ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 1370 - (show annotations)
Wed Oct 9 10:18:26 2013 UTC (7 years, 11 months ago) by ph10
File size: 131702 byte(s)
Error occurred while calculating annotation data.
Add \o{} and tidy up \x{} handling. Minor update to RunTest.
1 .TH PCREPATTERN 3 "08 October 2013" "PCRE 8.34"
3 PCRE - Perl-compatible regular expressions
5 .rs
6 .sp
7 The syntax and semantics of the regular expressions that are supported by PCRE
8 are described in detail below. There is a quick-reference syntax summary in the
9 .\" HREF
10 \fBpcresyntax\fP
11 .\"
12 page. PCRE tries to match Perl syntax and semantics as closely as it can. PCRE
13 also supports some alternative regular expression syntax (which does not
14 conflict with the Perl syntax) in order to provide some compatibility with
15 regular expressions in Python, .NET, and Oniguruma.
16 .P
17 Perl's regular expressions are described in its own documentation, and
18 regular expressions in general are covered in a number of books, some of which
19 have copious examples. Jeffrey Friedl's "Mastering Regular Expressions",
20 published by O'Reilly, covers regular expressions in great detail. This
21 description of PCRE's regular expressions is intended as reference material.
22 .P
23 This document discusses the patterns that are supported by PCRE when one its
24 main matching functions, \fBpcre_exec()\fP (8-bit) or \fBpcre[16|32]_exec()\fP
25 (16- or 32-bit), is used. PCRE also has alternative matching functions,
26 \fBpcre_dfa_exec()\fP and \fBpcre[16|32_dfa_exec()\fP, which match using a
27 different algorithm that is not Perl-compatible. Some of the features discussed
28 below are not available when DFA matching is used. The advantages and
29 disadvantages of the alternative functions, and how they differ from the normal
30 functions, are discussed in the
31 .\" HREF
32 \fBpcrematching\fP
33 .\"
34 page.
35 .
36 .
38 .rs
39 .sp
40 A number of options that can be passed to \fBpcre_compile()\fP can also be set
41 by special items at the start of a pattern. These are not Perl-compatible, but
42 are provided to make these options accessible to pattern writers who are not
43 able to change the program that processes the pattern. Any number of these
44 items may appear, but they must all be together right at the start of the
45 pattern string, and the letters must be in upper case.
46 .
47 .
48 .SS "UTF support"
49 .rs
50 .sp
51 The original operation of PCRE was on strings of one-byte characters. However,
52 there is now also support for UTF-8 strings in the original library, an
53 extra library that supports 16-bit and UTF-16 character strings, and a
54 third library that supports 32-bit and UTF-32 character strings. To use these
55 features, PCRE must be built to include appropriate support. When using UTF
56 strings you must either call the compiling function with the PCRE_UTF8,
57 PCRE_UTF16, or PCRE_UTF32 option, or the pattern must start with one of
58 these special sequences:
59 .sp
60 (*UTF8)
61 (*UTF16)
62 (*UTF32)
63 (*UTF)
64 .sp
65 (*UTF) is a generic sequence that can be used with any of the libraries.
66 Starting a pattern with such a sequence is equivalent to setting the relevant
67 option. How setting a UTF mode affects pattern matching is mentioned in several
68 places below. There is also a summary of features in the
69 .\" HREF
70 \fBpcreunicode\fP
71 .\"
72 page.
73 .P
74 Some applications that allow their users to supply patterns may wish to
75 restrict them to non-UTF data for security reasons. If the PCRE_NEVER_UTF
76 option is set at compile time, (*UTF) etc. are not allowed, and their
77 appearance causes an error.
78 .
79 .
80 .SS "Unicode property support"
81 .rs
82 .sp
83 Another special sequence that may appear at the start of a pattern is
84 .sp
85 (*UCP)
86 .sp
87 This has the same effect as setting the PCRE_UCP option: it causes sequences
88 such as \ed and \ew to use Unicode properties to determine character types,
89 instead of recognizing only characters with codes less than 128 via a lookup
90 table.
91 .
92 .
93 .SS "Disabling start-up optimizations"
94 .rs
95 .sp
96 If a pattern starts with (*NO_START_OPT), it has the same effect as setting the
97 PCRE_NO_START_OPTIMIZE option either at compile or matching time.
98 .
99 .
100 .\" HTML <a name="newlines"></a>
101 .SS "Newline conventions"
102 .rs
103 .sp
104 PCRE supports five different conventions for indicating line breaks in
105 strings: a single CR (carriage return) character, a single LF (linefeed)
106 character, the two-character sequence CRLF, any of the three preceding, or any
107 Unicode newline sequence. The
108 .\" HREF
109 \fBpcreapi\fP
110 .\"
111 page has
112 .\" HTML <a href="pcreapi.html#newlines">
113 .\" </a>
114 further discussion
115 .\"
116 about newlines, and shows how to set the newline convention in the
117 \fIoptions\fP arguments for the compiling and matching functions.
118 .P
119 It is also possible to specify a newline convention by starting a pattern
120 string with one of the following five sequences:
121 .sp
122 (*CR) carriage return
123 (*LF) linefeed
124 (*CRLF) carriage return, followed by linefeed
125 (*ANYCRLF) any of the three above
126 (*ANY) all Unicode newline sequences
127 .sp
128 These override the default and the options given to the compiling function. For
129 example, on a Unix system where LF is the default newline sequence, the pattern
130 .sp
131 (*CR)a.b
132 .sp
133 changes the convention to CR. That pattern matches "a\enb" because LF is no
134 longer a newline. If more than one of these settings is present, the last one
135 is used.
136 .P
137 The newline convention affects where the circumflex and dollar assertions are
138 true. It also affects the interpretation of the dot metacharacter when
139 PCRE_DOTALL is not set, and the behaviour of \eN. However, it does not affect
140 what the \eR escape sequence matches. By default, this is any Unicode newline
141 sequence, for Perl compatibility. However, this can be changed; see the
142 description of \eR in the section entitled
143 .\" HTML <a href="#newlineseq">
144 .\" </a>
145 "Newline sequences"
146 .\"
147 below. A change of \eR setting can be combined with a change of newline
148 convention.
149 .
150 .
