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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 247 - (show annotations) (download)
Mon Sep 17 09:38:32 2007 UTC (7 years, 1 month ago) by ph10
File size: 89067 byte(s)
(1) Update tests 2,5,7,8 to run when --enable-bsr-anycrlf is used.
(2) Updates files changed by building test release.

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. Perl's regular expressions are described in its own documentation, and
13 regular expressions in general are covered in a number of books, some of which
14 have copious examples. Jeffrey Friedl's "Mastering Regular Expressions",
15 published by O'Reilly, covers regular expressions in great detail. This
16 description of PCRE's regular expressions is intended as reference material.
17 .P
18 The original operation of PCRE was on strings of one-byte characters. However,
19 there is now also support for UTF-8 character strings. To use this, you must
20 build PCRE to include UTF-8 support, and then call \fBpcre_compile()\fP with
21 the PCRE_UTF8 option. How this affects pattern matching is mentioned in several
22 places below. There is also a summary of UTF-8 features in the
23 .\" HTML <a href="pcre.html#utf8support">
24 .\" </a>
25 section on UTF-8 support
26 .\"
27 in the main
28 .\" HREF
29 \fBpcre\fP
30 .\"
31 page.
32 .P
33 The remainder of this document discusses the patterns that are supported by
34 PCRE when its main matching function, \fBpcre_exec()\fP, is used.
35 From release 6.0, PCRE offers a second matching function,
36 \fBpcre_dfa_exec()\fP, which matches using a different algorithm that is not
37 Perl-compatible. Some of the features discussed below are not available when
38 \fBpcre_dfa_exec()\fP is used. The advantages and disadvantages of the
39 alternative function, and how it differs from the normal function, are
40 discussed in the
41 .\" HREF
42 \fBpcrematching\fP
43 .\"
44 page.
45 .
46 .
47 .SH "NEWLINE CONVENTIONS"
48 .rs
49 .sp
50 PCRE supports five different conventions for indicating line breaks in
51 strings: a single CR (carriage return) character, a single LF (linefeed)
52 character, the two-character sequence CRLF, any of the three preceding, or any
53 Unicode newline sequence. The
54 .\" HREF
55 \fBpcreapi\fP
56 .\"
57 page has
58 .\" HTML <a href="pcreapi.html#newlines">
59 .\" </a>
60 further discussion
61 .\"
62 about newlines, and shows how to set the newline convention in the
63 \fIoptions\fP arguments for the compiling and matching functions.
64 .P
65 It is also possible to specify a newline convention by starting a pattern
66 string with one of the following five sequences:
67 .sp
68 (*CR) carriage return
69 (*LF) linefeed
70 (*CRLF) carriage return, followed by linefeed
71 (*ANYCRLF) any of the three above
72 (*ANY) all Unicode newline sequences
73 .sp
74 These override the default and the options given to \fBpcre_compile()\fP. For
75 example, on a Unix system where LF is the default newline sequence, the pattern
76 .sp
77 (*CR)a.b
78 .sp
79 changes the convention to CR. That pattern matches "a\enb" because LF is no
80 longer a newline. Note that these special settings, which are not
81 Perl-compatible, are recognized only at the very start of a pattern, and that
82 they must be in upper case. If more than one of them is present, the last one
83 is used.
84 .P
85 The newline convention does not affect what the \eR escape sequence matches. By
86 default, this is any Unicode newline sequence, for Perl compatibility. However,
87 this can be changed; see the description of \eR in the section entitled
88 .\" HTML <a href="#newlineseq">
89 .\" </a>
90 "Newline sequences"
91 .\"
92 below. A change of \eR setting can be combined with a change of newline
93 convention.
94 .
95 .
96 .SH "CHARACTERS AND METACHARACTERS"
97 .rs
98 .sp
99 A regular expression is a pattern that is matched against a subject string from
100 left to right. Most characters stand for themselves in a pattern, and match the
101 corresponding characters in the subject. As a trivial example, the pattern
102 .sp
103 The quick brown fox
104 .sp
105 matches a portion of a subject string that is identical to itself. When
106 caseless matching is specified (the PCRE_CASELESS option), letters are matched
107 independently of case. In UTF-8 mode, PCRE always understands the concept of
108 case for characters whose values are less than 128, so caseless matching is
109 always possible. For characters with higher values, the concept of case is
110 supported if PCRE is compiled with Unicode property support, but not otherwise.
111 If you want to use caseless matching for characters 128 and above, you must
112 ensure that PCRE is compiled with Unicode property support as well as with
113 UTF-8 support.
114 .P
115 The power of regular expressions comes from the ability to include alternatives
116 and repetitions in the pattern. These are encoded in the pattern by the use of
117 \fImetacharacters\fP, which do not stand for themselves but instead are
118 interpreted in some special way.
119 .P
120 There are two different sets of metacharacters: those that are recognized
121 anywhere in the pattern except within square brackets, and those that are
122 recognized within square brackets. Outside square brackets, the metacharacters
123 are as follows:
124 .sp
125 \e general escape character with several uses
126 ^ assert start of string (or line, in multiline mode)
127 $ assert end of string (or line, in multiline mode)
128 . match any character except newline (by default)
129 [ start character class definition
130 | start of alternative branch
131 ( start subpattern
132 ) end subpattern
133 ? extends the meaning of (
134 also 0 or 1 quantifier
135 also quantifier minimizer
136 * 0 or more quantifier
137 + 1 or more quantifier
138 also "possessive quantifier"
139 { start min/max quantifier
140 .sp
141 Part of a pattern that is in square brackets is called a "character class". In
142 a character class the only metacharacters are:
143 .sp
144 \e general escape character
145 ^ negate the class, but only if the first character
146 - indicates character range
147 .\" JOIN
148 [ POSIX character class (only if followed by POSIX
149 syntax)
150 ] terminates the character class
151 .sp
152 The following sections describe the use of each of the metacharacters.
153 .
154 .
155 .SH BACKSLASH
156 .rs
157 .sp
158 The backslash character has several uses. Firstly, if it is followed by a
159 non-alphanumeric character, it takes away any special meaning that character
160 may have. This use of backslash as an escape character applies both inside and
161 outside character classes.
162 .P
163 For example, if you want to match a * character, you write \e* in the pattern.
164 This escaping action applies whether or not the following character would
165 otherwise be interpreted as a metacharacter, so it is always safe to precede a
166 non-alphanumeric with backslash to specify that it stands for itself. In
167 particular, if you want to match a backslash, you write \e\e.
168 .P
169 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
170 pattern (other than in a character class) and characters between a # outside
171 a character class and the next newline are ignored. An escaping backslash can
172 be used to include a whitespace or # character as part of the pattern.
173 .P
174 If you want to remove the special meaning from a sequence of characters, you
175 can do so by putting them between \eQ and \eE. This is different from Perl in
176 that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
177 Perl, $ and @ cause variable interpolation. Note the following examples:
178 .sp
179 Pattern PCRE matches Perl matches
180 .sp
181 .\" JOIN
182 \eQabc$xyz\eE abc$xyz abc followed by the
183 contents of $xyz
184 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
185 \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
186 .sp
187 The \eQ...\eE sequence is recognized both inside and outside character classes.
188 .
189 .
190 .\" HTML <a name="digitsafterbackslash"></a>
191 .SS "Non-printing characters"
192 .rs
193 .sp
194 A second use of backslash provides a way of encoding non-printing characters
195 in patterns in a visible manner. There is no restriction on the appearance of
196 non-printing characters, apart from the binary zero that terminates a pattern,
197 but when a pattern is being prepared by text editing, it is usually easier to
198 use one of the following escape sequences than the binary character it
199 represents:
200 .sp
201 \ea alarm, that is, the BEL character (hex 07)
202 \ecx "control-x", where x is any character
203 \ee escape (hex 1B)
204 \ef formfeed (hex 0C)
205 \en linefeed (hex 0A)
206 \er carriage return (hex 0D)
207 \et tab (hex 09)
208 \eddd character with octal code ddd, or backreference
209 \exhh character with hex code hh
210 \ex{hhh..} character with hex code hhh..
211 .sp
212 The precise effect of \ecx is as follows: if x is a lower case letter, it
213 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
214 Thus \ecz becomes hex 1A, but \ec{ becomes hex 3B, while \ec; becomes hex
215 7B.
216 .P
217 After \ex, from zero to two hexadecimal digits are read (letters can be in
218 upper or lower case). Any number of hexadecimal digits may appear between \ex{
219 and }, but the value of the character code must be less than 256 in non-UTF-8
220 mode, and less than 2**31 in UTF-8 mode. That is, the maximum value in
221 hexadecimal is 7FFFFFFF. Note that this is bigger than the largest Unicode code
222 point, which is 10FFFF.
223 .P
224 If characters other than hexadecimal digits appear between \ex{ and }, or if
225 there is no terminating }, this form of escape is not recognized. Instead, the
226 initial \ex will be interpreted as a basic hexadecimal escape, with no
227 following digits, giving a character whose value is zero.
228 .P
229 Characters whose value is less than 256 can be defined by either of the two
230 syntaxes for \ex. There is no difference in the way they are handled. For
231 example, \exdc is exactly the same as \ex{dc}.
232 .P
233 After \e0 up to two further octal digits are read. If there are fewer than two
234 digits, just those that are present are used. Thus the sequence \e0\ex\e07
235 specifies two binary zeros followed by a BEL character (code value 7). Make
236 sure you supply two digits after the initial zero if the pattern character that
237 follows is itself an octal digit.
238 .P
239 The handling of a backslash followed by a digit other than 0 is complicated.
240 Outside a character class, PCRE reads it and any following digits as a decimal
241 number. If the number is less than 10, or if there have been at least that many
242 previous capturing left parentheses in the expression, the entire sequence is
243 taken as a \fIback reference\fP. A description of how this works is given
244 .\" HTML <a href="#backreferences">
245 .\" </a>
246 later,
247 .\"
248 following the discussion of
249 .\" HTML <a href="#subpattern">
250 .\" </a>
251 parenthesized subpatterns.
252 .\"
253 .P
254 Inside a character class, or if the decimal number is greater than 9 and there
255 have not been that many capturing subpatterns, PCRE re-reads up to three octal
256 digits following the backslash, and uses them to generate a data character. Any
257 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
258 character specified in octal must be less than \e400. In UTF-8 mode, values up
259 to \e777 are permitted. For example:
260 .sp
261 \e040 is another way of writing a space
262 .\" JOIN
263 \e40 is the same, provided there are fewer than 40
264 previous capturing subpatterns
265 \e7 is always a back reference
266 .\" JOIN
267 \e11 might be a back reference, or another way of
268 writing a tab
269 \e011 is always a tab
270 \e0113 is a tab followed by the character "3"
271 .\" JOIN
272 \e113 might be a back reference, otherwise the
273 character with octal code 113
274 .\" JOIN
275 \e377 might be a back reference, otherwise
276 the byte consisting entirely of 1 bits
277 .\" JOIN
278 \e81 is either a back reference, or a binary zero
279 followed by the two characters "8" and "1"
280 .sp
281 Note that octal values of 100 or greater must not be introduced by a leading
282 zero, because no more than three octal digits are ever read.
