}
1110 [:alpha:] becomes \ep{L}
1111 [:blank:] becomes \eh
1112 [:digit:] becomes \ep{Nd}
1113 [:lower:] becomes \ep{Ll}
1114 [:space:] becomes \ep{Xps}
1115 [:upper:] becomes \ep{Lu}
1116 [:word:] becomes \ep{Xwd}
1117 .sp
1118 Negated versions, such as [:^alpha:] use \eP instead of \ep. The other POSIX
1119 classes are unchanged, and match only characters with code points less than
1120 128.
1121 .
1122 .
1123 .SH "VERTICAL BAR"
1124 .rs
1125 .sp
1126 Vertical bar characters are used to separate alternative patterns. For example,
1127 the pattern
1128 .sp
1129 gilbert|sullivan
1130 .sp
1131 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1132 and an empty alternative is permitted (matching the empty string). The matching
1133 process tries each alternative in turn, from left to right, and the first one
1134 that succeeds is used. If the alternatives are within a subpattern
1135 .\" HTML <a href="#subpattern">
1136 .\" </a>
1137 (defined below),
1138 .\"
1139 "succeeds" means matching the rest of the main pattern as well as the
1140 alternative in the subpattern.
1141 .
1142 .
1143 .SH "INTERNAL OPTION SETTING"
1144 .rs
1145 .sp
1146 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1147 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1148 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1149 The option letters are
1150 .sp
1151 i for PCRE_CASELESS
1152 m for PCRE_MULTILINE
1153 s for PCRE_DOTALL
1154 x for PCRE_EXTENDED
1155 .sp
1156 For example, (?im) sets caseless, multiline matching. It is also possible to
1157 unset these options by preceding the letter with a hyphen, and a combined
1158 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1159 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1160 permitted. If a letter appears both before and after the hyphen, the option is
1161 unset.
1162 .P
1163 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1164 changed in the same way as the Perl-compatible options by using the characters
1165 J, U and X respectively.
1166 .P
1167 When one of these option changes occurs at top level (that is, not inside
1168 subpattern parentheses), the change applies to the remainder of the pattern
1169 that follows. If the change is placed right at the start of a pattern, PCRE
1170 extracts it into the global options (and it will therefore show up in data
1171 extracted by the \fBpcre_fullinfo()\fP function).
1172 .P
1173 An option change within a subpattern (see below for a description of
1174 subpatterns) affects only that part of the current pattern that follows it, so
1175 .sp
1176 (a(?i)b)c
1177 .sp
1178 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1179 By this means, options can be made to have different settings in different
1180 parts of the pattern. Any changes made in one alternative do carry on
1181 into subsequent branches within the same subpattern. For example,
1182 .sp
1183 (a(?i)b|c)
1184 .sp
1185 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1186 branch is abandoned before the option setting. This is because the effects of
1187 option settings happen at compile time. There would be some very weird
1188 behaviour otherwise.
1189 .P
1190 \fBNote:\fP There are other PCRE-specific options that can be set by the
1191 application when the compile or match functions are called. In some cases the
1192 pattern can contain special leading sequences such as (*CRLF) to override what
1193 the application has set or what has been defaulted. Details are given in the
1194 section entitled
1195 .\" HTML <a href="#newlineseq">
1196 .\" </a>
1197 "Newline sequences"
1198 .\"
1199 above. There are also the (*UTF8) and (*UCP) leading sequences that can be used
1200 to set UTF-8 and Unicode property modes; they are equivalent to setting the
1201 PCRE_UTF8 and the PCRE_UCP options, respectively.
1202 .
1203 .
1204 .\" HTML <a name="subpattern"></a>
1205 .SH SUBPATTERNS
1206 .rs
1207 .sp
1208 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1209 Turning part of a pattern into a subpattern does two things:
1210 .sp
1211 1. It localizes a set of alternatives. For example, the pattern
1212 .sp
1213 cat(aract|erpillar|)
1214 .sp
1215 matches one of the words "cat", "cataract", or "caterpillar". Without the
1216 parentheses, it would match "cataract", "erpillar" or an empty string.
1217 .sp
1218 2. It sets up the subpattern as a capturing subpattern. This means that, when
1219 the whole pattern matches, that portion of the subject string that matched the
1220 subpattern is passed back to the caller via the \fIovector\fP argument of
1221 \fBpcre_exec()\fP. Opening parentheses are counted from left to right (starting
1222 from 1) to obtain numbers for the capturing subpatterns.
1223 .P
1224 For example, if the string "the red king" is matched against the pattern
1225 .sp
1226 the ((red|white) (king|queen))
1227 .sp
1228 the captured substrings are "red king", "red", and "king", and are numbered 1,
1229 2, and 3, respectively.
1230 .P
1231 The fact that plain parentheses fulfil two functions is not always helpful.
1232 There are often times when a grouping subpattern is required without a
1233 capturing requirement. If an opening parenthesis is followed by a question mark
1234 and a colon, the subpattern does not do any capturing, and is not counted when
1235 computing the number of any subsequent capturing subpatterns. For example, if
1236 the string "the white queen" is matched against the pattern
1237 .sp
1238 the ((?:red|white) (king|queen))
1239 .sp
1240 the captured substrings are "white queen" and "queen", and are numbered 1 and
1241 2. The maximum number of capturing subpatterns is 65535.
1242 .P
1243 As a convenient shorthand, if any option settings are required at the start of
1244 a non-capturing subpattern, the option letters may appear between the "?" and
1245 the ":". Thus the two patterns
1246 .sp
1247 (?i:saturday|sunday)
1248 (?:(?i)saturday|sunday)
1249 .sp
1250 match exactly the same set of strings. Because alternative branches are tried
1251 from left to right, and options are not reset until the end of the subpattern
1252 is reached, an option setting in one branch does affect subsequent branches, so
1253 the above patterns match "SUNDAY" as well as "Saturday".
1254 .
1255 .
1256 .\" HTML <a name="dupsubpatternnumber"></a>
1257 .SH "DUPLICATE SUBPATTERN NUMBERS"
1258 .rs
1259 .sp
1260 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1261 the same numbers for its capturing parentheses. Such a subpattern starts with
1262 (?| and is itself a non-capturing subpattern. For example, consider this
1263 pattern:
1264 .sp
1265 (?|(Sat)ur|(Sun))day
1266 .sp
1267 Because the two alternatives are inside a (?| group, both sets of capturing
1268 parentheses are numbered one. Thus, when the pattern matches, you can look
1269 at captured substring number one, whichever alternative matched. This construct
1270 is useful when you want to capture part, but not all, of one of a number of
1271 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1272 number is reset at the start of each branch. The numbers of any capturing
1273 buffers that follow the subpattern start after the highest number used in any
1274 branch. The following example is taken from the Perl documentation.
1275 The numbers underneath show in which buffer the captured content will be
1276 stored.
1277 .sp
1278 # before ---------------branch-reset----------- after
1279 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1280 # 1 2 2 3 2 3 4
1281 .sp
1282 A back reference to a numbered subpattern uses the most recent value that is
1283 set for that number by any subpattern. The following pattern matches "abcabc"
1284 or "defdef":
1285 .sp
1286 /(?|(abc)|(def))\e1/
1287 .sp
1288 In contrast, a recursive or "subroutine" call to a numbered subpattern always
1289 refers to the first one in the pattern with the given number. The following
1290 pattern matches "abcabc" or "defabc":
1291 .sp
1292 /(?|(abc)|(def))(?1)/
1293 .sp
1294 If a
1295 .\" HTML <a href="#conditions">
1296 .\" </a>
1297 condition test
1298 .\"
1299 for a subpattern's having matched refers to a non-unique number, the test is
1300 true if any of the subpatterns of that number have matched.
1301 .P
1302 An alternative approach to using this "branch reset" feature is to use
1303 duplicate named subpatterns, as described in the next section.
1304 .
1305 .
1306 .SH "NAMED SUBPATTERNS"
1307 .rs
1308 .sp
1309 Identifying capturing parentheses by number is simple, but it can be very hard
1310 to keep track of the numbers in complicated regular expressions. Furthermore,
1311 if an expression is modified, the numbers may change. To help with this
1312 difficulty, PCRE supports the naming of subpatterns. This feature was not
1313 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1314 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1315 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1316 have different names, but PCRE does not.