151 .SS "Setting match and recursion limits"
152 .rs
153 .sp
154 The caller of \fBpcre_exec()\fP can set a limit on the number of times the
155 internal \fBmatch()\fP function is called and on the maximum depth of
156 recursive calls. These facilities are provided to catch runaway matches that
157 are provoked by patterns with huge matching trees (a typical example is a
158 pattern with nested unlimited repeats) and to avoid running out of system stack
159 by too much recursion. When one of these limits is reached, \fBpcre_exec()\fP
160 gives an error return. The limits can also be set by items at the start of the
161 pattern of the form
162 .sp
163 (*LIMIT_MATCH=d)
165 .sp
166 where d is any number of decimal digits. However, the value of the setting must
167 be less than the value set by the caller of \fBpcre_exec()\fP for it to have
168 any effect. In other words, the pattern writer can lower the limit set by the
169 programmer, but not raise it. If there is more than one setting of one of these
170 limits, the lower value is used.
171 .
172 .
174 .rs
175 .sp
176 PCRE can be compiled to run in an environment that uses EBCDIC as its character
177 code rather than ASCII or Unicode (typically a mainframe system). In the
178 sections below, character code values are ASCII or Unicode; in an EBCDIC
179 environment these characters may have different code values, and there are no
180 code points greater than 255.
181 .
182 .
184 .rs
185 .sp
186 A regular expression is a pattern that is matched against a subject string from
187 left to right. Most characters stand for themselves in a pattern, and match the
188 corresponding characters in the subject. As a trivial example, the pattern
189 .sp
190 The quick brown fox
191 .sp
192 matches a portion of a subject string that is identical to itself. When
193 caseless matching is specified (the PCRE_CASELESS option), letters are matched
194 independently of case. In a UTF mode, PCRE always understands the concept of
195 case for characters whose values are less than 128, so caseless matching is
196 always possible. For characters with higher values, the concept of case is
197 supported if PCRE is compiled with Unicode property support, but not otherwise.
198 If you want to use caseless matching for characters 128 and above, you must
199 ensure that PCRE is compiled with Unicode property support as well as with
200 UTF support.
201 .P
202 The power of regular expressions comes from the ability to include alternatives
203 and repetitions in the pattern. These are encoded in the pattern by the use of
204 \fImetacharacters\fP, which do not stand for themselves but instead are
205 interpreted in some special way.
206 .P
207 There are two different sets of metacharacters: those that are recognized
208 anywhere in the pattern except within square brackets, and those that are
209 recognized within square brackets. Outside square brackets, the metacharacters
210 are as follows:
211 .sp
212 \e general escape character with several uses
213 ^ assert start of string (or line, in multiline mode)
214 $ assert end of string (or line, in multiline mode)
215 . match any character except newline (by default)
216 [ start character class definition
217 | start of alternative branch
218 ( start subpattern
219 ) end subpattern
220 ? extends the meaning of (
221 also 0 or 1 quantifier
222 also quantifier minimizer
223 * 0 or more quantifier
224 + 1 or more quantifier
225 also "possessive quantifier"
226 { start min/max quantifier
227 .sp
228 Part of a pattern that is in square brackets is called a "character class". In
229 a character class the only metacharacters are:
230 .sp
231 \e general escape character
232 ^ negate the class, but only if the first character
233 - indicates character range
234 .\" JOIN
235 [ POSIX character class (only if followed by POSIX
236 syntax)
237 ] terminates the character class
238 .sp
239 The following sections describe the use of each of the metacharacters.
240 .
241 .
243 .rs
244 .sp
245 The backslash character has several uses. Firstly, if it is followed by a
246 character that is not a number or a letter, it takes away any special meaning
247 that character may have. This use of backslash as an escape character applies
248 both inside and outside character classes.
249 .P
250 For example, if you want to match a * character, you write \e* in the pattern.
251 This escaping action applies whether or not the following character would
252 otherwise be interpreted as a metacharacter, so it is always safe to precede a
253 non-alphanumeric with backslash to specify that it stands for itself. In
254 particular, if you want to match a backslash, you write \e\e.
255 .P
256 In a UTF mode, only ASCII numbers and letters have any special meaning after a
257 backslash. All other characters (in particular, those whose codepoints are
258 greater than 127) are treated as literals.
259 .P
260 If a pattern is compiled with the PCRE_EXTENDED option, white space in the
261 pattern (other than in a character class) and characters between a # outside
262 a character class and the next newline are ignored. An escaping backslash can
263 be used to include a white space or # character as part of the pattern.
264 .P
265 If you want to remove the special meaning from a sequence of characters, you
266 can do so by putting them between \eQ and \eE. This is different from Perl in
267 that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
268 Perl, $ and @ cause variable interpolation. Note the following examples:
269 .sp
270 Pattern PCRE matches Perl matches
271 .sp
272 .\" JOIN
273 \eQabc$xyz\eE abc$xyz abc followed by the
274 contents of $xyz
275 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
276 \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
277 .sp
278 The \eQ...\eE sequence is recognized both inside and outside character classes.
279 An isolated \eE that is not preceded by \eQ is ignored. If \eQ is not followed
280 by \eE later in the pattern, the literal interpretation continues to the end of
281 the pattern (that is, \eE is assumed at the end). If the isolated \eQ is inside
282 a character class, this causes an error, because the character class is not
283 terminated.
284 .
285 .
286 .\" HTML <a name="digitsafterbackslash"></a>
287 .SS "Non-printing characters"
288 .rs
289 .sp
290 A second use of backslash provides a way of encoding non-printing characters
291 in patterns in a visible manner. There is no restriction on the appearance of
292 non-printing characters, apart from the binary zero that terminates a pattern,
293 but when a pattern is being prepared by text editing, it is often easier to use
294 one of the following escape sequences than the binary character it represents:
295 .sp
296 \ea alarm, that is, the BEL character (hex 07)
297 \ecx "control-x", where x is any ASCII character
298 \ee escape (hex 1B)
299 \ef form feed (hex 0C)
300 \en linefeed (hex 0A)
301 \er carriage return (hex 0D)
302 \et tab (hex 09)
303 \e0dd character with octal code 0dd
304 \eddd character with octal code ddd, or back reference
305 \eo{ddd..} character with octal code ddd..
306 \exhh character with hex code hh
307 \ex{hhh..} character with hex code hhh.. (non-JavaScript mode)
308 \euhhhh character with hex code hhhh (JavaScript mode only)
309 .sp
310 The precise effect of \ecx on ASCII characters is as follows: if x is a lower
311 case letter, it is converted to upper case. Then bit 6 of the character (hex
312 40) is inverted. Thus \ecA to \ecZ become hex 01 to hex 1A (A is 41, Z is 5A),
313 but \ec{ becomes hex 3B ({ is 7B), and \ec; becomes hex 7B (; is 3B). If the
314 data item (byte or 16-bit value) following \ec has a value greater than 127, a
315 compile-time error occurs. This locks out non-ASCII characters in all modes.