283 .P
284 All the sequences that define a single character value can be used both inside
285 and outside character classes. In addition, inside a character class, the
286 sequence \eb is interpreted as the backspace character (hex 08), and the
287 sequences \eR and \eX are interpreted as the characters "R" and "X",
288 respectively. Outside a character class, these sequences have different
289 meanings
290 .\" HTML <a href="#uniextseq">
291 .\" </a>
292 (see below).
293 .\"
294 .
295 .
296 .SS "Absolute and relative back references"
297 .rs
298 .sp
299 The sequence \eg followed by an unsigned or a negative number, optionally
300 enclosed in braces, is an absolute or relative back reference. A named back
301 reference can be coded as \eg{name}. Back references are discussed
302 .\" HTML <a href="#backreferences">
303 .\" </a>
304 later,
305 .\"
306 following the discussion of
307 .\" HTML <a href="#subpattern">
308 .\" </a>
309 parenthesized subpatterns.
310 .\"
311 .
312 .
313 .SS "Generic character types"
314 .rs
315 .sp
316 Another use of backslash is for specifying generic character types. The
317 following are always recognized:
318 .sp
319 \ed any decimal digit
320 \eD any character that is not a decimal digit
321 \eh any horizontal whitespace character
322 \eH any character that is not a horizontal whitespace character
323 \es any whitespace character
324 \eS any character that is not a whitespace character
325 \ev any vertical whitespace character
326 \eV any character that is not a vertical whitespace character
327 \ew any "word" character
328 \eW any "non-word" character
329 .sp
330 Each pair of escape sequences partitions the complete set of characters into
331 two disjoint sets. Any given character matches one, and only one, of each pair.
332 .P
333 These character type sequences can appear both inside and outside character
334 classes. They each match one character of the appropriate type. If the current
335 matching point is at the end of the subject string, all of them fail, since
336 there is no character to match.
337 .P
338 For compatibility with Perl, \es does not match the VT character (code 11).
339 This makes it different from the the POSIX "space" class. The \es characters
340 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
341 included in a Perl script, \es may match the VT character. In PCRE, it never
342 does.
343 .P
344 In UTF-8 mode, characters with values greater than 128 never match \ed, \es, or
345 \ew, and always match \eD, \eS, and \eW. This is true even when Unicode
346 character property support is available. These sequences retain their original
347 meanings from before UTF-8 support was available, mainly for efficiency
348 reasons.
349 .P
350 The sequences \eh, \eH, \ev, and \eV are Perl 5.10 features. In contrast to the
351 other sequences, these do match certain high-valued codepoints in UTF-8 mode.
352 The horizontal space characters are:
353 .sp
354 U+0009 Horizontal tab
355 U+0020 Space
356 U+00A0 Non-break space
357 U+1680 Ogham space mark
358 U+180E Mongolian vowel separator
359 U+2000 En quad
360 U+2001 Em quad
361 U+2002 En space
362 U+2003 Em space
363 U+2004 Three-per-em space
364 U+2005 Four-per-em space
365 U+2006 Six-per-em space
366 U+2007 Figure space
367 U+2008 Punctuation space
368 U+2009 Thin space
369 U+200A Hair space
370 U+202F Narrow no-break space
371 U+205F Medium mathematical space
372 U+3000 Ideographic space
373 .sp
374 The vertical space characters are:
375 .sp
376 U+000A Linefeed
377 U+000B Vertical tab
378 U+000C Formfeed
379 U+000D Carriage return
380 U+0085 Next line
381 U+2028 Line separator
382 U+2029 Paragraph separator
383 .P
384 A "word" character is an underscore or any character less than 256 that is a
385 letter or digit. The definition of letters and digits is controlled by PCRE's
386 low-valued character tables, and may vary if locale-specific matching is taking
387 place (see
388 .\" HTML <a href="pcreapi.html#localesupport">
389 .\" </a>
390 "Locale support"
391 .\"
392 in the
393 .\" HREF
394 \fBpcreapi\fP
395 .\"
396 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
397 or "french" in Windows, some character codes greater than 128 are used for
398 accented letters, and these are matched by \ew. The use of locales with Unicode
399 is discouraged.
400 .
401 .
402 .\" HTML <a name="newlineseq"></a>
403 .SS "Newline sequences"
404 .rs
405 .sp
406 Outside a character class, by default, the escape sequence \eR matches any
407 Unicode newline sequence. This is a Perl 5.10 feature. In non-UTF-8 mode \eR is
408 equivalent to the following:
409 .sp
410 (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
411 .sp
412 This is an example of an "atomic group", details of which are given
413 .\" HTML <a href="#atomicgroup">
414 .\" </a>
415 below.
416 .\"
417 This particular group matches either the two-character sequence CR followed by
418 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
419 U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
420 line, U+0085). The two-character sequence is treated as a single unit that
421 cannot be split.
422 .P
423 In UTF-8 mode, two additional characters whose codepoints are greater than 255
424 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
425 Unicode character property support is not needed for these characters to be
426 recognized.
427 .P
428 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
429 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
430 either at compile time or when the pattern is matched. (BSR is an abbrevation
431 for "backslash R".) This can be made the default when PCRE is built; if this is
432 the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
433 It is also possible to specify these settings by starting a pattern string with
434 one of the following sequences:
435 .sp
436 (*BSR_ANYCRLF) CR, LF, or CRLF only
437 (*BSR_UNICODE) any Unicode newline sequence
438 .sp
439 These override the default and the options given to \fBpcre_compile()\fP, but
440 they can be overridden by options given to \fBpcre_exec()\fP. Note that these
441 special settings, which are not Perl-compatible, are recognized only at the
442 very start of a pattern, and that they must be in upper case. If more than one
443 of them is present, the last one is used. They can be combined with a change of
444 newline convention, for example, a pattern can start with:
445 .sp
446 (*ANY)(*BSR_ANYCRLF)
447 .sp
448 Inside a character class, \eR matches the letter "R".
449 .
450 .
451 .\" HTML <a name="uniextseq"></a>
452 .SS Unicode character properties
453 .rs
454 .sp
455 When PCRE is built with Unicode character property support, three additional
456 escape sequences that match characters with specific properties are available.
457 When not in UTF-8 mode, these sequences are of course limited to testing
458 characters whose codepoints are less than 256, but they do work in this mode.
459 The extra escape sequences are:
460 .sp
461 \ep{\fIxx\fP} a character with the \fIxx\fP property
462 \eP{\fIxx\fP} a character without the \fIxx\fP property
463 \eX an extended Unicode sequence
464 .sp
465 The property names represented by \fIxx\fP above are limited to the Unicode
466 script names, the general category properties, and "Any", which matches any
467 character (including newline). Other properties such as "InMusicalSymbols" are
468 not currently supported by PCRE. Note that \eP{Any} does not match any
469 characters, so always causes a match failure.
470 .P
471 Sets of Unicode characters are defined as belonging to certain scripts. A
472 character from one of these sets can be matched using a script name. For
473 example:
474 .sp
475 \ep{Greek}
476 \eP{Han}
477 .sp
478 Those that are not part of an identified script are lumped together as
479 "Common". The current list of scripts is:
480 .P
481 Arabic,
482 Armenian,
483 Balinese,
484 Bengali,
485 Bopomofo,
486 Braille,
487 Buginese,
488 Buhid,
489 Canadian_Aboriginal,
490 Cherokee,
491 Common,
492 Coptic,
493 Cuneiform,
494 Cypriot,
495 Cyrillic,
496 Deseret,
497 Devanagari,
498 Ethiopic,
499 Georgian,
500 Glagolitic,
501 Gothic,
502 Greek,
503 Gujarati,
504 Gurmukhi,
505 Han,
506 Hangul,
507 Hanunoo,
508 Hebrew,
509 Hiragana,
510 Inherited,
511 Kannada,
512 Katakana,
513 Kharoshthi,
514 Khmer,
515 Lao,
516 Latin,
517 Limbu,
518 Linear_B,
519 Malayalam,
520 Mongolian,
521 Myanmar,
522 New_Tai_Lue,
523 Nko,
524 Ogham,
525 Old_Italic,
526 Old_Persian,
527 Oriya,
528 Osmanya,
529 Phags_Pa,
530 Phoenician,
531 Runic,
532 Shavian,
533 Sinhala,
534 Syloti_Nagri,
535 Syriac,
536 Tagalog,
537 Tagbanwa,
538 Tai_Le,
539 Tamil,
540 Telugu,
541 Thaana,
542 Thai,
543 Tibetan,
544 Tifinagh,
545 Ugaritic,
546 Yi.
547 .P
548 Each character has exactly one general category property, specified by a
549 two-letter abbreviation. For compatibility with Perl, negation can be specified
550 by including a circumflex between the opening brace and the property name. For
551 example, \ep{^Lu} is the same as \eP{Lu}.
552 .P
553 If only one letter is specified with \ep or \eP, it includes all the general
554 category properties that start with that letter. In this case, in the absence
555 of negation, the curly brackets in the escape sequence are optional; these two
556 examples have the same effect:
557 .sp
558 \ep{L}
559 \epL
560 .sp
561 The following general category property codes are supported:
562 .sp
563 C Other
564 Cc Control
565 Cf Format
566 Cn Unassigned
567 Co Private use
568 Cs Surrogate
569 .sp
570 L Letter
571 Ll Lower case letter
572 Lm Modifier letter
573 Lo Other letter
574 Lt Title case letter
575 Lu Upper case letter
576 .sp
577 M Mark
578 Mc Spacing mark
579 Me Enclosing mark
580 Mn Non-spacing mark
581 .sp
582 N Number
583 Nd Decimal number
584 Nl Letter number
585 No Other number
586 .sp
587 P Punctuation
588 Pc Connector punctuation
589 Pd Dash punctuation
590 Pe Close punctuation
591 Pf Final punctuation
592 Pi Initial punctuation
593 Po Other punctuation
594 Ps Open punctuation
595 .sp
596 S Symbol
597 Sc Currency symbol
598 Sk Modifier symbol
599 Sm Mathematical symbol
600 So Other symbol
601 .sp
602 Z Separator
603 Zl Line separator
604 Zp Paragraph separator
605 Zs Space separator
606 .sp
607 The special property L& is also supported: it matches a character that has
608 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
609 a modifier or "other".