1317 .P
1318 In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1319 (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1320 parentheses from other parts of the pattern, such as
1321 .\" HTML <a href="#backreferences">
1322 .\" </a>
1323 back references,
1324 .\"
1325 .\" HTML <a href="#recursion">
1326 .\" </a>
1327 recursion,
1328 .\"
1329 and
1330 .\" HTML <a href="#conditions">
1331 .\" </a>
1332 conditions,
1333 .\"
1334 can be made by name as well as by number.
1335 .P
1336 Names consist of up to 32 alphanumeric characters and underscores. Named
1337 capturing parentheses are still allocated numbers as well as names, exactly as
1338 if the names were not present. The PCRE API provides function calls for
1339 extracting the name-to-number translation table from a compiled pattern. There
1340 is also a convenience function for extracting a captured substring by name.
1341 .P
1342 By default, a name must be unique within a pattern, but it is possible to relax
1343 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1344 names are also always permitted for subpatterns with the same number, set up as
1345 described in the previous section.) Duplicate names can be useful for patterns
1346 where only one instance of the named parentheses can match. Suppose you want to
1347 match the name of a weekday, either as a 3-letter abbreviation or as the full
1348 name, and in both cases you want to extract the abbreviation. This pattern
1349 (ignoring the line breaks) does the job:
1350 .sp
1351 (?<DN>Mon|Fri|Sun)(?:day)?|
1352 (?<DN>Tue)(?:sday)?|
1353 (?<DN>Wed)(?:nesday)?|
1354 (?<DN>Thu)(?:rsday)?|
1355 (?<DN>Sat)(?:urday)?
1356 .sp
1357 There are five capturing substrings, but only one is ever set after a match.
1358 (An alternative way of solving this problem is to use a "branch reset"
1359 subpattern, as described in the previous section.)
1360 .P
1361 The convenience function for extracting the data by name returns the substring
1362 for the first (and in this example, the only) subpattern of that name that
1363 matched. This saves searching to find which numbered subpattern it was.
1364 .P
1365 If you make a back reference to a non-unique named subpattern from elsewhere in
1366 the pattern, the one that corresponds to the first occurrence of the name is
1367 used. In the absence of duplicate numbers (see the previous section) this is
1368 the one with the lowest number. If you use a named reference in a condition
1369 test (see the
1370 .\"
1371 .\" HTML <a href="#conditions">
1372 .\" </a>
1373 section about conditions
1374 .\"
1375 below), either to check whether a subpattern has matched, or to check for
1376 recursion, all subpatterns with the same name are tested. If the condition is
1377 true for any one of them, the overall condition is true. This is the same
1378 behaviour as testing by number. For further details of the interfaces for
1379 handling named subpatterns, see the
1380 .\" HREF
1381 \fBpcreapi\fP
1382 .\"
1383 documentation.
1384 .P
1385 \fBWarning:\fP You cannot use different names to distinguish between two
1386 subpatterns with the same number because PCRE uses only the numbers when
1387 matching. For this reason, an error is given at compile time if different names
1388 are given to subpatterns with the same number. However, you can give the same
1389 name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
1390 .
1391 .
1392 .SH REPETITION
1393 .rs
1394 .sp
1395 Repetition is specified by quantifiers, which can follow any of the following
1396 items:
1397 .sp
1398 a literal data character
1399 the dot metacharacter
1400 the \eC escape sequence
1401 the \eX escape sequence (in UTF-8 mode with Unicode properties)
1402 the \eR escape sequence
1403 an escape such as \ed that matches a single character
1404 a character class
1405 a back reference (see next section)
1406 a parenthesized subpattern (unless it is an assertion)
1407 a recursive or "subroutine" call to a subpattern
1408 .sp
1409 The general repetition quantifier specifies a minimum and maximum number of
1410 permitted matches, by giving the two numbers in curly brackets (braces),
1411 separated by a comma. The numbers must be less than 65536, and the first must
1412 be less than or equal to the second. For example:
1413 .sp
1414 z{2,4}
1415 .sp
1416 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1417 character. If the second number is omitted, but the comma is present, there is
1418 no upper limit; if the second number and the comma are both omitted, the
1419 quantifier specifies an exact number of required matches. Thus
1420 .sp
1421 [aeiou]{3,}
1422 .sp
1423 matches at least 3 successive vowels, but may match many more, while
1424 .sp
1425 \ed{8}
1426 .sp
1427 matches exactly 8 digits. An opening curly bracket that appears in a position
1428 where a quantifier is not allowed, or one that does not match the syntax of a
1429 quantifier, is taken as a literal character. For example, {,6} is not a
1430 quantifier, but a literal string of four characters.
1431 .P
1432 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1433 bytes. Thus, for example, \ex{100}{2} matches two UTF-8 characters, each of
1434 which is represented by a two-byte sequence. Similarly, when Unicode property
1435 support is available, \eX{3} matches three Unicode extended sequences, each of
1436 which may be several bytes long (and they may be of different lengths).
1437 .P
1438 The quantifier {0} is permitted, causing the expression to behave as if the
1439 previous item and the quantifier were not present. This may be useful for
1440 subpatterns that are referenced as
1441 .\" HTML <a href="#subpatternsassubroutines">
1442 .\" </a>
1443 subroutines
1444 .\"
1445 from elsewhere in the pattern. Items other than subpatterns that have a {0}
1446 quantifier are omitted from the compiled pattern.
1447 .P
1448 For convenience, the three most common quantifiers have single-character
1449 abbreviations:
1450 .sp
1451 * is equivalent to {0,}
1452 + is equivalent to {1,}
1453 ? is equivalent to {0,1}
1454 .sp
1455 It is possible to construct infinite loops by following a subpattern that can
1456 match no characters with a quantifier that has no upper limit, for example:
1457 .sp
1458 (a?)*
1459 .sp
1460 Earlier versions of Perl and PCRE used to give an error at compile time for
1461 such patterns. However, because there are cases where this can be useful, such
1462 patterns are now accepted, but if any repetition of the subpattern does in fact
1463 match no characters, the loop is forcibly broken.
1464 .P
1465 By default, the quantifiers are "greedy", that is, they match as much as
1466 possible (up to the maximum number of permitted times), without causing the
1467 rest of the pattern to fail. The classic example of where this gives problems
1468 is in trying to match comments in C programs. These appear between /* and */
1469 and within the comment, individual * and / characters may appear. An attempt to
1470 match C comments by applying the pattern
1471 .sp
1472 /\e*.*\e*/
1473 .sp
1474 to the string
1475 .sp
1476 /* first comment */ not comment /* second comment */
1477 .sp
1478 fails, because it matches the entire string owing to the greediness of the .*
1479 item.
1480 .P
1481 However, if a quantifier is followed by a question mark, it ceases to be
1482 greedy, and instead matches the minimum number of times possible, so the
1483 pattern
1484 .sp
1485 /\e*.*?\e*/
1486 .sp
1487 does the right thing with the C comments. The meaning of the various
1488 quantifiers is not otherwise changed, just the preferred number of matches.
1489 Do not confuse this use of question mark with its use as a quantifier in its
1490 own right. Because it has two uses, it can sometimes appear doubled, as in
1491 .sp
1492 \ed??\ed
1493 .sp
1494 which matches one digit by preference, but can match two if that is the only
1495 way the rest of the pattern matches.
1496 .P
1497 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1498 the quantifiers are not greedy by default, but individual ones can be made
1499 greedy by following them with a question mark. In other words, it inverts the
1500 default behaviour.
1501 .P
1502 When a parenthesized subpattern is quantified with a minimum repeat count that
1503 is greater than 1 or with a limited maximum, more memory is required for the
1504 compiled pattern, in proportion to the size of the minimum or maximum.
1505 .P
1506 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1507 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1508 implicitly anchored, because whatever follows will be tried against every
1509 character position in the subject string, so there is no point in retrying the
1510 overall match at any position after the first. PCRE normally treats such a
1511 pattern as though it were preceded by \eA.