316 .P
317 The \ec facility was designed for use with ASCII characters, but with the
318 extension to Unicode it is even less useful than it once was. It is, however,
319 recognized when PCRE is compiled in EBCDIC mode, where data items are always
320 bytes. In this mode, all values are valid after \ec. If the next character is a
321 lower case letter, it is converted to upper case. Then the 0xc0 bits of the
322 byte are inverted. Thus \ecA becomes hex 01, as in ASCII (A is C1), but because
323 the EBCDIC letters are disjoint, \ecZ becomes hex 29 (Z is E9), and other
324 characters also generate different values.
325 .P
326 After \e0 up to two further octal digits are read. If there are fewer than two
327 digits, just those that are present are used. Thus the sequence \e0\ex\e07
328 specifies two binary zeros followed by a BEL character (code value 7). Make
329 sure you supply two digits after the initial zero if the pattern character that
330 follows is itself an octal digit.
331 .P
332 The escape \eo must be followed by a sequence of octal digits, enclosed in
333 braces. An error occurs if this is not the case. This escape is a recent
334 addition to Perl; it provides way of specifying character code points as octal
335 numbers greater than 0777, and it also allows octal numbers and back references
336 to be unambiguously specified.
337 .P
338 For greater clarity and unambiguity, it is best to avoid following \e by a
339 digit greater than zero. Instead, use \eo{} or \ex{} to specify character
340 numbers, and \eg{} to specify back references. The following paragraphs
341 describe the old, ambiguous syntax.
342 .P
343 The handling of a backslash followed by a digit other than 0 is complicated,
344 and Perl has changed in recent releases, causing PCRE also to change. Outside a
345 character class, PCRE reads the digit and any following digits as a decimal
346 number. If the number is less than 8, or if there have been at least that many
347 previous capturing left parentheses in the expression, the entire sequence is
348 taken as a \fIback reference\fP. A description of how this works is given
349 .\" HTML <a href="#backreferences">
350 .\" </a>
351 later,
352 .\"
353 following the discussion of
354 .\" HTML <a href="#subpattern">
355 .\" </a>
356 parenthesized subpatterns.
357 .\"
358 .P
359 Inside a character class, or if the decimal number following \e is greater than
360 7 and there have not been that many capturing subpatterns, PCRE handles \e8 and
361 \e9 as the literal characters "8" and "9", and otherwise re-reads up to three
362 octal digits following the backslash, using them to generate a data character.
363 Any subsequent digits stand for themselves. For example:
364 .sp
365 \e040 is another way of writing an ASCII space
366 .\" JOIN
367 \e40 is the same, provided there are fewer than 40
368 previous capturing subpatterns
369 \e7 is always a back reference
370 .\" JOIN
371 \e11 might be a back reference, or another way of
372 writing a tab
373 \e011 is always a tab
374 \e0113 is a tab followed by the character "3"
375 .\" JOIN
376 \e113 might be a back reference, otherwise the
377 character with octal code 113
378 .\" JOIN
379 \e377 might be a back reference, otherwise
380 the value 255 (decimal)
381 .\" JOIN
382 \e81 is either a back reference, or the two
383 characters "8" and "1"
384 .sp
385 Note that octal values of 100 or greater that are specified using this syntax
386 must not be introduced by a leading zero, because no more than three octal
387 digits are ever read.
388 .P
389 By default, after \ex that is not followed by {, from zero to two hexadecimal
390 digits are read (letters can be in upper or lower case). Any number of
391 hexadecimal digits may appear between \ex{ and }. If a character other than
392 a hexadecimal digit appears between \ex{ and }, or if there is no terminating
393 }, an error occurs.
394 .P
395 If the PCRE_JAVASCRIPT_COMPAT option is set, the interpretation of \ex is
396 as just described only when it is followed by two hexadecimal digits.
397 Otherwise, it matches a literal "x" character. In JavaScript mode, support for
398 code points greater than 256 is provided by \eu, which must be followed by
399 four hexadecimal digits; otherwise it matches a literal "u" character.
400 .P
401 Characters whose value is less than 256 can be defined by either of the two
402 syntaxes for \ex (or by \eu in JavaScript mode). There is no difference in the
403 way they are handled. For example, \exdc is exactly the same as \ex{dc} (or
404 \eu00dc in JavaScript mode).
405 .
406 .
407 .SS "Constraints on character values"
408 .rs
409 .sp
410 Characters that are specified using octal or hexadecimal numbers are
411 limited to certain values, as follows:
412 .sp
413 8-bit non-UTF mode less than 0x100
414 8-bit UTF-8 mode less than 0x10ffff and a valid codepoint
415 16-bit non-UTF mode less than 0x10000
416 16-bit UTF-16 mode less than 0x10ffff and a valid codepoint
417 32-bit non-UTF mode less than 0x80000000
418 32-bit UTF-32 mode less than 0x10ffff and a valid codepoint
419 .sp
420 Invalid Unicode codepoints are the range 0xd800 to 0xdfff (the so-called
421 "surrogate" codepoints), and 0xffef.
422 .
423 .
424 .SS "Escape sequences in character classes"
425 .rs
426 .sp
427 All the sequences that define a single character value can be used both inside
428 and outside character classes. In addition, inside a character class, \eb is
429 interpreted as the backspace character (hex 08).
430 .P
431 \eN is not allowed in a character class. \eB, \eR, and \eX are not special
432 inside a character class. Like other unrecognized escape sequences, they are
433 treated as the literal characters "B", "R", and "X" by default, but cause an
434 error if the PCRE_EXTRA option is set. Outside a character class, these
435 sequences have different meanings.
436 .
437 .
438 .SS "Unsupported escape sequences"
439 .rs
440 .sp
441 In Perl, the sequences \el, \eL, \eu, and \eU are recognized by its string
442 handler and used to modify the case of following characters. By default, PCRE
443 does not support these escape sequences. However, if the PCRE_JAVASCRIPT_COMPAT
444 option is set, \eU matches a "U" character, and \eu can be used to define a
445 character by code point, as described in the previous section.
446 .
447 .