610 .P
611 The Cs (Surrogate) property applies only to characters in the range U+D800 to
612 U+DFFF. Such characters are not valid in UTF-8 strings (see RFC 3629) and so
613 cannot be tested by PCRE, unless UTF-8 validity checking has been turned off
614 (see the discussion of PCRE_NO_UTF8_CHECK in the
615 .\" HREF
616 \fBpcreapi\fP
617 .\"
618 page).
619 .P
620 The long synonyms for these properties that Perl supports (such as \ep{Letter})
621 are not supported by PCRE, nor is it permitted to prefix any of these
622 properties with "Is".
623 .P
624 No character that is in the Unicode table has the Cn (unassigned) property.
625 Instead, this property is assumed for any code point that is not in the
626 Unicode table.
627 .P
628 Specifying caseless matching does not affect these escape sequences. For
629 example, \ep{Lu} always matches only upper case letters.
630 .P
631 The \eX escape matches any number of Unicode characters that form an extended
632 Unicode sequence. \eX is equivalent to
633 .sp
634 (?>\ePM\epM*)
635 .sp
636 That is, it matches a character without the "mark" property, followed by zero
637 or more characters with the "mark" property, and treats the sequence as an
638 atomic group
639 .\" HTML <a href="#atomicgroup">
640 .\" </a>
641 (see below).
642 .\"
643 Characters with the "mark" property are typically accents that affect the
644 preceding character. None of them have codepoints less than 256, so in
645 non-UTF-8 mode \eX matches any one character.
646 .P
647 Matching characters by Unicode property is not fast, because PCRE has to search
648 a structure that contains data for over fifteen thousand characters. That is
649 why the traditional escape sequences such as \ed and \ew do not use Unicode
650 properties in PCRE.
651 .
652 .
653 .\" HTML <a name="resetmatchstart"></a>
654 .SS "Resetting the match start"
655 .rs
656 .sp
657 The escape sequence \eK, which is a Perl 5.10 feature, causes any previously
658 matched characters not to be included in the final matched sequence. For
659 example, the pattern:
660 .sp
661 foo\eKbar
662 .sp
663 matches "foobar", but reports that it has matched "bar". This feature is
664 similar to a lookbehind assertion
665 .\" HTML <a href="#lookbehind">
666 .\" </a>
667 (described below).
668 .\"
669 However, in this case, the part of the subject before the real match does not
670 have to be of fixed length, as lookbehind assertions do. The use of \eK does
671 not interfere with the setting of
672 .\" HTML <a href="#subpattern">
673 .\" </a>
674 captured substrings.
675 .\"
676 For example, when the pattern
677 .sp
678 (foo)\eKbar
679 .sp
680 matches "foobar", the first substring is still set to "foo".
681 .
682 .
683 .\" HTML <a name="smallassertions"></a>
684 .SS "Simple assertions"
685 .rs
686 .sp
687 The final use of backslash is for certain simple assertions. An assertion
688 specifies a condition that has to be met at a particular point in a match,
689 without consuming any characters from the subject string. The use of
690 subpatterns for more complicated assertions is described
691 .\" HTML <a href="#bigassertions">
692 .\" </a>
693 below.
694 .\"
695 The backslashed assertions are:
696 .sp
697 \eb matches at a word boundary
698 \eB matches when not at a word boundary
699 \eA matches at the start of the subject
700 \eZ matches at the end of the subject
701 also matches before a newline at the end of the subject
702 \ez matches only at the end of the subject
703 \eG matches at the first matching position in the subject
704 .sp
705 These assertions may not appear in character classes (but note that \eb has a
706 different meaning, namely the backspace character, inside a character class).
707 .P
708 A word boundary is a position in the subject string where the current character
709 and the previous character do not both match \ew or \eW (i.e. one matches
710 \ew and the other matches \eW), or the start or end of the string if the
711 first or last character matches \ew, respectively.
712 .P
713 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
714 dollar (described in the next section) in that they only ever match at the very
715 start and end of the subject string, whatever options are set. Thus, they are
716 independent of multiline mode. These three assertions are not affected by the
717 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
718 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
719 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
720 at a point other than the beginning of the subject, \eA can never match. The
721 difference between \eZ and \ez is that \eZ matches before a newline at the end
722 of the string as well as at the very end, whereas \ez matches only at the end.
723 .P
724 The \eG assertion is true only when the current matching position is at the
725 start point of the match, as specified by the \fIstartoffset\fP argument of
726 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
727 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
728 arguments, you can mimic Perl's /g option, and it is in this kind of
729 implementation where \eG can be useful.
730 .P
731 Note, however, that PCRE's interpretation of \eG, as the start of the current
732 match, is subtly different from Perl's, which defines it as the end of the
733 previous match. In Perl, these can be different when the previously matched
734 string was empty. Because PCRE does just one match at a time, it cannot
735 reproduce this behaviour.
736 .P
737 If all the alternatives of a pattern begin with \eG, the expression is anchored
738 to the starting match position, and the "anchored" flag is set in the compiled
739 regular expression.
740 .
741 .
742 .SH "CIRCUMFLEX AND DOLLAR"
743 .rs
744 .sp
745 Outside a character class, in the default matching mode, the circumflex
746 character is an assertion that is true only if the current matching point is
747 at the start of the subject string. If the \fIstartoffset\fP argument of
748 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
749 option is unset. Inside a character class, circumflex has an entirely different
750 meaning
751 .\" HTML <a href="#characterclass">
752 .\" </a>
753 (see below).
754 .\"
755 .P
756 Circumflex need not be the first character of the pattern if a number of
757 alternatives are involved, but it should be the first thing in each alternative
758 in which it appears if the pattern is ever to match that branch. If all
759 possible alternatives start with a circumflex, that is, if the pattern is
760 constrained to match only at the start of the subject, it is said to be an
761 "anchored" pattern. (There are also other constructs that can cause a pattern
762 to be anchored.)
763 .P
764 A dollar character is an assertion that is true only if the current matching
765 point is at the end of the subject string, or immediately before a newline
766 at the end of the string (by default). Dollar need not be the last character of
767 the pattern if a number of alternatives are involved, but it should be the last
768 item in any branch in which it appears. Dollar has no special meaning in a
769 character class.
770 .P
771 The meaning of dollar can be changed so that it matches only at the very end of
772 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
773 does not affect the \eZ assertion.
774 .P
775 The meanings of the circumflex and dollar characters are changed if the
776 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
777 immediately after internal newlines as well as at the start of the subject
778 string. It does not match after a newline that ends the string. A dollar
779 matches before any newlines in the string, as well as at the very end, when
780 PCRE_MULTILINE is set. When newline is specified as the two-character
781 sequence CRLF, isolated CR and LF characters do not indicate newlines.
782 .P
783 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
784 \en represents a newline) in multiline mode, but not otherwise. Consequently,
785 patterns that are anchored in single line mode because all branches start with
786 ^ are not anchored in multiline mode, and a match for circumflex is possible
787 when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
788 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
789 .P
790 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
791 end of the subject in both modes, and if all branches of a pattern start with
792 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
793 .
794 .
795 .SH "FULL STOP (PERIOD, DOT)"
796 .rs
797 .sp
798 Outside a character class, a dot in the pattern matches any one character in
799 the subject string except (by default) a character that signifies the end of a
800 line. In UTF-8 mode, the matched character may be more than one byte long.
801 .P
802 When a line ending is defined as a single character, dot never matches that
803 character; when the two-character sequence CRLF is used, dot does not match CR
804 if it is immediately followed by LF, but otherwise it matches all characters
805 (including isolated CRs and LFs). When any Unicode line endings are being
806 recognized, dot does not match CR or LF or any of the other line ending
807 characters.
808 .P
809 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
810 option is set, a dot matches any one character, without exception. If the
811 two-character sequence CRLF is present in the subject string, it takes two dots
812 to match it.
813 .P
814 The handling of dot is entirely independent of the handling of circumflex and
815 dollar, the only relationship being that they both involve newlines. Dot has no
816 special meaning in a character class.
817 .
818 .
819 .SH "MATCHING A SINGLE BYTE"
820 .rs
821 .sp
822 Outside a character class, the escape sequence \eC matches any one byte, both
823 in and out of UTF-8 mode. Unlike a dot, it always matches any line-ending
824 characters. The feature is provided in Perl in order to match individual bytes
825 in UTF-8 mode. Because it breaks up UTF-8 characters into individual bytes,
826 what remains in the string may be a malformed UTF-8 string. For this reason,
827 the \eC escape sequence is best avoided.
828 .P
829 PCRE does not allow \eC to appear in lookbehind assertions
830 .\" HTML <a href="#lookbehind">
831 .\" </a>
832 (described below),
833 .\"
834 because in UTF-8 mode this would make it impossible to calculate the length of
835 the lookbehind.
836 .
837 .
838 .\" HTML <a name="characterclass"></a>
839 .SH "SQUARE BRACKETS AND CHARACTER CLASSES"
840 .rs
841 .sp
842 An opening square bracket introduces a character class, terminated by a closing
843 square bracket. A closing square bracket on its own is not special. If a
844 closing square bracket is required as a member of the class, it should be the
845 first data character in the class (after an initial circumflex, if present) or
846 escaped with a backslash.
847 .P
848 A character class matches a single character in the subject. In UTF-8 mode, the
849 character may occupy more than one byte. A matched character must be in the set
850 of characters defined by the class, unless the first character in the class
851 definition is a circumflex, in which case the subject character must not be in
852 the set defined by the class. If a circumflex is actually required as a member
853 of the class, ensure it is not the first character, or escape it with a
854 backslash.
855 .P
856 For example, the character class [aeiou] matches any lower case vowel, while
857 [^aeiou] matches any character that is not a lower case vowel. Note that a
858 circumflex is just a convenient notation for specifying the characters that
859 are in the class by enumerating those that are not. A class that starts with a
860 circumflex is not an assertion: it still consumes a character from the subject
861 string, and therefore it fails if the current pointer is at the end of the
862 string.
863 .P
864 In UTF-8 mode, characters with values greater than 255 can be included in a
865 class as a literal string of bytes, or by using the \ex{ escaping mechanism.
866 .P
867 When caseless matching is set, any letters in a class represent both their
868 upper case and lower case versions, so for example, a caseless [aeiou] matches
869 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
870 caseful version would. In UTF-8 mode, PCRE always understands the concept of
871 case for characters whose values are less than 128, so caseless matching is
872 always possible. For characters with higher values, the concept of case is
873 supported if PCRE is compiled with Unicode property support, but not otherwise.