1512 .P
1513 In cases where it is known that the subject string contains no newlines, it is
1514 worth setting PCRE_DOTALL in order to obtain this optimization, or
1515 alternatively using ^ to indicate anchoring explicitly.
1516 .P
1517 However, there is one situation where the optimization cannot be used. When .*
1518 is inside capturing parentheses that are the subject of a back reference
1519 elsewhere in the pattern, a match at the start may fail where a later one
1520 succeeds. Consider, for example:
1521 .sp
1522 (.*)abc\e1
1523 .sp
1524 If the subject is "xyz123abc123" the match point is the fourth character. For
1525 this reason, such a pattern is not implicitly anchored.
1526 .P
1527 When a capturing subpattern is repeated, the value captured is the substring
1528 that matched the final iteration. For example, after
1529 .sp
1530 (tweedle[dume]{3}\es*)+
1531 .sp
1532 has matched "tweedledum tweedledee" the value of the captured substring is
1533 "tweedledee". However, if there are nested capturing subpatterns, the
1534 corresponding captured values may have been set in previous iterations. For
1535 example, after
1536 .sp
1537 /(a|(b))+/
1538 .sp
1539 matches "aba" the value of the second captured substring is "b".
1540 .
1541 .
1542 .\" HTML <a name="atomicgroup"></a>
1543 .SH "ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS"
1544 .rs
1545 .sp
1546 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1547 repetition, failure of what follows normally causes the repeated item to be
1548 re-evaluated to see if a different number of repeats allows the rest of the
1549 pattern to match. Sometimes it is useful to prevent this, either to change the
1550 nature of the match, or to cause it fail earlier than it otherwise might, when
1551 the author of the pattern knows there is no point in carrying on.
1552 .P
1553 Consider, for example, the pattern \ed+foo when applied to the subject line
1554 .sp
1555 123456bar
1556 .sp
1557 After matching all 6 digits and then failing to match "foo", the normal
1558 action of the matcher is to try again with only 5 digits matching the \ed+
1559 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1560 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1561 that once a subpattern has matched, it is not to be re-evaluated in this way.
1562 .P
1563 If we use atomic grouping for the previous example, the matcher gives up
1564 immediately on failing to match "foo" the first time. The notation is a kind of
1565 special parenthesis, starting with (?> as in this example:
1566 .sp
1567 (?>\ed+)foo
1568 .sp
1569 This kind of parenthesis "locks up" the part of the pattern it contains once
1570 it has matched, and a failure further into the pattern is prevented from
1571 backtracking into it. Backtracking past it to previous items, however, works as
1572 normal.
1573 .P
1574 An alternative description is that a subpattern of this type matches the string
1575 of characters that an identical standalone pattern would match, if anchored at
1576 the current point in the subject string.
1577 .P
1578 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1579 the above example can be thought of as a maximizing repeat that must swallow
1580 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1581 number of digits they match in order to make the rest of the pattern match,
1582 (?>\ed+) can only match an entire sequence of digits.
1583 .P
1584 Atomic groups in general can of course contain arbitrarily complicated
1585 subpatterns, and can be nested. However, when the subpattern for an atomic
1586 group is just a single repeated item, as in the example above, a simpler
1587 notation, called a "possessive quantifier" can be used. This consists of an
1588 additional + character following a quantifier. Using this notation, the
1589 previous example can be rewritten as
1590 .sp
1591 \ed++foo
1592 .sp
1593 Note that a possessive quantifier can be used with an entire group, for
1594 example:
1595 .sp
1596 (abc|xyz){2,3}+
1597 .sp
1598 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1599 option is ignored. They are a convenient notation for the simpler forms of
1600 atomic group. However, there is no difference in the meaning of a possessive
1601 quantifier and the equivalent atomic group, though there may be a performance
1602 difference; possessive quantifiers should be slightly faster.
1603 .P
1604 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1605 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1606 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1607 package, and PCRE copied it from there. It ultimately found its way into Perl
1608 at release 5.10.
1609 .P
1610 PCRE has an optimization that automatically "possessifies" certain simple
1611 pattern constructs. For example, the sequence A+B is treated as A++B because
1612 there is no point in backtracking into a sequence of A's when B must follow.
1613 .P
1614 When a pattern contains an unlimited repeat inside a subpattern that can itself
1615 be repeated an unlimited number of times, the use of an atomic group is the
1616 only way to avoid some failing matches taking a very long time indeed. The
1617 pattern
1618 .sp
1619 (\eD+|<\ed+>)*[!?]
1620 .sp
1621 matches an unlimited number of substrings that either consist of non-digits, or
1622 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1623 quickly. However, if it is applied to
1624 .sp
1625 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1626 .sp
1627 it takes a long time before reporting failure. This is because the string can
1628 be divided between the internal \eD+ repeat and the external * repeat in a
1629 large number of ways, and all have to be tried. (The example uses [!?] rather
1630 than a single character at the end, because both PCRE and Perl have an
1631 optimization that allows for fast failure when a single character is used. They
1632 remember the last single character that is required for a match, and fail early
1633 if it is not present in the string.) If the pattern is changed so that it uses
1634 an atomic group, like this:
1635 .sp
1636 ((?>\eD+)|<\ed+>)*[!?]
1637 .sp
1638 sequences of non-digits cannot be broken, and failure happens quickly.
1639 .
1640 .
1641 .\" HTML <a name="backreferences"></a>
1642 .SH "BACK REFERENCES"
1643 .rs
1644 .sp
1645 Outside a character class, a backslash followed by a digit greater than 0 (and
1646 possibly further digits) is a back reference to a capturing subpattern earlier
1647 (that is, to its left) in the pattern, provided there have been that many
1648 previous capturing left parentheses.
1649 .P
1650 However, if the decimal number following the backslash is less than 10, it is
1651 always taken as a back reference, and causes an error only if there are not
1652 that many capturing left parentheses in the entire pattern. In other words, the
1653 parentheses that are referenced need not be to the left of the reference for
1654 numbers less than 10. A "forward back reference" of this type can make sense
1655 when a repetition is involved and the subpattern to the right has participated
1656 in an earlier iteration.
1657 .P
1658 It is not possible to have a numerical "forward back reference" to a subpattern
1659 whose number is 10 or more using this syntax because a sequence such as \e50 is
1660 interpreted as a character defined in octal. See the subsection entitled
1661 "Non-printing characters"
1662 .\" HTML <a href="#digitsafterbackslash">
1663 .\" </a>
1664 above
1665 .\"
1666 for further details of the handling of digits following a backslash. There is
1667 no such problem when named parentheses are used. A back reference to any
1668 subpattern is possible using named parentheses (see below).
1669 .P
1670 Another way of avoiding the ambiguity inherent in the use of digits following a
1671 backslash is to use the \eg escape sequence, which is a feature introduced in
1672 Perl 5.10. This escape must be followed by an unsigned number or a negative
1673 number, optionally enclosed in braces. These examples are all identical:
1674 .sp
1675 (ring), \e1
1676 (ring), \eg1
1677 (ring), \eg{1}
1678 .sp
1679 An unsigned number specifies an absolute reference without the ambiguity that
1680 is present in the older syntax. It is also useful when literal digits follow
1681 the reference. A negative number is a relative reference. Consider this
1682 example:
1683 .sp
1684 (abc(def)ghi)\eg{-1}
1685 .sp
1686 The sequence \eg{-1} is a reference to the most recently started capturing
1687 subpattern before \eg, that is, is it equivalent to \e2. Similarly, \eg{-2}
1688 would be equivalent to \e1. The use of relative references can be helpful in
1689 long patterns, and also in patterns that are created by joining together
1690 fragments that contain references within themselves.