448 .SS "Absolute and relative back references"
449 .rs
450 .sp
451 The sequence \eg followed by an unsigned or a negative number, optionally
452 enclosed in braces, is an absolute or relative back reference. A named back
453 reference can be coded as \eg{name}. Back references are discussed
454 .\" HTML <a href="#backreferences">
455 .\" </a>
456 later,
457 .\"
458 following the discussion of
459 .\" HTML <a href="#subpattern">
460 .\" </a>
461 parenthesized subpatterns.
462 .\"
463 .
464 .
465 .SS "Absolute and relative subroutine calls"
466 .rs
467 .sp
468 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
469 a number enclosed either in angle brackets or single quotes, is an alternative
470 syntax for referencing a subpattern as a "subroutine". Details are discussed
471 .\" HTML <a href="#onigurumasubroutines">
472 .\" </a>
473 later.
474 .\"
475 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
476 synonymous. The former is a back reference; the latter is a
477 .\" HTML <a href="#subpatternsassubroutines">
478 .\" </a>
479 subroutine
480 .\"
481 call.
482 .
483 .
484 .\" HTML <a name="genericchartypes"></a>
485 .SS "Generic character types"
486 .rs
487 .sp
488 Another use of backslash is for specifying generic character types:
489 .sp
490 \ed any decimal digit
491 \eD any character that is not a decimal digit
492 \eh any horizontal white space character
493 \eH any character that is not a horizontal white space character
494 \es any white space character
495 \eS any character that is not a white space character
496 \ev any vertical white space character
497 \eV any character that is not a vertical white space character
498 \ew any "word" character
499 \eW any "non-word" character
500 .sp
501 There is also the single sequence \eN, which matches a non-newline character.
502 This is the same as
503 .\" HTML <a href="#fullstopdot">
504 .\" </a>
505 the "." metacharacter
506 .\"
507 when PCRE_DOTALL is not set. Perl also uses \eN to match characters by name;
508 PCRE does not support this.
509 .P
510 Each pair of lower and upper case escape sequences partitions the complete set
511 of characters into two disjoint sets. Any given character matches one, and only
512 one, of each pair. The sequences can appear both inside and outside character
513 classes. They each match one character of the appropriate type. If the current
514 matching point is at the end of the subject string, all of them fail, because
515 there is no character to match.
516 .P
517 For compatibility with Perl, \es did not used to match the VT character (code
518 11), which made it different from the the POSIX "space" class. However, Perl
519 added VT at release 5.18, and PCRE followed suit at release 8.34. The \es
520 characters are now HT (9), LF (10), VT (11), FF (12), CR (13), and space (32).
521 .P
522 A "word" character is an underscore or any character that is a letter or digit.
523 By default, the definition of letters and digits is controlled by PCRE's
524 low-valued character tables, and may vary if locale-specific matching is taking
525 place (see
526 .\" HTML <a href="pcreapi.html#localesupport">
527 .\" </a>
528 "Locale support"
529 .\"
530 in the
531 .\" HREF
532 \fBpcreapi\fP
533 .\"
534 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
535 or "french" in Windows, some character codes greater than 128 are used for
536 accented letters, and these are then matched by \ew. The use of locales with
537 Unicode is discouraged.
538 .P
539 By default, in a UTF mode, characters with values greater than 128 never match
540 \ed, \es, or \ew, and always match \eD, \eS, and \eW. These sequences retain
541 their original meanings from before UTF support was available, mainly for
542 efficiency reasons. However, if PCRE is compiled with Unicode property support,
543 and the PCRE_UCP option is set, the behaviour is changed so that Unicode
544 properties are used to determine character types, as follows:
545 .sp
546 \ed any character that \ep{Nd} matches (decimal digit)
547 \es any character that \ep{Z} matches, plus HT, LF, FF, CR
548 \ew any character that \ep{L} or \ep{N} matches, plus underscore
549 .sp
550 The upper case escapes match the inverse sets of characters. Note that \ed
551 matches only decimal digits, whereas \ew matches any Unicode digit, as well as
552 any Unicode letter, and underscore. Note also that PCRE_UCP affects \eb, and
553 \eB because they are defined in terms of \ew and \eW. Matching these sequences
554 is noticeably slower when PCRE_UCP is set.
555 .P
556 The sequences \eh, \eH, \ev, and \eV are features that were added to Perl at
557 release 5.10. In contrast to the other sequences, which match only ASCII
558 characters by default, these always match certain high-valued codepoints,
559 whether or not PCRE_UCP is set. The horizontal space characters are:
560 .sp
561 U+0009 Horizontal tab (HT)
562 U+0020 Space
563 U+00A0 Non-break space
564 U+1680 Ogham space mark
565 U+180E Mongolian vowel separator
566 U+2000 En quad
567 U+2001 Em quad
568 U+2002 En space
569 U+2003 Em space
570 U+2004 Three-per-em space
571 U+2005 Four-per-em space
572 U+2006 Six-per-em space
573 U+2007 Figure space
574 U+2008 Punctuation space
575 U+2009 Thin space
576 U+200A Hair space
577 U+202F Narrow no-break space
578 U+205F Medium mathematical space
579 U+3000 Ideographic space
580 .sp
581 The vertical space characters are:
582 .sp
583 U+000A Linefeed (LF)
584 U+000B Vertical tab (VT)
585 U+000C Form feed (FF)
586 U+000D Carriage return (CR)
587 U+0085 Next line (NEL)
588 U+2028 Line separator
589 U+2029 Paragraph separator
590 .sp
591 In 8-bit, non-UTF-8 mode, only the characters with codepoints less than 256 are
592 relevant.
593 .
594 .
595 .\" HTML <a name="newlineseq"></a>
596 .SS "Newline sequences"
597 .rs
598 .sp
599 Outside a character class, by default, the escape sequence \eR matches any
600 Unicode newline sequence. In 8-bit non-UTF-8 mode \eR is equivalent to the
601 following:
602 .sp
603 (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
604 .sp
605 This is an example of an "atomic group", details of which are given
606 .\" HTML <a href="#atomicgroup">
607 .\" </a>
608 below.
609 .\"
610 This particular group matches either the two-character sequence CR followed by
611 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
612 U+000B), FF (form feed, U+000C), CR (carriage return, U+000D), or NEL (next
613 line, U+0085). The two-character sequence is treated as a single unit that
614 cannot be split.
615 .P
616 In other modes, two additional characters whose codepoints are greater than 255
617 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
618 Unicode character property support is not needed for these characters to be
619 recognized.