874 If you want to use caseless matching for characters 128 and above, you must
875 ensure that PCRE is compiled with Unicode property support as well as with
876 UTF-8 support.
877 .P
878 Characters that might indicate line breaks are never treated in any special way
879 when matching character classes, whatever line-ending sequence is in use, and
880 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
881 such as [^a] always matches one of these characters.
882 .P
883 The minus (hyphen) character can be used to specify a range of characters in a
884 character class. For example, [d-m] matches any letter between d and m,
885 inclusive. If a minus character is required in a class, it must be escaped with
886 a backslash or appear in a position where it cannot be interpreted as
887 indicating a range, typically as the first or last character in the class.
888 .P
889 It is not possible to have the literal character "]" as the end character of a
890 range. A pattern such as [W-]46] is interpreted as a class of two characters
891 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
892 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
893 the end of range, so [W-\e]46] is interpreted as a class containing a range
894 followed by two other characters. The octal or hexadecimal representation of
895 "]" can also be used to end a range.
896 .P
897 Ranges operate in the collating sequence of character values. They can also be
898 used for characters specified numerically, for example [\e000-\e037]. In UTF-8
899 mode, ranges can include characters whose values are greater than 255, for
900 example [\ex{100}-\ex{2ff}].
901 .P
902 If a range that includes letters is used when caseless matching is set, it
903 matches the letters in either case. For example, [W-c] is equivalent to
904 [][\e\e^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
905 tables for a French locale are in use, [\exc8-\excb] matches accented E
906 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
907 characters with values greater than 128 only when it is compiled with Unicode
908 property support.
909 .P
910 The character types \ed, \eD, \ep, \eP, \es, \eS, \ew, and \eW may also appear
911 in a character class, and add the characters that they match to the class. For
912 example, [\edABCDEF] matches any hexadecimal digit. A circumflex can
913 conveniently be used with the upper case character types to specify a more
914 restricted set of characters than the matching lower case type. For example,
915 the class [^\eW_] matches any letter or digit, but not underscore.
916 .P
917 The only metacharacters that are recognized in character classes are backslash,
918 hyphen (only where it can be interpreted as specifying a range), circumflex
919 (only at the start), opening square bracket (only when it can be interpreted as
920 introducing a POSIX class name - see the next section), and the terminating
921 closing square bracket. However, escaping other non-alphanumeric characters
922 does no harm.
923 .
924 .
925 .SH "POSIX CHARACTER CLASSES"
926 .rs
927 .sp
928 Perl supports the POSIX notation for character classes. This uses names
929 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
930 this notation. For example,
931 .sp
932 [01[:alpha:]%]
933 .sp
934 matches "0", "1", any alphabetic character, or "%". The supported class names
935 are
936 .sp
937 alnum letters and digits
938 alpha letters
939 ascii character codes 0 - 127
940 blank space or tab only
941 cntrl control characters
942 digit decimal digits (same as \ed)
943 graph printing characters, excluding space
944 lower lower case letters
945 print printing characters, including space
946 punct printing characters, excluding letters and digits
947 space white space (not quite the same as \es)
948 upper upper case letters
949 word "word" characters (same as \ew)
950 xdigit hexadecimal digits
951 .sp
952 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
953 space (32). Notice that this list includes the VT character (code 11). This
954 makes "space" different to \es, which does not include VT (for Perl
955 compatibility).
956 .P
957 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
958 5.8. Another Perl extension is negation, which is indicated by a ^ character
959 after the colon. For example,
960 .sp
961 [12[:^digit:]]
962 .sp
963 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
964 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
965 supported, and an error is given if they are encountered.
966 .P
967 In UTF-8 mode, characters with values greater than 128 do not match any of
968 the POSIX character classes.
969 .
970 .
971 .SH "VERTICAL BAR"
972 .rs
973 .sp
974 Vertical bar characters are used to separate alternative patterns. For example,
975 the pattern
976 .sp
977 gilbert|sullivan
978 .sp
979 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
980 and an empty alternative is permitted (matching the empty string). The matching
981 process tries each alternative in turn, from left to right, and the first one
982 that succeeds is used. If the alternatives are within a subpattern
983 .\" HTML <a href="#subpattern">
984 .\" </a>
985 (defined below),
986 .\"
987 "succeeds" means matching the rest of the main pattern as well as the
988 alternative in the subpattern.
989 .
990 .
991 .SH "INTERNAL OPTION SETTING"
992 .rs
993 .sp
994 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
995 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
996 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
997 The option letters are
998 .sp
999 i for PCRE_CASELESS
1000 m for PCRE_MULTILINE
1001 s for PCRE_DOTALL
1002 x for PCRE_EXTENDED
1003 .sp
1004 For example, (?im) sets caseless, multiline matching. It is also possible to
1005 unset these options by preceding the letter with a hyphen, and a combined
1006 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1007 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1008 permitted. If a letter appears both before and after the hyphen, the option is
1009 unset.
1010 .P
1011 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1012 changed in the same way as the Perl-compatible options by using the characters
1013 J, U and X respectively.
1014 .P
1015 When an option change occurs at top level (that is, not inside subpattern
1016 parentheses), the change applies to the remainder of the pattern that follows.
1017 If the change is placed right at the start of a pattern, PCRE extracts it into
1018 the global options (and it will therefore show up in data extracted by the
1019 \fBpcre_fullinfo()\fP function).
1020 .P
1021 An option change within a subpattern (see below for a description of
1022 subpatterns) affects only that part of the current pattern that follows it, so
1023 .sp
1024 (a(?i)b)c
1025 .sp
1026 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1027 By this means, options can be made to have different settings in different
1028 parts of the pattern. Any changes made in one alternative do carry on
1029 into subsequent branches within the same subpattern. For example,
1030 .sp
1031 (a(?i)b|c)
1032 .sp
1033 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1034 branch is abandoned before the option setting. This is because the effects of
1035 option settings happen at compile time. There would be some very weird
1036 behaviour otherwise.
1037 .
1038 .
1039 .\" HTML <a name="subpattern"></a>
1040 .SH SUBPATTERNS
1041 .rs
1042 .sp
1043 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1044 Turning part of a pattern into a subpattern does two things:
1045 .sp
1046 1. It localizes a set of alternatives. For example, the pattern
1047 .sp
1048 cat(aract|erpillar|)
1049 .sp
1050 matches one of the words "cat", "cataract", or "caterpillar". Without the
1051 parentheses, it would match "cataract", "erpillar" or an empty string.
1052 .sp
1053 2. It sets up the subpattern as a capturing subpattern. This means that, when
1054 the whole pattern matches, that portion of the subject string that matched the
1055 subpattern is passed back to the caller via the \fIovector\fP argument of
1056 \fBpcre_exec()\fP. Opening parentheses are counted from left to right (starting
1057 from 1) to obtain numbers for the capturing subpatterns.
1058 .P
1059 For example, if the string "the red king" is matched against the pattern
1060 .sp
1061 the ((red|white) (king|queen))
1062 .sp
1063 the captured substrings are "red king", "red", and "king", and are numbered 1,
1064 2, and 3, respectively.
1065 .P
1066 The fact that plain parentheses fulfil two functions is not always helpful.
1067 There are often times when a grouping subpattern is required without a
1068 capturing requirement. If an opening parenthesis is followed by a question mark
1069 and a colon, the subpattern does not do any capturing, and is not counted when
1070 computing the number of any subsequent capturing subpatterns. For example, if
1071 the string "the white queen" is matched against the pattern
1072 .sp
1073 the ((?:red|white) (king|queen))
1074 .sp
1075 the captured substrings are "white queen" and "queen", and are numbered 1 and
1076 2. The maximum number of capturing subpatterns is 65535.
1077 .P
1078 As a convenient shorthand, if any option settings are required at the start of
1079 a non-capturing subpattern, the option letters may appear between the "?" and
1080 the ":". Thus the two patterns
1081 .sp
1082 (?i:saturday|sunday)
1083 (?:(?i)saturday|sunday)
1084 .sp
1085 match exactly the same set of strings. Because alternative branches are tried
1086 from left to right, and options are not reset until the end of the subpattern
1087 is reached, an option setting in one branch does affect subsequent branches, so
1088 the above patterns match "SUNDAY" as well as "Saturday".
1089 .
1090 .
1091 .SH "DUPLICATE SUBPATTERN NUMBERS"
1092 .rs
1093 .sp
1094 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1095 the same numbers for its capturing parentheses. Such a subpattern starts with
1096 (?| and is itself a non-capturing subpattern. For example, consider this
1097 pattern:
1098 .sp
1099 (?|(Sat)ur|(Sun))day
1100 .sp
1101 Because the two alternatives are inside a (?| group, both sets of capturing
1102 parentheses are numbered one. Thus, when the pattern matches, you can look
1103 at captured substring number one, whichever alternative matched. This construct
1104 is useful when you want to capture part, but not all, of one of a number of
1105 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1106 number is reset at the start of each branch. The numbers of any capturing
1107 buffers that follow the subpattern start after the highest number used in any
1108 branch. The following example is taken from the Perl documentation.
1109 The numbers underneath show in which buffer the captured content will be
1110 stored.
1111 .sp
1112 # before ---------------branch-reset----------- after
1113 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1114 # 1 2 2 3 2 3 4
1115 .sp
1116 A backreference or a recursive call to a numbered subpattern always refers to
1117 the first one in the pattern with the given number.
1118 .P
1119 An alternative approach to using this "branch reset" feature is to use
1120 duplicate named subpatterns, as described in the next section.
1121 .
1122 .
1123 .SH "NAMED SUBPATTERNS"
1124 .rs
1125 .sp
1126 Identifying capturing parentheses by number is simple, but it can be very hard
1127 to keep track of the numbers in complicated regular expressions. Furthermore,
1128 if an expression is modified, the numbers may change. To help with this
1129 difficulty, PCRE supports the naming of subpatterns. This feature was not
1130 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1131 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1132 the Perl and the Python syntax.
1133 .P
1134 In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1135 (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1136 parentheses from other parts of the pattern, such as
1137 .\" HTML <a href="#backreferences">
1138 .\" </a>
1139 backreferences,
1140 .\"
1141 .\" HTML <a href="#recursion">
1142 .\" </a>
1143 recursion,
1144 .\"
1145 and
1146 .\" HTML <a href="#conditions">
1147 .\" </a>
1148 conditions,
1149 .\"
1150 can be made by name as well as by number.
1151 .P
1152 Names consist of up to 32 alphanumeric characters and underscores. Named
1153 capturing parentheses are still allocated numbers as well as names, exactly as
1154 if the names were not present. The PCRE API provides function calls for
1155 extracting the name-to-number translation table from a compiled pattern. There
1156 is also a convenience function for extracting a captured substring by name.