1691 .P
1692 A back reference matches whatever actually matched the capturing subpattern in
1693 the current subject string, rather than anything matching the subpattern
1694 itself (see
1695 .\" HTML <a href="#subpatternsassubroutines">
1696 .\" </a>
1697 "Subpatterns as subroutines"
1698 .\"
1699 below for a way of doing that). So the pattern
1700 .sp
1701 (sens|respons)e and \e1ibility
1702 .sp
1703 matches "sense and sensibility" and "response and responsibility", but not
1704 "sense and responsibility". If caseful matching is in force at the time of the
1705 back reference, the case of letters is relevant. For example,
1706 .sp
1707 ((?i)rah)\es+\e1
1708 .sp
1709 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1710 capturing subpattern is matched caselessly.
1711 .P
1712 There are several different ways of writing back references to named
1713 subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1714 \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1715 back reference syntax, in which \eg can be used for both numeric and named
1716 references, is also supported. We could rewrite the above example in any of
1717 the following ways:
1718 .sp
1719 (?<p1>(?i)rah)\es+\ek<p1>
1720 (?'p1'(?i)rah)\es+\ek{p1}
1721 (?P<p1>(?i)rah)\es+(?P=p1)
1722 (?<p1>(?i)rah)\es+\eg{p1}
1723 .sp
1724 A subpattern that is referenced by name may appear in the pattern before or
1725 after the reference.
1726 .P
1727 There may be more than one back reference to the same subpattern. If a
1728 subpattern has not actually been used in a particular match, any back
1729 references to it always fail by default. For example, the pattern
1730 .sp
1731 (a|(bc))\e2
1732 .sp
1733 always fails if it starts to match "a" rather than "bc". However, if the
1734 PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1735 unset value matches an empty string.
1736 .P
1737 Because there may be many capturing parentheses in a pattern, all digits
1738 following a backslash are taken as part of a potential back reference number.
1739 If the pattern continues with a digit character, some delimiter must be used to
1740 terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1741 whitespace. Otherwise, the \eg{ syntax or an empty comment (see
1742 .\" HTML <a href="#comments">
1743 .\" </a>
1744 "Comments"
1745 .\"
1746 below) can be used.
1747 .
1748 .SS "Recursive back references"
1749 .rs
1750 .sp
1751 A back reference that occurs inside the parentheses to which it refers fails
1752 when the subpattern is first used, so, for example, (a\e1) never matches.
1753 However, such references can be useful inside repeated subpatterns. For
1754 example, the pattern
1755 .sp
1756 (a|b\e1)+
1757 .sp
1758 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1759 the subpattern, the back reference matches the character string corresponding
1760 to the previous iteration. In order for this to work, the pattern must be such
1761 that the first iteration does not need to match the back reference. This can be
1762 done using alternation, as in the example above, or by a quantifier with a
1763 minimum of zero.
1764 .P
1765 Back references of this type cause the group that they reference to be treated
1766 as an
1767 .\" HTML <a href="#atomicgroup">
1768 .\" </a>
1769 atomic group.
1770 .\"
1771 Once the whole group has been matched, a subsequent matching failure cannot
1772 cause backtracking into the middle of the group.
1773 .
1774 .
1775 .\" HTML <a name="bigassertions"></a>
1776 .SH ASSERTIONS
1777 .rs
1778 .sp
1779 An assertion is a test on the characters following or preceding the current
1780 matching point that does not actually consume any characters. The simple
1781 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1782 .\" HTML <a href="#smallassertions">
1783 .\" </a>
1784 above.
1785 .\"
1786 .P
1787 More complicated assertions are coded as subpatterns. There are two kinds:
1788 those that look ahead of the current position in the subject string, and those
1789 that look behind it. An assertion subpattern is matched in the normal way,
1790 except that it does not cause the current matching position to be changed.
1791 .P
1792 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1793 because it makes no sense to assert the same thing several times. If any kind
1794 of assertion contains capturing subpatterns within it, these are counted for
1795 the purposes of numbering the capturing subpatterns in the whole pattern.
1796 However, substring capturing is carried out only for positive assertions,
1797 because it does not make sense for negative assertions.
1798 .
1799 .
1800 .SS "Lookahead assertions"
1801 .rs
1802 .sp
1803 Lookahead assertions start with (?= for positive assertions and (?! for
1804 negative assertions. For example,
1805 .sp
1806 \ew+(?=;)
1807 .sp
1808 matches a word followed by a semicolon, but does not include the semicolon in
1809 the match, and
1810 .sp
1811 foo(?!bar)
1812 .sp
1813 matches any occurrence of "foo" that is not followed by "bar". Note that the
1814 apparently similar pattern
1815 .sp
1816 (?!foo)bar
1817 .sp
1818 does not find an occurrence of "bar" that is preceded by something other than
1819 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1820 (?!foo) is always true when the next three characters are "bar". A
1821 lookbehind assertion is needed to achieve the other effect.
1822 .P
1823 If you want to force a matching failure at some point in a pattern, the most
1824 convenient way to do it is with (?!) because an empty string always matches, so
1825 an assertion that requires there not to be an empty string must always fail.
1826 The Perl 5.10 backtracking control verb (*FAIL) or (*F) is essentially a
1827 synonym for (?!).
1828 .
1829 .
1830 .\" HTML <a name="lookbehind"></a>
1831 .SS "Lookbehind assertions"
1832 .rs
1833 .sp
1834 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1835 negative assertions. For example,
1836 .sp
1837 (?<!foo)bar
1838 .sp
1839 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1840 a lookbehind assertion are restricted such that all the strings it matches must
1841 have a fixed length. However, if there are several top-level alternatives, they
1842 do not all have to have the same fixed length. Thus
1843 .sp
1844 (?<=bullock|donkey)
1845 .sp
1846 is permitted, but
1847 .sp
1848 (?<!dogs?|cats?)
1849 .sp
1850 causes an error at compile time. Branches that match different length strings
1851 are permitted only at the top level of a lookbehind assertion. This is an
1852 extension compared with Perl (5.8 and 5.10), which requires all branches to
1853 match the same length of string. An assertion such as
1854 .sp
1855 (?<=ab(c|de))
1856 .sp
1857 is not permitted, because its single top-level branch can match two different
1858 lengths, but it is acceptable to PCRE if rewritten to use two top-level
1859 branches:
1860 .sp
1861 (?<=abc|abde)
1862 .sp
1863 In some cases, the Perl 5.10 escape sequence \eK
1864 .\" HTML <a href="#resetmatchstart">
1865 .\" </a>
1866 (see above)
1867 .\"
1868 can be used instead of a lookbehind assertion to get round the fixed-length
1869 restriction.
1870 .P
1871 The implementation of lookbehind assertions is, for each alternative, to
1872 temporarily move the current position back by the fixed length and then try to
1873 match. If there are insufficient characters before the current position, the
1874 assertion fails.
1875 .P
1876 PCRE does not allow the \eC escape (which matches a single byte in UTF-8 mode)
1877 to appear in lookbehind assertions, because it makes it impossible to calculate
1878 the length of the lookbehind. The \eX and \eR escapes, which can match
1879 different numbers of bytes, are also not permitted.
1880 .P
1881 .\" HTML <a href="#subpatternsassubroutines">
1882 .\" </a>
1883 "Subroutine"
1884 .\"
1885 calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
1886 as the subpattern matches a fixed-length string.
1887 .\" HTML <a href="#recursion">
1888 .\" </a>
1889 Recursion,
1890 .\"
1891 however, is not supported.
1892 .P
1893 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1894 specify efficient matching of fixed-length strings at the end of subject
1895 strings. Consider a simple pattern such as
1896 .sp
1897 abcd$
1898 .sp
1899 when applied to a long string that does not match. Because matching proceeds
1900 from left to right, PCRE will look for each "a" in the subject and then see if
1901 what follows matches the rest of the pattern. If the pattern is specified as
1902 .sp
1903 ^.*abcd$
1904 .sp
1905 the initial .* matches the entire string at first, but when this fails (because
1906 there is no following "a"), it backtracks to match all but the last character,
1907 then all but the last two characters, and so on. Once again the search for "a"
1908 covers the entire string, from right to left, so we are no better off. However,
1909 if the pattern is written as
1910 .sp
1911 ^.*+(?<=abcd)
1912 .sp
1913 there can be no backtracking for the .*+ item; it can match only the entire
1914 string. The subsequent lookbehind assertion does a single test on the last four
1915 characters. If it fails, the match fails immediately. For long strings, this
1916 approach makes a significant difference to the processing time.