620 .P
621 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
622 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
623 either at compile time or when the pattern is matched. (BSR is an abbrevation
624 for "backslash R".) This can be made the default when PCRE is built; if this is
625 the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
626 It is also possible to specify these settings by starting a pattern string with
627 one of the following sequences:
628 .sp
629 (*BSR_ANYCRLF) CR, LF, or CRLF only
630 (*BSR_UNICODE) any Unicode newline sequence
631 .sp
632 These override the default and the options given to the compiling function, but
633 they can themselves be overridden by options given to a matching function. Note
634 that these special settings, which are not Perl-compatible, are recognized only
635 at the very start of a pattern, and that they must be in upper case. If more
636 than one of them is present, the last one is used. They can be combined with a
637 change of newline convention; for example, a pattern can start with:
638 .sp
640 .sp
641 They can also be combined with the (*UTF8), (*UTF16), (*UTF32), (*UTF) or
642 (*UCP) special sequences. Inside a character class, \eR is treated as an
643 unrecognized escape sequence, and so matches the letter "R" by default, but
644 causes an error if PCRE_EXTRA is set.
645 .
646 .
647 .\" HTML <a name="uniextseq"></a>
648 .SS Unicode character properties
649 .rs
650 .sp
651 When PCRE is built with Unicode character property support, three additional
652 escape sequences that match characters with specific properties are available.
653 When in 8-bit non-UTF-8 mode, these sequences are of course limited to testing
654 characters whose codepoints are less than 256, but they do work in this mode.
655 The extra escape sequences are:
656 .sp
657 \ep{\fIxx\fP} a character with the \fIxx\fP property
658 \eP{\fIxx\fP} a character without the \fIxx\fP property
659 \eX a Unicode extended grapheme cluster
660 .sp
661 The property names represented by \fIxx\fP above are limited to the Unicode
662 script names, the general category properties, "Any", which matches any
663 character (including newline), and some special PCRE properties (described
664 in the
665 .\" HTML <a href="#extraprops">
666 .\" </a>
667 next section).
668 .\"
669 Other Perl properties such as "InMusicalSymbols" are not currently supported by
670 PCRE. Note that \eP{Any} does not match any characters, so always causes a
671 match failure.
672 .P
673 Sets of Unicode characters are defined as belonging to certain scripts. A
674 character from one of these sets can be matched using a script name. For
675 example:
676 .sp
677 \ep{Greek}
678 \eP{Han}
679 .sp
680 Those that are not part of an identified script are lumped together as
681 "Common". The current list of scripts is:
682 .P
683 Arabic,
684 Armenian,
685 Avestan,
686 Balinese,
687 Bamum,
688 Batak,
689 Bengali,
690 Bopomofo,
691 Brahmi,
692 Braille,
693 Buginese,
694 Buhid,
695 Canadian_Aboriginal,
696 Carian,
697 Chakma,
698 Cham,
699 Cherokee,
700 Common,
701 Coptic,
702 Cuneiform,
703 Cypriot,
704 Cyrillic,
705 Deseret,
706 Devanagari,
707 Egyptian_Hieroglyphs,
708 Ethiopic,
709 Georgian,
710 Glagolitic,
711 Gothic,
712 Greek,
713 Gujarati,
714 Gurmukhi,
715 Han,
716 Hangul,
717 Hanunoo,
718 Hebrew,
719 Hiragana,
720 Imperial_Aramaic,
721 Inherited,
722 Inscriptional_Pahlavi,
723 Inscriptional_Parthian,
724 Javanese,
725 Kaithi,
726 Kannada,
727 Katakana,
728 Kayah_Li,
729 Kharoshthi,
730 Khmer,
731 Lao,
732 Latin,
733 Lepcha,
734 Limbu,
735 Linear_B,
736 Lisu,
737 Lycian,
738 Lydian,
739 Malayalam,
740 Mandaic,
741 Meetei_Mayek,
742 Meroitic_Cursive,
743 Meroitic_Hieroglyphs,
744 Miao,
745 Mongolian,
746 Myanmar,
747 New_Tai_Lue,
748 Nko,
749 Ogham,
750 Old_Italic,
751 Old_Persian,
752 Old_South_Arabian,
753 Old_Turkic,
754 Ol_Chiki,
755 Oriya,
756 Osmanya,
757 Phags_Pa,
758 Phoenician,
759 Rejang,
760 Runic,
761 Samaritan,
762 Saurashtra,
763 Sharada,
764 Shavian,
765 Sinhala,
766 Sora_Sompeng,
767 Sundanese,
768 Syloti_Nagri,
769 Syriac,
770 Tagalog,
771 Tagbanwa,
772 Tai_Le,
773 Tai_Tham,
774 Tai_Viet,
775 Takri,
776 Tamil,
777 Telugu,
778 Thaana,
779 Thai,
780 Tibetan,
781 Tifinagh,
782 Ugaritic,
783 Vai,
784 Yi.
785 .P
786 Each character has exactly one Unicode general category property, specified by
787 a two-letter abbreviation. For compatibility with Perl, negation can be
788 specified by including a circumflex between the opening brace and the property
789 name. For example, \ep{^Lu} is the same as \eP{Lu}.
790 .P
791 If only one letter is specified with \ep or \eP, it includes all the general
792 category properties that start with that letter. In this case, in the absence
793 of negation, the curly brackets in the escape sequence are optional; these two
794 examples have the same effect:
795 .sp
796 \ep{L}
797 \epL
798 .sp
799 The following general category property codes are supported:
800 .sp
801 C Other
802 Cc Control
803 Cf Format
804 Cn Unassigned
805 Co Private use
806 Cs Surrogate
807 .sp
808 L Letter
809 Ll Lower case letter
810 Lm Modifier letter
811 Lo Other letter
812 Lt Title case letter
813 Lu Upper case letter
814 .sp
815 M Mark
816 Mc Spacing mark
817 Me Enclosing mark
818 Mn Non-spacing mark
819 .sp
820 N Number
821 Nd Decimal number
822 Nl Letter number
823 No Other number
824 .sp
825 P Punctuation
826 Pc Connector punctuation
827 Pd Dash punctuation
828 Pe Close punctuation
829 Pf Final punctuation
830 Pi Initial punctuation
831 Po Other punctuation
832 Ps Open punctuation
833 .sp
834 S Symbol
835 Sc Currency symbol
836 Sk Modifier symbol
837 Sm Mathematical symbol
838 So Other symbol
839 .sp
840 Z Separator
841 Zl Line separator
842 Zp Paragraph separator
843 Zs Space separator
844 .sp
845 The special property L& is also supported: it matches a character that has
846 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
847 a modifier or "other".