1157 .P
1158 By default, a name must be unique within a pattern, but it is possible to relax
1159 this constraint by setting the PCRE_DUPNAMES option at compile time. This can
1160 be useful for patterns where only one instance of the named parentheses can
1161 match. Suppose you want to match the name of a weekday, either as a 3-letter
1162 abbreviation or as the full name, and in both cases you want to extract the
1163 abbreviation. This pattern (ignoring the line breaks) does the job:
1164 .sp
1165 (?<DN>Mon|Fri|Sun)(?:day)?|
1166 (?<DN>Tue)(?:sday)?|
1167 (?<DN>Wed)(?:nesday)?|
1168 (?<DN>Thu)(?:rsday)?|
1169 (?<DN>Sat)(?:urday)?
1170 .sp
1171 There are five capturing substrings, but only one is ever set after a match.
1172 (An alternative way of solving this problem is to use a "branch reset"
1173 subpattern, as described in the previous section.)
1174 .P
1175 The convenience function for extracting the data by name returns the substring
1176 for the first (and in this example, the only) subpattern of that name that
1177 matched. This saves searching to find which numbered subpattern it was. If you
1178 make a reference to a non-unique named subpattern from elsewhere in the
1179 pattern, the one that corresponds to the lowest number is used. For further
1180 details of the interfaces for handling named subpatterns, see the
1181 .\" HREF
1182 \fBpcreapi\fP
1183 .\"
1184 documentation.
1185 .
1186 .
1187 .SH REPETITION
1188 .rs
1189 .sp
1190 Repetition is specified by quantifiers, which can follow any of the following
1191 items:
1192 .sp
1193 a literal data character
1194 the dot metacharacter
1195 the \eC escape sequence
1196 the \eX escape sequence (in UTF-8 mode with Unicode properties)
1197 the \eR escape sequence
1198 an escape such as \ed that matches a single character
1199 a character class
1200 a back reference (see next section)
1201 a parenthesized subpattern (unless it is an assertion)
1202 .sp
1203 The general repetition quantifier specifies a minimum and maximum number of
1204 permitted matches, by giving the two numbers in curly brackets (braces),
1205 separated by a comma. The numbers must be less than 65536, and the first must
1206 be less than or equal to the second. For example:
1207 .sp
1208 z{2,4}
1209 .sp
1210 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1211 character. If the second number is omitted, but the comma is present, there is
1212 no upper limit; if the second number and the comma are both omitted, the
1213 quantifier specifies an exact number of required matches. Thus
1214 .sp
1215 [aeiou]{3,}
1216 .sp
1217 matches at least 3 successive vowels, but may match many more, while
1218 .sp
1219 \ed{8}
1220 .sp
1221 matches exactly 8 digits. An opening curly bracket that appears in a position
1222 where a quantifier is not allowed, or one that does not match the syntax of a
1223 quantifier, is taken as a literal character. For example, {,6} is not a
1224 quantifier, but a literal string of four characters.
1225 .P
1226 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1227 bytes. Thus, for example, \ex{100}{2} matches two UTF-8 characters, each of
1228 which is represented by a two-byte sequence. Similarly, when Unicode property
1229 support is available, \eX{3} matches three Unicode extended sequences, each of
1230 which may be several bytes long (and they may be of different lengths).
1231 .P
1232 The quantifier {0} is permitted, causing the expression to behave as if the
1233 previous item and the quantifier were not present.
1234 .P
1235 For convenience, the three most common quantifiers have single-character
1236 abbreviations:
1237 .sp
1238 * is equivalent to {0,}
1239 + is equivalent to {1,}
1240 ? is equivalent to {0,1}
1241 .sp
1242 It is possible to construct infinite loops by following a subpattern that can
1243 match no characters with a quantifier that has no upper limit, for example:
1244 .sp
1245 (a?)*
1246 .sp
1247 Earlier versions of Perl and PCRE used to give an error at compile time for
1248 such patterns. However, because there are cases where this can be useful, such
1249 patterns are now accepted, but if any repetition of the subpattern does in fact
1250 match no characters, the loop is forcibly broken.
1251 .P
1252 By default, the quantifiers are "greedy", that is, they match as much as
1253 possible (up to the maximum number of permitted times), without causing the
1254 rest of the pattern to fail. The classic example of where this gives problems
1255 is in trying to match comments in C programs. These appear between /* and */
1256 and within the comment, individual * and / characters may appear. An attempt to
1257 match C comments by applying the pattern
1258 .sp
1259 /\e*.*\e*/
1260 .sp
1261 to the string
1262 .sp
1263 /* first comment */ not comment /* second comment */
1264 .sp
1265 fails, because it matches the entire string owing to the greediness of the .*
1266 item.
1267 .P
1268 However, if a quantifier is followed by a question mark, it ceases to be
1269 greedy, and instead matches the minimum number of times possible, so the
1270 pattern
1271 .sp
1272 /\e*.*?\e*/
1273 .sp
1274 does the right thing with the C comments. The meaning of the various
1275 quantifiers is not otherwise changed, just the preferred number of matches.
1276 Do not confuse this use of question mark with its use as a quantifier in its
1277 own right. Because it has two uses, it can sometimes appear doubled, as in
1278 .sp
1279 \ed??\ed
1280 .sp
1281 which matches one digit by preference, but can match two if that is the only
1282 way the rest of the pattern matches.
1283 .P
1284 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1285 the quantifiers are not greedy by default, but individual ones can be made
1286 greedy by following them with a question mark. In other words, it inverts the
1287 default behaviour.
1288 .P
1289 When a parenthesized subpattern is quantified with a minimum repeat count that
1290 is greater than 1 or with a limited maximum, more memory is required for the
1291 compiled pattern, in proportion to the size of the minimum or maximum.
1292 .P
1293 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1294 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1295 implicitly anchored, because whatever follows will be tried against every
1296 character position in the subject string, so there is no point in retrying the
1297 overall match at any position after the first. PCRE normally treats such a
1298 pattern as though it were preceded by \eA.
1299 .P
1300 In cases where it is known that the subject string contains no newlines, it is
1301 worth setting PCRE_DOTALL in order to obtain this optimization, or
1302 alternatively using ^ to indicate anchoring explicitly.
1303 .P
1304 However, there is one situation where the optimization cannot be used. When .*
1305 is inside capturing parentheses that are the subject of a backreference
1306 elsewhere in the pattern, a match at the start may fail where a later one
1307 succeeds. Consider, for example:
1308 .sp
1309 (.*)abc\e1
1310 .sp
1311 If the subject is "xyz123abc123" the match point is the fourth character. For
1312 this reason, such a pattern is not implicitly anchored.
1313 .P
1314 When a capturing subpattern is repeated, the value captured is the substring
1315 that matched the final iteration. For example, after
1316 .sp
1317 (tweedle[dume]{3}\es*)+
1318 .sp
1319 has matched "tweedledum tweedledee" the value of the captured substring is
1320 "tweedledee". However, if there are nested capturing subpatterns, the
1321 corresponding captured values may have been set in previous iterations. For
1322 example, after
1323 .sp
1324 /(a|(b))+/
1325 .sp
1326 matches "aba" the value of the second captured substring is "b".
1327 .
1328 .
1329 .\" HTML <a name="atomicgroup"></a>
1330 .SH "ATOMIC GROUPING AND POSSESSIVE QUANTIFIERS"
1331 .rs
1332 .sp
1333 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1334 repetition, failure of what follows normally causes the repeated item to be
1335 re-evaluated to see if a different number of repeats allows the rest of the
1336 pattern to match. Sometimes it is useful to prevent this, either to change the
1337 nature of the match, or to cause it fail earlier than it otherwise might, when
1338 the author of the pattern knows there is no point in carrying on.
1339 .P
1340 Consider, for example, the pattern \ed+foo when applied to the subject line
1341 .sp
1342 123456bar
1343 .sp
1344 After matching all 6 digits and then failing to match "foo", the normal
1345 action of the matcher is to try again with only 5 digits matching the \ed+
1346 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1347 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1348 that once a subpattern has matched, it is not to be re-evaluated in this way.
1349 .P
1350 If we use atomic grouping for the previous example, the matcher gives up
1351 immediately on failing to match "foo" the first time. The notation is a kind of
1352 special parenthesis, starting with (?> as in this example:
1353 .sp
1354 (?>\ed+)foo
1355 .sp
1356 This kind of parenthesis "locks up" the part of the pattern it contains once
1357 it has matched, and a failure further into the pattern is prevented from
1358 backtracking into it. Backtracking past it to previous items, however, works as
1359 normal.
1360 .P
1361 An alternative description is that a subpattern of this type matches the string
1362 of characters that an identical standalone pattern would match, if anchored at
1363 the current point in the subject string.
1364 .P
1365 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1366 the above example can be thought of as a maximizing repeat that must swallow
1367 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1368 number of digits they match in order to make the rest of the pattern match,
1369 (?>\ed+) can only match an entire sequence of digits.
1370 .P
1371 Atomic groups in general can of course contain arbitrarily complicated
1372 subpatterns, and can be nested. However, when the subpattern for an atomic
1373 group is just a single repeated item, as in the example above, a simpler
1374 notation, called a "possessive quantifier" can be used. This consists of an
1375 additional + character following a quantifier. Using this notation, the
1376 previous example can be rewritten as
1377 .sp
1378 \ed++foo
1379 .sp
1380 Note that a possessive quantifier can be used with an entire group, for
1381 example:
1382 .sp
1383 (abc|xyz){2,3}+
1384 .sp
1385 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1386 option is ignored. They are a convenient notation for the simpler forms of
1387 atomic group. However, there is no difference in the meaning of a possessive
1388 quantifier and the equivalent atomic group, though there may be a performance
1389 difference; possessive quantifiers should be slightly faster.
1390 .P
1391 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1392 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1393 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1394 package, and PCRE copied it from there. It ultimately found its way into Perl
1395 at release 5.10.
1396 .P
1397 PCRE has an optimization that automatically "possessifies" certain simple
1398 pattern constructs. For example, the sequence A+B is treated as A++B because
1399 there is no point in backtracking into a sequence of A's when B must follow.
1400 .P
1401 When a pattern contains an unlimited repeat inside a subpattern that can itself
1402 be repeated an unlimited number of times, the use of an atomic group is the
1403 only way to avoid some failing matches taking a very long time indeed. The
1404 pattern
1405 .sp
1406 (\eD+|<\ed+>)*[!?]