1917 .
1918 .
1919 .SS "Using multiple assertions"
1920 .rs
1921 .sp
1922 Several assertions (of any sort) may occur in succession. For example,
1923 .sp
1924 (?<=\ed{3})(?<!999)foo
1925 .sp
1926 matches "foo" preceded by three digits that are not "999". Notice that each of
1927 the assertions is applied independently at the same point in the subject
1928 string. First there is a check that the previous three characters are all
1929 digits, and then there is a check that the same three characters are not "999".
1930 This pattern does \fInot\fP match "foo" preceded by six characters, the first
1931 of which are digits and the last three of which are not "999". For example, it
1932 doesn't match "123abcfoo". A pattern to do that is
1933 .sp
1934 (?<=\ed{3}...)(?<!999)foo
1935 .sp
1936 This time the first assertion looks at the preceding six characters, checking
1937 that the first three are digits, and then the second assertion checks that the
1938 preceding three characters are not "999".
1939 .P
1940 Assertions can be nested in any combination. For example,
1941 .sp
1942 (?<=(?<!foo)bar)baz
1943 .sp
1944 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1945 preceded by "foo", while
1946 .sp
1947 (?<=\ed{3}(?!999)...)foo
1948 .sp
1949 is another pattern that matches "foo" preceded by three digits and any three
1950 characters that are not "999".
1951 .
1952 .
1953 .\" HTML <a name="conditions"></a>
1954 .SH "CONDITIONAL SUBPATTERNS"
1955 .rs
1956 .sp
1957 It is possible to cause the matching process to obey a subpattern
1958 conditionally or to choose between two alternative subpatterns, depending on
1959 the result of an assertion, or whether a specific capturing subpattern has
1960 already been matched. The two possible forms of conditional subpattern are:
1961 .sp
1962 (?(condition)yes-pattern)
1963 (?(condition)yes-pattern|no-pattern)
1964 .sp
1965 If the condition is satisfied, the yes-pattern is used; otherwise the
1966 no-pattern (if present) is used. If there are more than two alternatives in the
1967 subpattern, a compile-time error occurs.
1968 .P
1969 There are four kinds of condition: references to subpatterns, references to
1970 recursion, a pseudo-condition called DEFINE, and assertions.
1971 .
1972 .SS "Checking for a used subpattern by number"
1973 .rs
1974 .sp
1975 If the text between the parentheses consists of a sequence of digits, the
1976 condition is true if a capturing subpattern of that number has previously
1977 matched. If there is more than one capturing subpattern with the same number
1978 (see the earlier
1979 .\"
1980 .\" HTML <a href="#recursion">
1981 .\" </a>
1982 section about duplicate subpattern numbers),
1983 .\"
1984 the condition is true if any of them have been set. An alternative notation is
1985 to precede the digits with a plus or minus sign. In this case, the subpattern
1986 number is relative rather than absolute. The most recently opened parentheses
1987 can be referenced by (?(-1), the next most recent by (?(-2), and so on. In
1988 looping constructs it can also make sense to refer to subsequent groups with
1989 constructs such as (?(+2).
1990 .P
1991 Consider the following pattern, which contains non-significant white space to
1992 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
1993 three parts for ease of discussion:
1994 .sp
1995 ( \e( )? [^()]+ (?(1) \e) )
1996 .sp
1997 The first part matches an optional opening parenthesis, and if that
1998 character is present, sets it as the first captured substring. The second part
1999 matches one or more characters that are not parentheses. The third part is a
2000 conditional subpattern that tests whether the first set of parentheses matched
2001 or not. If they did, that is, if subject started with an opening parenthesis,
2002 the condition is true, and so the yes-pattern is executed and a closing
2003 parenthesis is required. Otherwise, since no-pattern is not present, the
2004 subpattern matches nothing. In other words, this pattern matches a sequence of
2005 non-parentheses, optionally enclosed in parentheses.
2006 .P
2007 If you were embedding this pattern in a larger one, you could use a relative
2008 reference:
2009 .sp
2010 ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
2011 .sp
2012 This makes the fragment independent of the parentheses in the larger pattern.
2013 .
2014 .SS "Checking for a used subpattern by name"
2015 .rs
2016 .sp
2017 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
2018 subpattern by name. For compatibility with earlier versions of PCRE, which had
2019 this facility before Perl, the syntax (?(name)...) is also recognized. However,
2020 there is a possible ambiguity with this syntax, because subpattern names may
2021 consist entirely of digits. PCRE looks first for a named subpattern; if it
2022 cannot find one and the name consists entirely of digits, PCRE looks for a
2023 subpattern of that number, which must be greater than zero. Using subpattern
2024 names that consist entirely of digits is not recommended.
2025 .P
2026 Rewriting the above example to use a named subpattern gives this:
2027 .sp
2028 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
2029 .sp
2030 If the name used in a condition of this kind is a duplicate, the test is
2031 applied to all subpatterns of the same name, and is true if any one of them has
2032 matched.
2033 .
2034 .SS "Checking for pattern recursion"
2035 .rs
2036 .sp
2037 If the condition is the string (R), and there is no subpattern with the name R,
2038 the condition is true if a recursive call to the whole pattern or any
2039 subpattern has been made. If digits or a name preceded by ampersand follow the
2040 letter R, for example:
2041 .sp
2042 (?(R3)...) or (?(R&name)...)
2043 .sp
2044 the condition is true if the most recent recursion is into a subpattern whose
2045 number or name is given. This condition does not check the entire recursion
2046 stack. If the name used in a condition of this kind is a duplicate, the test is
2047 applied to all subpatterns of the same name, and is true if any one of them is
2048 the most recent recursion.
2049 .P
2050 At "top level", all these recursion test conditions are false.
2051 .\" HTML <a href="#recursion">
2052 .\" </a>
2053 The syntax for recursive patterns
2054 .\"
2055 is described below.
2056 .
2057 .SS "Defining subpatterns for use by reference only"
2058 .rs
2059 .sp
2060 If the condition is the string (DEFINE), and there is no subpattern with the
2061 name DEFINE, the condition is always false. In this case, there may be only one
2062 alternative in the subpattern. It is always skipped if control reaches this
2063 point in the pattern; the idea of DEFINE is that it can be used to define
2064 "subroutines" that can be referenced from elsewhere. (The use of
2065 .\" HTML <a href="#subpatternsassubroutines">
2066 .\" </a>
2067 "subroutines"
2068 .\"
2069 is described below.) For example, a pattern to match an IPv4 address could be
2070 written like this (ignore whitespace and line breaks):
2071 .sp
2072 (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
2073 \eb (?&byte) (\e.(?&byte)){3} \eb
2074 .sp
2075 The first part of the pattern is a DEFINE group inside which a another group
2076 named "byte" is defined. This matches an individual component of an IPv4
2077 address (a number less than 256). When matching takes place, this part of the
2078 pattern is skipped because DEFINE acts like a false condition. The rest of the
2079 pattern uses references to the named group to match the four dot-separated
2080 components of an IPv4 address, insisting on a word boundary at each end.
2081 .
2082 .SS "Assertion conditions"
2083 .rs
2084 .sp
2085 If the condition is not in any of the above formats, it must be an assertion.
2086 This may be a positive or negative lookahead or lookbehind assertion. Consider
2087 this pattern, again containing non-significant white space, and with the two
2088 alternatives on the second line:
2089 .sp
2090 (?(?=[^a-z]*[a-z])
2091 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
2092 .sp
2093 The condition is a positive lookahead assertion that matches an optional
2094 sequence of non-letters followed by a letter. In other words, it tests for the
2095 presence of at least one letter in the subject. If a letter is found, the
2096 subject is matched against the first alternative; otherwise it is matched
2097 against the second. This pattern matches strings in one of the two forms
2098 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2099 .
2100 .
2101 .\" HTML <a name="comments"></a>
2102 .SH COMMENTS
2103 .rs
2104 .sp
2105 The sequence (?# marks the start of a comment that continues up to the next
2106 closing parenthesis. Nested parentheses are not permitted. The characters
2107 that make up a comment play no part in the pattern matching at all.