848 .P
849 The Cs (Surrogate) property applies only to characters in the range U+D800 to
850 U+DFFF. Such characters are not valid in Unicode strings and so
851 cannot be tested by PCRE, unless UTF validity checking has been turned off
852 (see the discussion of PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK and
853 PCRE_NO_UTF32_CHECK in the
854 .\" HREF
855 \fBpcreapi\fP
856 .\"
857 page). Perl does not support the Cs property.
858 .P
859 The long synonyms for property names that Perl supports (such as \ep{Letter})
860 are not supported by PCRE, nor is it permitted to prefix any of these
861 properties with "Is".
862 .P
863 No character that is in the Unicode table has the Cn (unassigned) property.
864 Instead, this property is assumed for any code point that is not in the
865 Unicode table.
866 .P
867 Specifying caseless matching does not affect these escape sequences. For
868 example, \ep{Lu} always matches only upper case letters. This is different from
869 the behaviour of current versions of Perl.
870 .P
871 Matching characters by Unicode property is not fast, because PCRE has to do a
872 multistage table lookup in order to find a character's property. That is why
873 the traditional escape sequences such as \ed and \ew do not use Unicode
874 properties in PCRE by default, though you can make them do so by setting the
875 PCRE_UCP option or by starting the pattern with (*UCP).
876 .
877 .
878 .SS Extended grapheme clusters
879 .rs
880 .sp
881 The \eX escape matches any number of Unicode characters that form an "extended
882 grapheme cluster", and treats the sequence as an atomic group
883 .\" HTML <a href="#atomicgroup">
884 .\" </a>
885 (see below).
886 .\"
887 Up to and including release 8.31, PCRE matched an earlier, simpler definition
888 that was equivalent to
889 .sp
890 (?>\ePM\epM*)
891 .sp
892 That is, it matched a character without the "mark" property, followed by zero
893 or more characters with the "mark" property. Characters with the "mark"
894 property are typically non-spacing accents that affect the preceding character.
895 .P
896 This simple definition was extended in Unicode to include more complicated
897 kinds of composite character by giving each character a grapheme breaking
898 property, and creating rules that use these properties to define the boundaries
899 of extended grapheme clusters. In releases of PCRE later than 8.31, \eX matches
900 one of these clusters.
901 .P
902 \eX always matches at least one character. Then it decides whether to add
903 additional characters according to the following rules for ending a cluster:
904 .P
905 1. End at the end of the subject string.
906 .P
907 2. Do not end between CR and LF; otherwise end after any control character.
908 .P
909 3. Do not break Hangul (a Korean script) syllable sequences. Hangul characters
910 are of five types: L, V, T, LV, and LVT. An L character may be followed by an
911 L, V, LV, or LVT character; an LV or V character may be followed by a V or T
912 character; an LVT or T character may be follwed only by a T character.
913 .P
914 4. Do not end before extending characters or spacing marks. Characters with
915 the "mark" property always have the "extend" grapheme breaking property.
916 .P
917 5. Do not end after prepend characters.
918 .P
919 6. Otherwise, end the cluster.
920 .
921 .
922 .\" HTML <a name="extraprops"></a>
923 .SS PCRE's additional properties
924 .rs
925 .sp
926 As well as the standard Unicode properties described above, PCRE supports four
927 more that make it possible to convert traditional escape sequences such as \ew
928 and \es and POSIX character classes to use Unicode properties. PCRE uses these
929 non-standard, non-Perl properties internally when PCRE_UCP is set. However,
930 they may also be used explicitly. These properties are:
931 .sp
932 Xan Any alphanumeric character
933 Xps Any POSIX space character
934 Xsp Any Perl space character
935 Xwd Any Perl "word" character
936 .sp
937 Xan matches characters that have either the L (letter) or the N (number)
938 property. Xps matches the characters tab, linefeed, vertical tab, form feed, or
939 carriage return, and any other character that has the Z (separator) property.
940 Xsp is the same as Xps, except that vertical tab is excluded. Xwd matches the
941 same characters as Xan, plus underscore.
942 .P
943 There is another non-standard property, Xuc, which matches any character that
944 can be represented by a Universal Character Name in C++ and other programming
945 languages. These are the characters $, @, ` (grave accent), and all characters
946 with Unicode code points greater than or equal to U+00A0, except for the
947 surrogates U+D800 to U+DFFF. Note that most base (ASCII) characters are
948 excluded. (Universal Character Names are of the form \euHHHH or \eUHHHHHHHH
949 where H is a hexadecimal digit. Note that the Xuc property does not match these
950 sequences but the characters that they represent.)
951 .
952 .
953 .\" HTML <a name="resetmatchstart"></a>
954 .SS "Resetting the match start"
955 .rs
956 .sp
957 The escape sequence \eK causes any previously matched characters not to be
958 included in the final matched sequence. For example, the pattern:
959 .sp
960 foo\eKbar
961 .sp
962 matches "foobar", but reports that it has matched "bar". This feature is
963 similar to a lookbehind assertion
964 .\" HTML <a href="#lookbehind">
965 .\" </a>
966 (described below).
967 .\"
968 However, in this case, the part of the subject before the real match does not
969 have to be of fixed length, as lookbehind assertions do. The use of \eK does
970 not interfere with the setting of
971 .\" HTML <a href="#subpattern">
972 .\" </a>
973 captured substrings.
974 .\"
975 For example, when the pattern
976 .sp
977 (foo)\eKbar
978 .sp
979 matches "foobar", the first substring is still set to "foo".
980 .P
981 Perl documents that the use of \eK within assertions is "not well defined". In
982 PCRE, \eK is acted upon when it occurs inside positive assertions, but is
983 ignored in negative assertions.
984 .
985 .
986 .\" HTML <a name="smallassertions"></a>
987 .SS "Simple assertions"
988 .rs
989 .sp
990 The final use of backslash is for certain simple assertions. An assertion
991 specifies a condition that has to be met at a particular point in a match,
992 without consuming any characters from the subject string. The use of
993 subpatterns for more complicated assertions is described
994 .\" HTML <a href="#bigassertions">
995 .\" </a>
996 below.