1407 .sp
1408 matches an unlimited number of substrings that either consist of non-digits, or
1409 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1410 quickly. However, if it is applied to
1411 .sp
1412 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1413 .sp
1414 it takes a long time before reporting failure. This is because the string can
1415 be divided between the internal \eD+ repeat and the external * repeat in a
1416 large number of ways, and all have to be tried. (The example uses [!?] rather
1417 than a single character at the end, because both PCRE and Perl have an
1418 optimization that allows for fast failure when a single character is used. They
1419 remember the last single character that is required for a match, and fail early
1420 if it is not present in the string.) If the pattern is changed so that it uses
1421 an atomic group, like this:
1422 .sp
1423 ((?>\eD+)|<\ed+>)*[!?]
1424 .sp
1425 sequences of non-digits cannot be broken, and failure happens quickly.
1426 .
1427 .
1428 .\" HTML <a name="backreferences"></a>
1429 .SH "BACK REFERENCES"
1430 .rs
1431 .sp
1432 Outside a character class, a backslash followed by a digit greater than 0 (and
1433 possibly further digits) is a back reference to a capturing subpattern earlier
1434 (that is, to its left) in the pattern, provided there have been that many
1435 previous capturing left parentheses.
1436 .P
1437 However, if the decimal number following the backslash is less than 10, it is
1438 always taken as a back reference, and causes an error only if there are not
1439 that many capturing left parentheses in the entire pattern. In other words, the
1440 parentheses that are referenced need not be to the left of the reference for
1441 numbers less than 10. A "forward back reference" of this type can make sense
1442 when a repetition is involved and the subpattern to the right has participated
1443 in an earlier iteration.
1444 .P
1445 It is not possible to have a numerical "forward back reference" to a subpattern
1446 whose number is 10 or more using this syntax because a sequence such as \e50 is
1447 interpreted as a character defined in octal. See the subsection entitled
1448 "Non-printing characters"
1449 .\" HTML <a href="#digitsafterbackslash">
1450 .\" </a>
1451 above
1452 .\"
1453 for further details of the handling of digits following a backslash. There is
1454 no such problem when named parentheses are used. A back reference to any
1455 subpattern is possible using named parentheses (see below).
1456 .P
1457 Another way of avoiding the ambiguity inherent in the use of digits following a
1458 backslash is to use the \eg escape sequence, which is a feature introduced in
1459 Perl 5.10. This escape must be followed by an unsigned number or a negative
1460 number, optionally enclosed in braces. These examples are all identical:
1461 .sp
1462 (ring), \e1
1463 (ring), \eg1
1464 (ring), \eg{1}
1465 .sp
1466 An unsigned number specifies an absolute reference without the ambiguity that
1467 is present in the older syntax. It is also useful when literal digits follow
1468 the reference. A negative number is a relative reference. Consider this
1469 example:
1470 .sp
1471 (abc(def)ghi)\eg{-1}
1472 .sp
1473 The sequence \eg{-1} is a reference to the most recently started capturing
1474 subpattern before \eg, that is, is it equivalent to \e2. Similarly, \eg{-2}
1475 would be equivalent to \e1. The use of relative references can be helpful in
1476 long patterns, and also in patterns that are created by joining together
1477 fragments that contain references within themselves.
1478 .P
1479 A back reference matches whatever actually matched the capturing subpattern in
1480 the current subject string, rather than anything matching the subpattern
1481 itself (see
1482 .\" HTML <a href="#subpatternsassubroutines">
1483 .\" </a>
1484 "Subpatterns as subroutines"
1485 .\"
1486 below for a way of doing that). So the pattern
1487 .sp
1488 (sens|respons)e and \e1ibility
1489 .sp
1490 matches "sense and sensibility" and "response and responsibility", but not
1491 "sense and responsibility". If caseful matching is in force at the time of the
1492 back reference, the case of letters is relevant. For example,
1493 .sp
1494 ((?i)rah)\es+\e1
1495 .sp
1496 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1497 capturing subpattern is matched caselessly.
1498 .P
1499 There are several different ways of writing back references to named
1500 subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1501 \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1502 back reference syntax, in which \eg can be used for both numeric and named
1503 references, is also supported. We could rewrite the above example in any of
1504 the following ways:
1505 .sp
1506 (?<p1>(?i)rah)\es+\ek<p1>
1507 (?'p1'(?i)rah)\es+\ek{p1}
1508 (?P<p1>(?i)rah)\es+(?P=p1)
1509 (?<p1>(?i)rah)\es+\eg{p1}
1510 .sp
1511 A subpattern that is referenced by name may appear in the pattern before or
1512 after the reference.
1513 .P
1514 There may be more than one back reference to the same subpattern. If a
1515 subpattern has not actually been used in a particular match, any back
1516 references to it always fail. For example, the pattern
1517 .sp
1518 (a|(bc))\e2
1519 .sp
1520 always fails if it starts to match "a" rather than "bc". Because there may be
1521 many capturing parentheses in a pattern, all digits following the backslash are
1522 taken as part of a potential back reference number. If the pattern continues
1523 with a digit character, some delimiter must be used to terminate the back
1524 reference. If the PCRE_EXTENDED option is set, this can be whitespace.
1525 Otherwise an empty comment (see
1526 .\" HTML <a href="#comments">
1527 .\" </a>
1528 "Comments"
1529 .\"
1530 below) can be used.
1531 .P
1532 A back reference that occurs inside the parentheses to which it refers fails
1533 when the subpattern is first used, so, for example, (a\e1) never matches.
1534 However, such references can be useful inside repeated subpatterns. For
1535 example, the pattern
1536 .sp
1537 (a|b\e1)+
1538 .sp
1539 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1540 the subpattern, the back reference matches the character string corresponding
1541 to the previous iteration. In order for this to work, the pattern must be such
1542 that the first iteration does not need to match the back reference. This can be
1543 done using alternation, as in the example above, or by a quantifier with a
1544 minimum of zero.
1545 .
1546 .
1547 .\" HTML <a name="bigassertions"></a>
1548 .SH ASSERTIONS
1549 .rs
1550 .sp
1551 An assertion is a test on the characters following or preceding the current
1552 matching point that does not actually consume any characters. The simple
1553 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1554 .\" HTML <a href="#smallassertions">
1555 .\" </a>
1556 above.
1557 .\"
1558 .P
1559 More complicated assertions are coded as subpatterns. There are two kinds:
1560 those that look ahead of the current position in the subject string, and those
1561 that look behind it. An assertion subpattern is matched in the normal way,
1562 except that it does not cause the current matching position to be changed.
1563 .P
1564 Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1565 because it makes no sense to assert the same thing several times. If any kind
1566 of assertion contains capturing subpatterns within it, these are counted for
1567 the purposes of numbering the capturing subpatterns in the whole pattern.
1568 However, substring capturing is carried out only for positive assertions,
1569 because it does not make sense for negative assertions.
1570 .
1571 .
1572 .SS "Lookahead assertions"
1573 .rs
1574 .sp
1575 Lookahead assertions start with (?= for positive assertions and (?! for
1576 negative assertions. For example,
1577 .sp
1578 \ew+(?=;)
1579 .sp
1580 matches a word followed by a semicolon, but does not include the semicolon in
1581 the match, and
1582 .sp
1583 foo(?!bar)
1584 .sp
1585 matches any occurrence of "foo" that is not followed by "bar". Note that the
1586 apparently similar pattern
1587 .sp
1588 (?!foo)bar
1589 .sp
1590 does not find an occurrence of "bar" that is preceded by something other than
1591 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1592 (?!foo) is always true when the next three characters are "bar". A
1593 lookbehind assertion is needed to achieve the other effect.
1594 .P
1595 If you want to force a matching failure at some point in a pattern, the most
1596 convenient way to do it is with (?!) because an empty string always matches, so
1597 an assertion that requires there not to be an empty string must always fail.
1598 .
1599 .
1600 .\" HTML <a name="lookbehind"></a>
1601 .SS "Lookbehind assertions"
1602 .rs
1603 .sp
1604 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1605 negative assertions. For example,
1606 .sp
1607 (?<!foo)bar
1608 .sp
1609 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1610 a lookbehind assertion are restricted such that all the strings it matches must
1611 have a fixed length. However, if there are several top-level alternatives, they
1612 do not all have to have the same fixed length. Thus
1613 .sp
1614 (?<=bullock|donkey)
1615 .sp
1616 is permitted, but
1617 .sp
1618 (?<!dogs?|cats?)
1619 .sp
1620 causes an error at compile time. Branches that match different length strings
1621 are permitted only at the top level of a lookbehind assertion. This is an
1622 extension compared with Perl (at least for 5.8), which requires all branches to
1623 match the same length of string. An assertion such as
1624 .sp
1625 (?<=ab(c|de))
1626 .sp
1627 is not permitted, because its single top-level branch can match two different
1628 lengths, but it is acceptable if rewritten to use two top-level branches:
1629 .sp
1630 (?<=abc|abde)
1631 .sp
1632 In some cases, the Perl 5.10 escape sequence \eK
1633 .\" HTML <a href="#resetmatchstart">
1634 .\" </a>
1635 (see above)
1636 .\"
1637 can be used instead of a lookbehind assertion; this is not restricted to a
1638 fixed-length.
1639 .P
1640 The implementation of lookbehind assertions is, for each alternative, to
1641 temporarily move the current position back by the fixed length and then try to
1642 match. If there are insufficient characters before the current position, the
1643 assertion fails.
1644 .P
1645 PCRE does not allow the \eC escape (which matches a single byte in UTF-8 mode)
1646 to appear in lookbehind assertions, because it makes it impossible to calculate
1647 the length of the lookbehind. The \eX and \eR escapes, which can match
1648 different numbers of bytes, are also not permitted.
1649 .P
1650 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1651 specify efficient matching at the end of the subject string. Consider a simple
1652 pattern such as
1653 .sp
1654 abcd$
1655 .sp
1656 when applied to a long string that does not match. Because matching proceeds
1657 from left to right, PCRE will look for each "a" in the subject and then see if
1658 what follows matches the rest of the pattern. If the pattern is specified as
1659 .sp
1660 ^.*abcd$
1661 .sp
1662 the initial .* matches the entire string at first, but when this fails (because
1663 there is no following "a"), it backtracks to match all but the last character,
1664 then all but the last two characters, and so on. Once again the search for "a"
1665 covers the entire string, from right to left, so we are no better off. However,
1666 if the pattern is written as
1667 .sp
1668 ^.*+(?<=abcd)
1669 .sp
1670 there can be no backtracking for the .*+ item; it can match only the entire
1671 string. The subsequent lookbehind assertion does a single test on the last four
1672 characters. If it fails, the match fails immediately. For long strings, this
1673 approach makes a significant difference to the processing time.
1674 .
1675 .
1676 .SS "Using multiple assertions"
1677 .rs
1678 .sp
1679 Several assertions (of any sort) may occur in succession. For example,
1680 .sp
1681 (?<=\ed{3})(?<!999)foo
1682 .sp
1683 matches "foo" preceded by three digits that are not "999". Notice that each of
1684 the assertions is applied independently at the same point in the subject
1685 string. First there is a check that the previous three characters are all
1686 digits, and then there is a check that the same three characters are not "999".