2108 .P
2109 If the PCRE_EXTENDED option is set, an unescaped # character outside a
2110 character class introduces a comment that continues to immediately after the
2111 next newline in the pattern.
2112 .
2113 .
2114 .\" HTML <a name="recursion"></a>
2115 .SH "RECURSIVE PATTERNS"
2116 .rs
2117 .sp
2118 Consider the problem of matching a string in parentheses, allowing for
2119 unlimited nested parentheses. Without the use of recursion, the best that can
2120 be done is to use a pattern that matches up to some fixed depth of nesting. It
2121 is not possible to handle an arbitrary nesting depth.
2122 .P
2123 For some time, Perl has provided a facility that allows regular expressions to
2124 recurse (amongst other things). It does this by interpolating Perl code in the
2125 expression at run time, and the code can refer to the expression itself. A Perl
2126 pattern using code interpolation to solve the parentheses problem can be
2127 created like this:
2128 .sp
2129 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
2130 .sp
2131 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2132 recursively to the pattern in which it appears.
2133 .P
2134 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2135 supports special syntax for recursion of the entire pattern, and also for
2136 individual subpattern recursion. After its introduction in PCRE and Python,
2137 this kind of recursion was subsequently introduced into Perl at release 5.10.
2138 .P
2139 A special item that consists of (? followed by a number greater than zero and a
2140 closing parenthesis is a recursive call of the subpattern of the given number,
2141 provided that it occurs inside that subpattern. (If not, it is a
2142 .\" HTML <a href="#subpatternsassubroutines">
2143 .\" </a>
2144 "subroutine"
2145 .\"
2146 call, which is described in the next section.) The special item (?R) or (?0) is
2147 a recursive call of the entire regular expression.
2148 .P
2149 This PCRE pattern solves the nested parentheses problem (assume the
2150 PCRE_EXTENDED option is set so that white space is ignored):
2151 .sp
2152 \e( ( [^()]++ | (?R) )* \e)
2153 .sp
2154 First it matches an opening parenthesis. Then it matches any number of
2155 substrings which can either be a sequence of non-parentheses, or a recursive
2156 match of the pattern itself (that is, a correctly parenthesized substring).
2157 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2158 to avoid backtracking into sequences of non-parentheses.
2159 .P
2160 If this were part of a larger pattern, you would not want to recurse the entire
2161 pattern, so instead you could use this:
2162 .sp
2163 ( \e( ( [^()]++ | (?1) )* \e) )
2164 .sp
2165 We have put the pattern into parentheses, and caused the recursion to refer to
2166 them instead of the whole pattern.
2167 .P
2168 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2169 is made easier by the use of relative references (a Perl 5.10 feature).
2170 Instead of (?1) in the pattern above you can write (?-2) to refer to the second
2171 most recently opened parentheses preceding the recursion. In other words, a
2172 negative number counts capturing parentheses leftwards from the point at which
2173 it is encountered.
2174 .P
2175 It is also possible to refer to subsequently opened parentheses, by writing
2176 references such as (?+2). However, these cannot be recursive because the
2177 reference is not inside the parentheses that are referenced. They are always
2178 .\" HTML <a href="#subpatternsassubroutines">
2179 .\" </a>
2180 "subroutine"
2181 .\"
2182 calls, as described in the next section.
2183 .P
2184 An alternative approach is to use named parentheses instead. The Perl syntax
2185 for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
2186 could rewrite the above example as follows:
2187 .sp
2188 (?<pn> \e( ( [^()]++ | (?&pn) )* \e) )
2189 .sp
2190 If there is more than one subpattern with the same name, the earliest one is
2191 used.
2192 .P
2193 This particular example pattern that we have been looking at contains nested
2194 unlimited repeats, and so the use of a possessive quantifier for matching
2195 strings of non-parentheses is important when applying the pattern to strings
2196 that do not match. For example, when this pattern is applied to
2197 .sp
2198 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2199 .sp
2200 it yields "no match" quickly. However, if a possessive quantifier is not used,
2201 the match runs for a very long time indeed because there are so many different
2202 ways the + and * repeats can carve up the subject, and all have to be tested
2203 before failure can be reported.
2204 .P
2205 At the end of a match, the values of capturing parentheses are those from
2206 the outermost level. If you want to obtain intermediate values, a callout
2207 function can be used (see below and the
2208 .\" HREF
2209 \fBpcrecallout\fP
2210 .\"
2211 documentation). If the pattern above is matched against
2212 .sp
2213 (ab(cd)ef)
2214 .sp
2215 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2216 the last value taken on at the top level. If a capturing subpattern is not
2217 matched at the top level, its final value is unset, even if it is (temporarily)
2218 set at a deeper level.
2219 .P
2220 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2221 obtain extra memory to store data during a recursion, which it does by using
2222 \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no memory can
2223 be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2224 .P
2225 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2226 Consider this pattern, which matches text in angle brackets, allowing for
2227 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2228 recursing), whereas any characters are permitted at the outer level.
2229 .sp
2230 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
2231 .sp
2232 In this pattern, (?(R) is the start of a conditional subpattern, with two
2233 different alternatives for the recursive and non-recursive cases. The (?R) item
2234 is the actual recursive call.
2235 .
2236 .
2237 .\" HTML <a name="recursiondifference"></a>
2238 .SS "Recursion difference from Perl"
2239 .rs
2240 .sp
2241 In PCRE (like Python, but unlike Perl), a recursive subpattern call is always
2242 treated as an atomic group. That is, once it has matched some of the subject
2243 string, it is never re-entered, even if it contains untried alternatives and
2244 there is a subsequent matching failure. This can be illustrated by the
2245 following pattern, which purports to match a palindromic string that contains
2246 an odd number of characters (for example, "a", "aba", "abcba", "abcdcba"):
2247 .sp
2248 ^(.|(.)(?1)\e2)$
2249 .sp
2250 The idea is that it either matches a single character, or two identical
2251 characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2252 it does not if the pattern is longer than three characters. Consider the
2253 subject string "abcba":
2254 .P
2255 At the top level, the first character is matched, but as it is not at the end
2256 of the string, the first alternative fails; the second alternative is taken
2257 and the recursion kicks in. The recursive call to subpattern 1 successfully
2258 matches the next character ("b"). (Note that the beginning and end of line
2259 tests are not part of the recursion).
2260 .P
2261 Back at the top level, the next character ("c") is compared with what
2262 subpattern 2 matched, which was "a". This fails. Because the recursion is
2263 treated as an atomic group, there are now no backtracking points, and so the
2264 entire match fails. (Perl is able, at this point, to re-enter the recursion and
2265 try the second alternative.) However, if the pattern is written with the
2266 alternatives in the other order, things are different:
2267 .sp
2268 ^((.)(?1)\e2|.)$
2269 .sp
2270 This time, the recursing alternative is tried first, and continues to recurse
2271 until it runs out of characters, at which point the recursion fails. But this
2272 time we do have another alternative to try at the higher level. That is the big
2273 difference: in the previous case the remaining alternative is at a deeper
2274 recursion level, which PCRE cannot use.
2275 .P
2276 To change the pattern so that matches all palindromic strings, not just those
2277 with an odd number of characters, it is tempting to change the pattern to this:
2278 .sp
2279 ^((.)(?1)\e2|.?)$
2280 .sp
2281 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2282 deeper recursion has matched a single character, it cannot be entered again in
2283 order to match an empty string. The solution is to separate the two cases, and
2284 write out the odd and even cases as alternatives at the higher level:
2285 .sp
2286 ^(?:((.)(?1)\e2|)|((.)(?3)\e4|.))
2287 .sp
2288 If you want to match typical palindromic phrases, the pattern has to ignore all
2289 non-word characters, which can be done like this:
2290 .sp
2291 ^\eW*+(?:((.)\eW*+(?1)\eW*+\e2|)|((.)\eW*+(?3)\eW*+\e4|\eW*+.\eW*+))\eW*+$
2292 .sp
2293 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2294 man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2295 the use of the possessive quantifier *+ to avoid backtracking into sequences of
2296 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2297 more) to match typical phrases, and Perl takes so long that you think it has
2298 gone into a loop.