997 .\"
998 The backslashed assertions are:
999 .sp
1000 \eb matches at a word boundary
1001 \eB matches when not at a word boundary
1002 \eA matches at the start of the subject
1003 \eZ matches at the end of the subject
1004 also matches before a newline at the end of the subject
1005 \ez matches only at the end of the subject
1006 \eG matches at the first matching position in the subject
1007 .sp
1008 Inside a character class, \eb has a different meaning; it matches the backspace
1009 character. If any other of these assertions appears in a character class, by
1010 default it matches the corresponding literal character (for example, \eB
1011 matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
1012 escape sequence" error is generated instead.
1013 .P
1014 A word boundary is a position in the subject string where the current character
1015 and the previous character do not both match \ew or \eW (i.e. one matches
1016 \ew and the other matches \eW), or the start or end of the string if the
1017 first or last character matches \ew, respectively. In a UTF mode, the meanings
1018 of \ew and \eW can be changed by setting the PCRE_UCP option. When this is
1019 done, it also affects \eb and \eB. Neither PCRE nor Perl has a separate "start
1020 of word" or "end of word" metasequence. However, whatever follows \eb normally
1021 determines which it is. For example, the fragment \eba matches "a" at the start
1022 of a word.
1023 .P
1024 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
1025 dollar (described in the next section) in that they only ever match at the very
1026 start and end of the subject string, whatever options are set. Thus, they are
1027 independent of multiline mode. These three assertions are not affected by the
1028 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
1029 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
1030 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
1031 at a point other than the beginning of the subject, \eA can never match. The
1032 difference between \eZ and \ez is that \eZ matches before a newline at the end
1033 of the string as well as at the very end, whereas \ez matches only at the end.
1034 .P
1035 The \eG assertion is true only when the current matching position is at the
1036 start point of the match, as specified by the \fIstartoffset\fP argument of
1037 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
1038 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
1039 arguments, you can mimic Perl's /g option, and it is in this kind of
1040 implementation where \eG can be useful.
1041 .P
1042 Note, however, that PCRE's interpretation of \eG, as the start of the current
1043 match, is subtly different from Perl's, which defines it as the end of the
1044 previous match. In Perl, these can be different when the previously matched
1045 string was empty. Because PCRE does just one match at a time, it cannot
1046 reproduce this behaviour.
1047 .P
1048 If all the alternatives of a pattern begin with \eG, the expression is anchored
1049 to the starting match position, and the "anchored" flag is set in the compiled
1050 regular expression.
1051 .
1052 .
1054 .rs
1055 .sp
1056 The circumflex and dollar metacharacters are zero-width assertions. That is,
1057 they test for a particular condition being true without consuming any
1058 characters from the subject string.
1059 .P
1060 Outside a character class, in the default matching mode, the circumflex
1061 character is an assertion that is true only if the current matching point is at
1062 the start of the subject string. If the \fIstartoffset\fP argument of
1063 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
1064 option is unset. Inside a character class, circumflex has an entirely different
1065 meaning
1066 .\" HTML <a href="#characterclass">
1067 .\" </a>
1068 (see below).
1069 .\"
1070 .P
1071 Circumflex need not be the first character of the pattern if a number of
1072 alternatives are involved, but it should be the first thing in each alternative
1073 in which it appears if the pattern is ever to match that branch. If all
1074 possible alternatives start with a circumflex, that is, if the pattern is
1075 constrained to match only at the start of the subject, it is said to be an
1076 "anchored" pattern. (There are also other constructs that can cause a pattern
1077 to be anchored.)
1078 .P
1079 The dollar character is an assertion that is true only if the current matching
1080 point is at the end of the subject string, or immediately before a newline at
1081 the end of the string (by default). Note, however, that it does not actually
1082 match the newline. Dollar need not be the last character of the pattern if a
1083 number of alternatives are involved, but it should be the last item in any
1084 branch in which it appears. Dollar has no special meaning in a character class.
1085 .P
1086 The meaning of dollar can be changed so that it matches only at the very end of
1087 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
1088 does not affect the \eZ assertion.
1089 .P
1090 The meanings of the circumflex and dollar characters are changed if the
1091 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
1092 immediately after internal newlines as well as at the start of the subject
1093 string. It does not match after a newline that ends the string. A dollar
1094 matches before any newlines in the string, as well as at the very end, when
1095 PCRE_MULTILINE is set. When newline is specified as the two-character
1096 sequence CRLF, isolated CR and LF characters do not indicate newlines.
1097 .P
1098 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
1099 \en represents a newline) in multiline mode, but not otherwise. Consequently,
1100 patterns that are anchored in single line mode because all branches start with
1101 ^ are not anchored in multiline mode, and a match for circumflex is possible
1102 when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
1103 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
1104 .P
1105 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
1106 end of the subject in both modes, and if all branches of a pattern start with
1107 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
1108 .
1109 .
1110 .\" HTML <a name="fullstopdot"></a>
1112 .rs
1113 .sp
1114 Outside a character class, a dot in the pattern matches any one character in
1115 the subject string except (by default) a character that signifies the end of a
1116 line.
1117 .P
1118 When a line ending is defined as a single character, dot never matches that
1119 character; when the two-character sequence CRLF is used, dot does not match CR
1120 if it is immediately followed by LF, but otherwise it matches all characters
1121 (including isolated CRs and LFs). When any Unicode line endings are being
1122 recognized, dot does not match CR or LF or any of the other line ending
1123 characters.
1124 .P
1125 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
1126 option is set, a dot matches any one character, without exception. If the
1127 two-character sequence CRLF is present in the subject string, it takes two dots
1128 to match it.
1129 .P
1130 The handling of dot is entirely independent of the handling of circumflex and
1131 dollar, the only relationship being that they both involve newlines. Dot has no
1132 special meaning in a character class.
1133 .P
1134 The escape sequence \eN behaves like a dot, except that it is not affected by
1135 the PCRE_DOTALL option. In other words, it matches any character except one
1136 that signifies the end of a line. Perl also uses \eN to match characters by
1137 name; PCRE does not support this.
1138 .
1139 .