1687 This pattern does \fInot\fP match "foo" preceded by six characters, the first
1688 of which are digits and the last three of which are not "999". For example, it
1689 doesn't match "123abcfoo". A pattern to do that is
1690 .sp
1691 (?<=\ed{3}...)(?<!999)foo
1692 .sp
1693 This time the first assertion looks at the preceding six characters, checking
1694 that the first three are digits, and then the second assertion checks that the
1695 preceding three characters are not "999".
1696 .P
1697 Assertions can be nested in any combination. For example,
1698 .sp
1699 (?<=(?<!foo)bar)baz
1700 .sp
1701 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1702 preceded by "foo", while
1703 .sp
1704 (?<=\ed{3}(?!999)...)foo
1705 .sp
1706 is another pattern that matches "foo" preceded by three digits and any three
1707 characters that are not "999".
1708 .
1709 .
1710 .\" HTML <a name="conditions"></a>
1711 .SH "CONDITIONAL SUBPATTERNS"
1712 .rs
1713 .sp
1714 It is possible to cause the matching process to obey a subpattern
1715 conditionally or to choose between two alternative subpatterns, depending on
1716 the result of an assertion, or whether a previous capturing subpattern matched
1717 or not. The two possible forms of conditional subpattern are
1718 .sp
1719 (?(condition)yes-pattern)
1720 (?(condition)yes-pattern|no-pattern)
1721 .sp
1722 If the condition is satisfied, the yes-pattern is used; otherwise the
1723 no-pattern (if present) is used. If there are more than two alternatives in the
1724 subpattern, a compile-time error occurs.
1725 .P
1726 There are four kinds of condition: references to subpatterns, references to
1727 recursion, a pseudo-condition called DEFINE, and assertions.
1728 .
1729 .SS "Checking for a used subpattern by number"
1730 .rs
1731 .sp
1732 If the text between the parentheses consists of a sequence of digits, the
1733 condition is true if the capturing subpattern of that number has previously
1734 matched. An alternative notation is to precede the digits with a plus or minus
1735 sign. In this case, the subpattern number is relative rather than absolute.
1736 The most recently opened parentheses can be referenced by (?(-1), the next most
1737 recent by (?(-2), and so on. In looping constructs it can also make sense to
1738 refer to subsequent groups with constructs such as (?(+2).
1739 .P
1740 Consider the following pattern, which contains non-significant white space to
1741 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
1742 three parts for ease of discussion:
1743 .sp
1744 ( \e( )? [^()]+ (?(1) \e) )
1745 .sp
1746 The first part matches an optional opening parenthesis, and if that
1747 character is present, sets it as the first captured substring. The second part
1748 matches one or more characters that are not parentheses. The third part is a
1749 conditional subpattern that tests whether the first set of parentheses matched
1750 or not. If they did, that is, if subject started with an opening parenthesis,
1751 the condition is true, and so the yes-pattern is executed and a closing
1752 parenthesis is required. Otherwise, since no-pattern is not present, the
1753 subpattern matches nothing. In other words, this pattern matches a sequence of
1754 non-parentheses, optionally enclosed in parentheses.
1755 .P
1756 If you were embedding this pattern in a larger one, you could use a relative
1757 reference:
1758 .sp
1759 ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
1760 .sp
1761 This makes the fragment independent of the parentheses in the larger pattern.
1762 .
1763 .SS "Checking for a used subpattern by name"
1764 .rs
1765 .sp
1766 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
1767 subpattern by name. For compatibility with earlier versions of PCRE, which had
1768 this facility before Perl, the syntax (?(name)...) is also recognized. However,
1769 there is a possible ambiguity with this syntax, because subpattern names may
1770 consist entirely of digits. PCRE looks first for a named subpattern; if it
1771 cannot find one and the name consists entirely of digits, PCRE looks for a
1772 subpattern of that number, which must be greater than zero. Using subpattern
1773 names that consist entirely of digits is not recommended.
1774 .P
1775 Rewriting the above example to use a named subpattern gives this:
1776 .sp
1777 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
1778 .sp
1779 .
1780 .SS "Checking for pattern recursion"
1781 .rs
1782 .sp
1783 If the condition is the string (R), and there is no subpattern with the name R,
1784 the condition is true if a recursive call to the whole pattern or any
1785 subpattern has been made. If digits or a name preceded by ampersand follow the
1786 letter R, for example:
1787 .sp
1788 (?(R3)...) or (?(R&name)...)
1789 .sp
1790 the condition is true if the most recent recursion is into the subpattern whose
1791 number or name is given. This condition does not check the entire recursion
1792 stack.
1793 .P
1794 At "top level", all these recursion test conditions are false. Recursive
1795 patterns are described below.
1796 .
1797 .SS "Defining subpatterns for use by reference only"
1798 .rs
1799 .sp
1800 If the condition is the string (DEFINE), and there is no subpattern with the
1801 name DEFINE, the condition is always false. In this case, there may be only one
1802 alternative in the subpattern. It is always skipped if control reaches this
1803 point in the pattern; the idea of DEFINE is that it can be used to define
1804 "subroutines" that can be referenced from elsewhere. (The use of "subroutines"
1805 is described below.) For example, a pattern to match an IPv4 address could be
1806 written like this (ignore whitespace and line breaks):
1807 .sp
1808 (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
1809 \eb (?&byte) (\e.(?&byte)){3} \eb
1810 .sp
1811 The first part of the pattern is a DEFINE group inside which a another group
1812 named "byte" is defined. This matches an individual component of an IPv4
1813 address (a number less than 256). When matching takes place, this part of the
1814 pattern is skipped because DEFINE acts like a false condition.
1815 .P
1816 The rest of the pattern uses references to the named group to match the four
1817 dot-separated components of an IPv4 address, insisting on a word boundary at
1818 each end.
1819 .
1820 .SS "Assertion conditions"
1821 .rs
1822 .sp
1823 If the condition is not in any of the above formats, it must be an assertion.
1824 This may be a positive or negative lookahead or lookbehind assertion. Consider
1825 this pattern, again containing non-significant white space, and with the two
1826 alternatives on the second line:
1827 .sp
1828 (?(?=[^a-z]*[a-z])
1829 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
1830 .sp
1831 The condition is a positive lookahead assertion that matches an optional
1832 sequence of non-letters followed by a letter. In other words, it tests for the
1833 presence of at least one letter in the subject. If a letter is found, the
1834 subject is matched against the first alternative; otherwise it is matched
1835 against the second. This pattern matches strings in one of the two forms
1836 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
1837 .
1838 .
1839 .\" HTML <a name="comments"></a>
1840 .SH COMMENTS
1841 .rs
1842 .sp
1843 The sequence (?# marks the start of a comment that continues up to the next
1844 closing parenthesis. Nested parentheses are not permitted. The characters
1845 that make up a comment play no part in the pattern matching at all.
1846 .P
1847 If the PCRE_EXTENDED option is set, an unescaped # character outside a
1848 character class introduces a comment that continues to immediately after the
1849 next newline in the pattern.
1850 .
1851 .
1852 .\" HTML <a name="recursion"></a>
1853 .SH "RECURSIVE PATTERNS"
1854 .rs
1855 .sp
1856 Consider the problem of matching a string in parentheses, allowing for
1857 unlimited nested parentheses. Without the use of recursion, the best that can
1858 be done is to use a pattern that matches up to some fixed depth of nesting. It
1859 is not possible to handle an arbitrary nesting depth.
1860 .P
1861 For some time, Perl has provided a facility that allows regular expressions to
1862 recurse (amongst other things). It does this by interpolating Perl code in the
1863 expression at run time, and the code can refer to the expression itself. A Perl
1864 pattern using code interpolation to solve the parentheses problem can be
1865 created like this:
1866 .sp
1867 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
1868 .sp
1869 The (?p{...}) item interpolates Perl code at run time, and in this case refers
1870 recursively to the pattern in which it appears.
1871 .P
1872 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
1873 supports special syntax for recursion of the entire pattern, and also for
1874 individual subpattern recursion. After its introduction in PCRE and Python,
1875 this kind of recursion was introduced into Perl at release 5.10.
1876 .P
1877 A special item that consists of (? followed by a number greater than zero and a
1878 closing parenthesis is a recursive call of the subpattern of the given number,
1879 provided that it occurs inside that subpattern. (If not, it is a "subroutine"
1880 call, which is described in the next section.) The special item (?R) or (?0) is
1881 a recursive call of the entire regular expression.
1882 .P
1883 In PCRE (like Python, but unlike Perl), a recursive subpattern call is always
1884 treated as an atomic group. That is, once it has matched some of the subject
1885 string, it is never re-entered, even if it contains untried alternatives and
1886 there is a subsequent matching failure.
1887 .P
1888 This PCRE pattern solves the nested parentheses problem (assume the
1889 PCRE_EXTENDED option is set so that white space is ignored):
1890 .sp
1891 \e( ( (?>[^()]+) | (?R) )* \e)
1892 .sp
1893 First it matches an opening parenthesis. Then it matches any number of
1894 substrings which can either be a sequence of non-parentheses, or a recursive
1895 match of the pattern itself (that is, a correctly parenthesized substring).
1896 Finally there is a closing parenthesis.
1897 .P
1898 If this were part of a larger pattern, you would not want to recurse the entire
1899 pattern, so instead you could use this:
1900 .sp
1901 ( \e( ( (?>[^()]+) | (?1) )* \e) )
1902 .sp
1903 We have put the pattern into parentheses, and caused the recursion to refer to
1904 them instead of the whole pattern.
1905 .P
1906 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
1907 is made easier by the use of relative references. (A Perl 5.10 feature.)
1908 Instead of (?1) in the pattern above you can write (?-2) to refer to the second
1909 most recently opened parentheses preceding the recursion. In other words, a
1910 negative number counts capturing parentheses leftwards from the point at which
1911 it is encountered.
1912 .P
1913 It is also possible to refer to subsequently opened parentheses, by writing
1914 references such as (?+2). However, these cannot be recursive because the
1915 reference is not inside the parentheses that are referenced. They are always
1916 "subroutine" calls, as described in the next section.
1917 .P
1918 An alternative approach is to use named parentheses instead. The Perl syntax
1919 for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
1920 could rewrite the above example as follows:
1921 .sp
1922 (?<pn> \e( ( (?>[^()]+) | (?&pn) )* \e) )
1923 .sp
1924 If there is more than one subpattern with the same name, the earliest one is
1925 used.