2299 .P
2300 \fBWARNING\fP: The palindrome-matching patterns above work only if the subject
2301 string does not start with a palindrome that is shorter than the entire string.
2302 For example, although "abcba" is correctly matched, if the subject is "ababa",
2303 PCRE finds the palindrome "aba" at the start, then fails at top level because
2304 the end of the string does not follow. Once again, it cannot jump back into the
2305 recursion to try other alternatives, so the entire match fails.
2306 .
2307 .
2308 .\" HTML <a name="subpatternsassubroutines"></a>
2309 .SH "SUBPATTERNS AS SUBROUTINES"
2310 .rs
2311 .sp
2312 If the syntax for a recursive subpattern reference (either by number or by
2313 name) is used outside the parentheses to which it refers, it operates like a
2314 subroutine in a programming language. The "called" subpattern may be defined
2315 before or after the reference. A numbered reference can be absolute or
2316 relative, as in these examples:
2317 .sp
2318 (...(absolute)...)...(?2)...
2319 (...(relative)...)...(?-1)...
2320 (...(?+1)...(relative)...
2321 .sp
2322 An earlier example pointed out that the pattern
2323 .sp
2324 (sens|respons)e and \e1ibility
2325 .sp
2326 matches "sense and sensibility" and "response and responsibility", but not
2327 "sense and responsibility". If instead the pattern
2328 .sp
2329 (sens|respons)e and (?1)ibility
2330 .sp
2331 is used, it does match "sense and responsibility" as well as the other two
2332 strings. Another example is given in the discussion of DEFINE above.
2333 .P
2334 Like recursive subpatterns, a subroutine call is always treated as an atomic
2335 group. That is, once it has matched some of the subject string, it is never
2336 re-entered, even if it contains untried alternatives and there is a subsequent
2337 matching failure. Any capturing parentheses that are set during the subroutine
2338 call revert to their previous values afterwards.
2339 .P
2340 When a subpattern is used as a subroutine, processing options such as
2341 case-independence are fixed when the subpattern is defined. They cannot be
2342 changed for different calls. For example, consider this pattern:
2343 .sp
2344 (abc)(?i:(?-1))
2345 .sp
2346 It matches "abcabc". It does not match "abcABC" because the change of
2347 processing option does not affect the called subpattern.
2348 .
2349 .
2350 .\" HTML <a name="onigurumasubroutines"></a>
2351 .SH "ONIGURUMA SUBROUTINE SYNTAX"
2352 .rs
2353 .sp
2354 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
2355 a number enclosed either in angle brackets or single quotes, is an alternative
2356 syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2357 are two of the examples used above, rewritten using this syntax:
2358 .sp
2359 (?<pn> \e( ( (?>[^()]+) | \eg<pn> )* \e) )
2360 (sens|respons)e and \eg'1'ibility
2361 .sp
2362 PCRE supports an extension to Oniguruma: if a number is preceded by a
2363 plus or a minus sign it is taken as a relative reference. For example:
2364 .sp
2365 (abc)(?i:\eg<-1>)
2366 .sp
2367 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
2368 synonymous. The former is a back reference; the latter is a subroutine call.
2369 .
2370 .
2371 .SH CALLOUTS
2372 .rs
2373 .sp
2374 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2375 code to be obeyed in the middle of matching a regular expression. This makes it
2376 possible, amongst other things, to extract different substrings that match the
2377 same pair of parentheses when there is a repetition.
2378 .P
2379 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2380 code. The feature is called "callout". The caller of PCRE provides an external
2381 function by putting its entry point in the global variable \fIpcre_callout\fP.
2382 By default, this variable contains NULL, which disables all calling out.
2383 .P
2384 Within a regular expression, (?C) indicates the points at which the external
2385 function is to be called. If you want to identify different callout points, you
2386 can put a number less than 256 after the letter C. The default value is zero.
2387 For example, this pattern has two callout points:
2388 .sp
2389 (?C1)abc(?C2)def
2390 .sp
2391 If the PCRE_AUTO_CALLOUT flag is passed to \fBpcre_compile()\fP, callouts are
2392 automatically installed before each item in the pattern. They are all numbered
2393 255.
2394 .P
2395 During matching, when PCRE reaches a callout point (and \fIpcre_callout\fP is
2396 set), the external function is called. It is provided with the number of the
2397 callout, the position in the pattern, and, optionally, one item of data
2398 originally supplied by the caller of \fBpcre_exec()\fP. The callout function
2399 may cause matching to proceed, to backtrack, or to fail altogether. A complete
2400 description of the interface to the callout function is given in the
2401 .\" HREF
2402 \fBpcrecallout\fP
2403 .\"
2404 documentation.
2405 .
2406 .
2407 .\" HTML <a name="backtrackcontrol"></a>
2408 .SH "BACKTRACKING CONTROL"
2409 .rs
2410 .sp
2411 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2412 are described in the Perl documentation as "experimental and subject to change
2413 or removal in a future version of Perl". It goes on to say: "Their usage in
2414 production code should be noted to avoid problems during upgrades." The same
2415 remarks apply to the PCRE features described in this section.
2416 .P
2417 Since these verbs are specifically related to backtracking, most of them can be
2418 used only when the pattern is to be matched using \fBpcre_exec()\fP, which uses
2419 a backtracking algorithm. With the exception of (*FAIL), which behaves like a
2420 failing negative assertion, they cause an error if encountered by
2421 \fBpcre_dfa_exec()\fP.
2422 .P
2423 If any of these verbs are used in an assertion or subroutine subpattern
2424 (including recursive subpatterns), their effect is confined to that subpattern;
2425 it does not extend to the surrounding pattern. Note that such subpatterns are
2426 processed as anchored at the point where they are tested.
2427 .P
2428 The new verbs make use of what was previously invalid syntax: an opening
2429 parenthesis followed by an asterisk. They are generally of the form
2430 (*VERB) or (*VERB:NAME). Some may take either form, with differing behaviour,
2431 depending on whether or not an argument is present. An name is a sequence of
2432 letters, digits, and underscores. If the name is empty, that is, if the closing
2433 parenthesis immediately follows the colon, the effect is as if the colon were
2434 not there. Any number of these verbs may occur in a pattern.
2435 .P
2436 PCRE contains some optimizations that are used to speed up matching by running
2437 some checks at the start of each match attempt. For example, it may know the
2438 minimum length of matching subject, or that a particular character must be
2439 present. When one of these optimizations suppresses the running of a match, any
2440 included backtracking verbs will not, of course, be processed. You can suppress
2441 the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2442 when calling \fBpcre_exec()\fP.
2443 .
2444 .
2445 .SS "Verbs that act immediately"
2446 .rs
2447 .sp
2448 The following verbs act as soon as they are encountered. They may not be
2449 followed by a name.
2450 .sp
2451 (*ACCEPT)
2452 .sp
2453 This verb causes the match to end successfully, skipping the remainder of the
2454 pattern. When inside a recursion, only the innermost pattern is ended
2455 immediately. If (*ACCEPT) is inside capturing parentheses, the data so far is
2456 captured. (This feature was added to PCRE at release 8.00.) For example:
2457 .sp
2458 A((?:A|B(*ACCEPT)|C)D)
2459 .sp
2460 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2461 the outer parentheses.
2462 .sp
2463 (*FAIL) or (*F)
2464 .sp
2465 This verb causes the match to fail, forcing backtracking to occur. It is
2466 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2467 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2468 Perl features that are not present in PCRE. The nearest equivalent is the
2469 callout feature, as for example in this pattern:
2470 .sp
2471 a+(?C)(*FAIL)
2472 .sp
2473 A match with the string "aaaa" always fails, but the callout is taken before
2474 each backtrack happens (in this example, 10 times).
2475 .
2476 .
2477 .SS "Recording which path was taken"
2478 .rs
2479 .sp
2480 There is one verb whose main purpose is to track how a match was arrived at,
2481 though it also has a secondary use in conjunction with advancing the match
2482 starting point (see (*SKIP) below).
2483 .sp
2484 (*MARK:NAME) or (*:NAME)
2485 .sp
2486 A name is always required with this verb. There may be as many instances of
2487 (*MARK) as you like in a pattern, and their names do not have to be unique.
2488 .P
2489 When a match succeeds, the name of the last-encountered (*MARK) is passed back
2490 to the caller via the \fIpcre_extra\fP data structure, as described in the
2491 .\" HTML <a href="pcreapi.html#extradata">
2492 .\" </a>
2493 section on \fIpcre_extra\fP
2494 .\"
2495 in the
2496 .\" HREF
2497 \fBpcreapi\fP
2498 .\"
2499 documentation. No data is returned for a partial match. Here is an example of
2500 \fBpcretest\fP output, where the /K modifier requests the retrieval and
2501 outputting of (*MARK) data:
2502 .sp
2503 /X(*MARK:A)Y|X(*MARK:B)Z/K
2504 XY
2505 0: XY
2506 MK: A
2507 XZ
2508 0: XZ
2509 MK: B
2510 .sp
2511 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2512 indicates which of the two alternatives matched. This is a more efficient way
2513 of obtaining this information than putting each alternative in its own
2514 capturing parentheses.
2515 .P
2516 A name may also be returned after a failed match if the final path through the
2517 pattern involves (*MARK). However, unless (*MARK) used in conjunction with
2518 (*COMMIT), this is unlikely to happen for an unanchored pattern because, as the
2519 starting point for matching is advanced, the final check is often with an empty
2520 string, causing a failure before (*MARK) is reached. For example:
2521 .sp
2522 /X(*MARK:A)Y|X(*MARK:B)Z/K
2523 XP
2524 No match
2525 .sp
2526 There are three potential starting points for this match (starting with X,
2527 starting with P, and with an empty string). If the pattern is anchored, the
2528 result is different:
2529 .sp
2530 /^X(*MARK:A)Y|^X(*MARK:B)Z/K
2531 XP
2532 No match, mark = B
2533 .sp
2534 PCRE's start-of-match optimizations can also interfere with this. For example,
2535 if, as a result of a call to \fBpcre_study()\fP, it knows the minimum
2536 subject length for a match, a shorter subject will not be scanned at all.
2537 .P
2538 Note that similar anomalies (though different in detail) exist in Perl, no
2539 doubt for the same reasons. The use of (*MARK) data after a failed match of an
2540 unanchored pattern is not recommended, unless (*COMMIT) is involved.
2541 .
2542 .
2543 .SS "Verbs that act after backtracking"
2544 .rs
2545 .sp
2546 The following verbs do nothing when they are encountered. Matching continues
2547 with what follows, but if there is no subsequent match, causing a backtrack to
2548 the verb, a failure is forced. That is, backtracking cannot pass to the left of
2549 the verb. However, when one of these verbs appears inside an atomic group, its
2550 effect is confined to that group, because once the group has been matched,
2551 there is never any backtracking into it. In this situation, backtracking can
2552 "jump back" to the left of the entire atomic group. (Remember also, as stated
2553 above, that this localization also applies in subroutine calls and assertions.)
2554 .P
2555 These verbs differ in exactly what kind of failure occurs when backtracking
2556 reaches them.
2557 .sp
2558 (*COMMIT)
2559 .sp
2560 This verb, which may not be followed by a name, causes the whole match to fail
2561 outright if the rest of the pattern does not match. Even if the pattern is
2562 unanchored, no further attempts to find a match by advancing the starting point
2563 take place. Once (*COMMIT) has been passed, \fBpcre_exec()\fP is committed to
2564 finding a match at the current starting point, or not at all. For example:
2565 .sp
2566 a+(*COMMIT)b
2567 .sp
2568 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2569 dynamic anchor, or "I've started, so I must finish." The name of the most
2570 recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2571 match failure.
2572 .P
2573 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2574 unless PCRE's start-of-match optimizations are turned off, as shown in this
2575 \fBpcretest\fP example:
2576 .sp
2577 /(*COMMIT)abc/
2578 xyzabc
2579 0: abc
2580 xyzabc\eY
2581 No match
2582 .sp
2583 PCRE knows that any match must start with "a", so the optimization skips along
2584 the subject to "a" before running the first match attempt, which succeeds. When
2585 the optimization is disabled by the \eY escape in the second subject, the match
2586 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2587 starting points.
2588 .sp
2589 (*PRUNE) or (*PRUNE:NAME)
2590 .sp
2591 This verb causes the match to fail at the current starting position in the
2592 subject if the rest of the pattern does not match. If the pattern is
2593 unanchored, the normal "bumpalong" advance to the next starting character then
2594 happens. Backtracking can occur as usual to the left of (*PRUNE), before it is
2595 reached, or when matching to the right of (*PRUNE), but if there is no match to
2596 the right, backtracking cannot cross (*PRUNE). In simple cases, the use of
2597 (*PRUNE) is just an alternative to an atomic group or possessive quantifier,
2598 but there are some uses of (*PRUNE) that cannot be expressed in any other way.
2599 The behaviour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE) when the
2600 match fails completely; the name is passed back if this is the final attempt.
2601 (*PRUNE:NAME) does not pass back a name if the match succeeds. In an anchored
2602 pattern (*PRUNE) has the same effect as (*COMMIT).
2603 .sp
2604 (*SKIP)
2605 .sp
2606 This verb, when given without a name, is like (*PRUNE), except that if the
2607 pattern is unanchored, the "bumpalong" advance is not to the next character,
2608 but to the position in the subject where (*SKIP) was encountered. (*SKIP)
2609 signifies that whatever text was matched leading up to it cannot be part of a
2610 successful match. Consider:
2611 .sp
2612 a+(*SKIP)b
2613 .sp
2614 If the subject is "aaaac...", after the first match attempt fails (starting at
2615 the first character in the string), the starting point skips on to start the
2616 next attempt at "c". Note that a possessive quantifer does not have the same
2617 effect as this example; although it would suppress backtracking during the
2618 first match attempt, the second attempt would start at the second character
2619 instead of skipping on to "c".
2620 .sp
2621 (*SKIP:NAME)
2622 .sp
2623 When (*SKIP) has an associated name, its behaviour is modified. If the
2624 following pattern fails to match, the previous path through the pattern is
2625 searched for the most recent (*MARK) that has the same name. If one is found,
2626 the "bumpalong" advance is to the subject position that corresponds to that
2627 (*MARK) instead of to where (*SKIP) was encountered. If no (*MARK) with a
2628 matching name is found, normal "bumpalong" of one character happens (the
2629 (*SKIP) is ignored).
2630 .sp
2631 (*THEN) or (*THEN:NAME)
2632 .sp
2633 This verb causes a skip to the next alternation if the rest of the pattern does
2634 not match. That is, it cancels pending backtracking, but only within the
2635 current alternation. Its name comes from the observation that it can be used
2636 for a pattern-based if-then-else block:
2637 .sp
2638 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2639 .sp
2640 If the COND1 pattern matches, FOO is tried (and possibly further items after
2641 the end of the group if FOO succeeds); on failure the matcher skips to the
2642 second alternative and tries COND2, without backtracking into COND1. The
2643 behaviour of (*THEN:NAME) is exactly the same as (*MARK:NAME)(*THEN) if the
2644 overall match fails. If (*THEN) is not directly inside an alternation, it acts
2645 like (*PRUNE).
2646 .
2647 .
2648 .SH "SEE ALSO"
2649 .rs
2650 .sp
2651 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3),
2652 \fBpcresyntax\fP(3), \fBpcre\fP(3).
2653 .
2654 .
2655 .SH AUTHOR
2656 .rs
2657 .sp
2658 .nf
2659 Philip Hazel
2660 University Computing Service
2661 Cambridge CB2 3QH, England.
2662 .fi
2663 .
2664 .
2665 .SH REVISION
2666 .rs
2667 .sp
2668 .nf
2669 Last updated: 18 May 2010
2670 Copyright (c) 1997-2010 University of Cambridge.
2671 .fi

Properties

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

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12