1926 .P
1927 This particular example pattern that we have been looking at contains nested
1928 unlimited repeats, and so the use of atomic grouping for matching strings of
1929 non-parentheses is important when applying the pattern to strings that do not
1930 match. For example, when this pattern is applied to
1931 .sp
1932 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
1933 .sp
1934 it yields "no match" quickly. However, if atomic grouping is not used,
1935 the match runs for a very long time indeed because there are so many different
1936 ways the + and * repeats can carve up the subject, and all have to be tested
1937 before failure can be reported.
1938 .P
1939 At the end of a match, the values set for any capturing subpatterns are those
1940 from the outermost level of the recursion at which the subpattern value is set.
1941 If you want to obtain intermediate values, a callout function can be used (see
1942 below and the
1943 .\" HREF
1944 \fBpcrecallout\fP
1945 .\"
1946 documentation). If the pattern above is matched against
1947 .sp
1948 (ab(cd)ef)
1949 .sp
1950 the value for the capturing parentheses is "ef", which is the last value taken
1951 on at the top level. If additional parentheses are added, giving
1952 .sp
1953 \e( ( ( (?>[^()]+) | (?R) )* ) \e)
1954 ^ ^
1955 ^ ^
1956 .sp
1957 the string they capture is "ab(cd)ef", the contents of the top level
1958 parentheses. If there are more than 15 capturing parentheses in a pattern, PCRE
1959 has to obtain extra memory to store data during a recursion, which it does by
1960 using \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no
1961 memory can be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
1962 .P
1963 Do not confuse the (?R) item with the condition (R), which tests for recursion.
1964 Consider this pattern, which matches text in angle brackets, allowing for
1965 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
1966 recursing), whereas any characters are permitted at the outer level.
1967 .sp
1968 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
1969 .sp
1970 In this pattern, (?(R) is the start of a conditional subpattern, with two
1971 different alternatives for the recursive and non-recursive cases. The (?R) item
1972 is the actual recursive call.
1973 .
1974 .
1975 .\" HTML <a name="subpatternsassubroutines"></a>
1976 .SH "SUBPATTERNS AS SUBROUTINES"
1977 .rs
1978 .sp
1979 If the syntax for a recursive subpattern reference (either by number or by
1980 name) is used outside the parentheses to which it refers, it operates like a
1981 subroutine in a programming language. The "called" subpattern may be defined
1982 before or after the reference. A numbered reference can be absolute or
1983 relative, as in these examples:
1984 .sp
1985 (...(absolute)...)...(?2)...
1986 (...(relative)...)...(?-1)...
1987 (...(?+1)...(relative)...
1988 .sp
1989 An earlier example pointed out that the pattern
1990 .sp
1991 (sens|respons)e and \e1ibility
1992 .sp
1993 matches "sense and sensibility" and "response and responsibility", but not
1994 "sense and responsibility". If instead the pattern
1995 .sp
1996 (sens|respons)e and (?1)ibility
1997 .sp
1998 is used, it does match "sense and responsibility" as well as the other two
1999 strings. Another example is given in the discussion of DEFINE above.
2000 .P
2001 Like recursive subpatterns, a "subroutine" call is always treated as an atomic
2002 group. That is, once it has matched some of the subject string, it is never
2003 re-entered, even if it contains untried alternatives and there is a subsequent
2004 matching failure.
2005 .P
2006 When a subpattern is used as a subroutine, processing options such as
2007 case-independence are fixed when the subpattern is defined. They cannot be
2008 changed for different calls. For example, consider this pattern:
2009 .sp
2010 (abc)(?i:(?-1))
2011 .sp
2012 It matches "abcabc". It does not match "abcABC" because the change of
2013 processing option does not affect the called subpattern.
2014 .
2015 .
2016 .SH CALLOUTS
2017 .rs
2018 .sp
2019 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2020 code to be obeyed in the middle of matching a regular expression. This makes it
2021 possible, amongst other things, to extract different substrings that match the
2022 same pair of parentheses when there is a repetition.
2023 .P
2024 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2025 code. The feature is called "callout". The caller of PCRE provides an external
2026 function by putting its entry point in the global variable \fIpcre_callout\fP.
2027 By default, this variable contains NULL, which disables all calling out.
2028 .P
2029 Within a regular expression, (?C) indicates the points at which the external
2030 function is to be called. If you want to identify different callout points, you
2031 can put a number less than 256 after the letter C. The default value is zero.
2032 For example, this pattern has two callout points:
2033 .sp
2034 (?C1)abc(?C2)def
2035 .sp
2036 If the PCRE_AUTO_CALLOUT flag is passed to \fBpcre_compile()\fP, callouts are
2037 automatically installed before each item in the pattern. They are all numbered
2038 255.
2039 .P
2040 During matching, when PCRE reaches a callout point (and \fIpcre_callout\fP is
2041 set), the external function is called. It is provided with the number of the
2042 callout, the position in the pattern, and, optionally, one item of data
2043 originally supplied by the caller of \fBpcre_exec()\fP. The callout function
2044 may cause matching to proceed, to backtrack, or to fail altogether. A complete
2045 description of the interface to the callout function is given in the
2046 .\" HREF
2047 \fBpcrecallout\fP
2048 .\"
2049 documentation.
2050 .
2051 .
2052 .SH "BACKTRACKING CONTROL"
2053 .rs
2054 .sp
2055 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2056 are described in the Perl documentation as "experimental and subject to change
2057 or removal in a future version of Perl". It goes on to say: "Their usage in
2058 production code should be noted to avoid problems during upgrades." The same
2059 remarks apply to the PCRE features described in this section.
2060 .P
2061 Since these verbs are specifically related to backtracking, they can be used
2062 only when the pattern is to be matched using \fBpcre_exec()\fP, which uses a
2063 backtracking algorithm. They cause an error if encountered by
2064 \fBpcre_dfa_exec()\fP.
2065 .P
2066 The new verbs make use of what was previously invalid syntax: an opening
2067 parenthesis followed by an asterisk. In Perl, they are generally of the form
2068 (*VERB:ARG) but PCRE does not support the use of arguments, so its general
2069 form is just (*VERB). Any number of these verbs may occur in a pattern. There
2070 are two kinds:
2071 .
2072 .SS "Verbs that act immediately"
2073 .rs
2074 .sp
2075 The following verbs act as soon as they are encountered:
2076 .sp
2077 (*ACCEPT)
2078 .sp
2079 This verb causes the match to end successfully, skipping the remainder of the
2080 pattern. When inside a recursion, only the innermost pattern is ended
2081 immediately. PCRE differs from Perl in what happens if the (*ACCEPT) is inside
2082 capturing parentheses. In Perl, the data so far is captured: in PCRE no data is
2083 captured. For example:
2084 .sp
2085 A(A|B(*ACCEPT)|C)D
2086 .sp
2087 This matches "AB", "AAD", or "ACD", but when it matches "AB", no data is
2088 captured.
2089 .sp
2090 (*FAIL) or (*F)
2091 .sp
2092 This verb causes the match to fail, forcing backtracking to occur. It is
2093 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2094 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2095 Perl features that are not present in PCRE. The nearest equivalent is the
2096 callout feature, as for example in this pattern:
2097 .sp
2098 a+(?C)(*FAIL)
2099 .sp
2100 A match with the string "aaaa" always fails, but the callout is taken before
2101 each backtrack happens (in this example, 10 times).
2102 .
2103 .SS "Verbs that act after backtracking"
2104 .rs
2105 .sp
2106 The following verbs do nothing when they are encountered. Matching continues
2107 with what follows, but if there is no subsequent match, a failure is forced.
2108 The verbs differ in exactly what kind of failure occurs.
2109 .sp
2110 (*COMMIT)
2111 .sp
2112 This verb causes the whole match to fail outright if the rest of the pattern
2113 does not match. Even if the pattern is unanchored, no further attempts to find
2114 a match by advancing the start point take place. Once (*COMMIT) has been
2115 passed, \fBpcre_exec()\fP is committed to finding a match at the current
2116 starting point, or not at all. For example:
2117 .sp
2118 a+(*COMMIT)b
2119 .sp
2120 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2121 dynamic anchor, or "I've started, so I must finish."
2122 .sp
2123 (*PRUNE)
2124 .sp
2125 This verb causes the match to fail at the current position if the rest of the
2126 pattern does not match. If the pattern is unanchored, the normal "bumpalong"
2127 advance to the next starting character then happens. Backtracking can occur as
2128 usual to the left of (*PRUNE), or when matching to the right of (*PRUNE), but
2129 if there is no match to the right, backtracking cannot cross (*PRUNE).
2130 In simple cases, the use of (*PRUNE) is just an alternative to an atomic
2131 group or possessive quantifier, but there are some uses of (*PRUNE) that cannot
2132 be expressed in any other way.
2133 .sp
2134 (*SKIP)
2135 .sp
2136 This verb is like (*PRUNE), except that if the pattern is unanchored, the
2137 "bumpalong" advance is not to the next character, but to the position in the
2138 subject where (*SKIP) was encountered. (*SKIP) signifies that whatever text
2139 was matched leading up to it cannot be part of a successful match. Consider:
2140 .sp
2141 a+(*SKIP)b
2142 .sp
2143 If the subject is "aaaac...", after the first match attempt fails (starting at
2144 the first character in the string), the starting point skips on to start the
2145 next attempt at "c". Note that a possessive quantifer does not have the same
2146 effect in this example; although it would suppress backtracking during the
2147 first match attempt, the second attempt would start at the second character
2148 instead of skipping on to "c".
2149 .sp
2150 (*THEN)
2151 .sp
2152 This verb causes a skip to the next alternation if the rest of the pattern does
2153 not match. That is, it cancels pending backtracking, but only within the
2154 current alternation. Its name comes from the observation that it can be used
2155 for a pattern-based if-then-else block:
2156 .sp
2157 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2158 .sp
2159 If the COND1 pattern matches, FOO is tried (and possibly further items after
2160 the end of the group if FOO succeeds); on failure the matcher skips to the
2161 second alternative and tries COND2, without backtracking into COND1. If (*THEN)
2162 is used outside of any alternation, it acts exactly like (*PRUNE).
2163 .
2164 .
2165 .SH "SEE ALSO"
2166 .rs
2167 .sp
2168 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3), \fBpcre\fP(3).
2169 .
2170 .
2171 .SH AUTHOR
2172 .rs
2173 .sp
2174 .nf
2175 Philip Hazel
2176 University Computing Service
2177 Cambridge CB2 3QH, England.
2178 .fi
2179 .
2180 .
2181 .SH REVISION
2182 .rs
2183 .sp
2184 .nf
2185 Last updated: 14 September 2007
2186 Copyright (c) 1997-2007 University of Cambridge.
2187 .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