1141 .rs
1142 .sp
1143 Outside a character class, the escape sequence \eC matches any one data unit,
1144 whether or not a UTF mode is set. In the 8-bit library, one data unit is one
1145 byte; in the 16-bit library it is a 16-bit unit; in the 32-bit library it is
1146 a 32-bit unit. Unlike a dot, \eC always
1147 matches line-ending characters. The feature is provided in Perl in order to
1148 match individual bytes in UTF-8 mode, but it is unclear how it can usefully be
1149 used. Because \eC breaks up characters into individual data units, matching one
1150 unit with \eC in a UTF mode means that the rest of the string may start with a
1151 malformed UTF character. This has undefined results, because PCRE assumes that
1152 it is dealing with valid UTF strings (and by default it checks this at the
1153 start of processing unless the PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK or
1154 PCRE_NO_UTF32_CHECK option is used).
1155 .P
1156 PCRE does not allow \eC to appear in lookbehind assertions
1157 .\" HTML <a href="#lookbehind">
1158 .\" </a>
1159 (described below)
1160 .\"
1161 in a UTF mode, because this would make it impossible to calculate the length of
1162 the lookbehind.
1163 .P
1164 In general, the \eC escape sequence is best avoided. However, one
1165 way of using it that avoids the problem of malformed UTF characters is to use a
1166 lookahead to check the length of the next character, as in this pattern, which
1167 could be used with a UTF-8 string (ignore white space and line breaks):
1168 .sp
1169 (?| (?=[\ex00-\ex7f])(\eC) |
1170 (?=[\ex80-\ex{7ff}])(\eC)(\eC) |
1171 (?=[\ex{800}-\ex{ffff}])(\eC)(\eC)(\eC) |
1172 (?=[\ex{10000}-\ex{1fffff}])(\eC)(\eC)(\eC)(\eC))
1173 .sp
1174 A group that starts with (?| resets the capturing parentheses numbers in each
1175 alternative (see
1176 .\" HTML <a href="#dupsubpatternnumber">
1177 .\" </a>
1178 "Duplicate Subpattern Numbers"
1179 .\"
1180 below). The assertions at the start of each branch check the next UTF-8
1181 character for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The
1182 character's individual bytes are then captured by the appropriate number of
1183 groups.
1184 .
1185 .
1186 .\" HTML <a name="characterclass"></a>
1188 .rs
1189 .sp
1190 An opening square bracket introduces a character class, terminated by a closing
1191 square bracket. A closing square bracket on its own is not special by default.
1192 However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
1193 bracket causes a compile-time error. If a closing square bracket is required as
1194 a member of the class, it should be the first data character in the class
1195 (after an initial circumflex, if present) or escaped with a backslash.
1196 .P
1197 A character class matches a single character in the subject. In a UTF mode, the
1198 character may be more than one data unit long. A matched character must be in
1199 the set of characters defined by the class, unless the first character in the
1200 class definition is a circumflex, in which case the subject character must not
1201 be in the set defined by the class. If a circumflex is actually required as a
1202 member of the class, ensure it is not the first character, or escape it with a
1203 backslash.
1204 .P
1205 For example, the character class [aeiou] matches any lower case vowel, while
1206 [^aeiou] matches any character that is not a lower case vowel. Note that a
1207 circumflex is just a convenient notation for specifying the characters that
1208 are in the class by enumerating those that are not. A class that starts with a
1209 circumflex is not an assertion; it still consumes a character from the subject
1210 string, and therefore it fails if the current pointer is at the end of the
1211 string.
1212 .P
1213 In UTF-8 (UTF-16, UTF-32) mode, characters with values greater than 255 (0xffff)
1214 can be included in a class as a literal string of data units, or by using the
1215 \ex{ escaping mechanism.
1216 .P
1217 When caseless matching is set, any letters in a class represent both their
1218 upper case and lower case versions, so for example, a caseless [aeiou] matches
1219 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1220 caseful version would. In a UTF mode, PCRE always understands the concept of
1221 case for characters whose values are less than 128, so caseless matching is
1222 always possible. For characters with higher values, the concept of case is
1223 supported if PCRE is compiled with Unicode property support, but not otherwise.
1224 If you want to use caseless matching in a UTF mode for characters 128 and
1225 above, you must ensure that PCRE is compiled with Unicode property support as
1226 well as with UTF support.
1227 .P
1228 Characters that might indicate line breaks are never treated in any special way
1229 when matching character classes, whatever line-ending sequence is in use, and
1230 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
1231 such as [^a] always matches one of these characters.
1232 .P
1233 The minus (hyphen) character can be used to specify a range of characters in a
1234 character class. For example, [d-m] matches any letter between d and m,
1235 inclusive. If a minus character is required in a class, it must be escaped with
1236 a backslash or appear in a position where it cannot be interpreted as
1237 indicating a range, typically as the first or last character in the class.
1238 .P
1239 It is not possible to have the literal character "]" as the end character of a
1240 range. A pattern such as [W-]46] is interpreted as a class of two characters
1241 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1242 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1243 the end of range, so [W-\e]46] is interpreted as a class containing a range
1244 followed by two other characters. The octal or hexadecimal representation of
1245 "]" can also be used to end a range.
1246 .P
1247 Ranges operate in the collating sequence of character values. They can also be
1248 used for characters specified numerically, for example [\e000-\e037]. Ranges
1249 can include any characters that are valid for the current mode.
1250 .P
1251 If a range that includes letters is used when caseless matching is set, it
1252 matches the letters in either case. For example, [W-c] is equivalent to
1253 [][\e\e^_`wxyzabc], matched caselessly, and in a non-UTF mode, if character
1254 tables for a French locale are in use, [\exc8-\excb] matches accented E
1255 characters in both cases. In UTF modes, PCRE supports the concept of case for
1256 characters with values greater than 128 only when it is compiled with Unicode
1257 property support.
1258 .P
1259 The character escape sequences \ed, \eD, \eh, \eH, \ep, \eP, \es, \eS, \ev,
1260 \eV, \ew, and \eW may appear in a character class, and add the characters that
1261 they match to the class. For example, [\edABCDEF] matches any hexadecimal
1262 digit. In UTF modes, the PCRE_UCP option affects the meanings of \ed, \es, \ew
1263 and their upper case partners, just as it does when they appear outside a
1264 character class, as described in the section entitled
1265 .\" HTML <a href="#genericchartypes">
1266 .\" </a>
1267 "Generic character types"
1268 .\"
1269 above. The escape sequence \eb has a different meaning inside a character
1270 class; it matches the backspace character. The sequences \eB, \eN, \eR, and \eX
1271 are not special inside a character class. Like any other unrecognized escape
1272 sequences, they are treated as the literal characters "B", "N", "R", and "X" by
1273 default, but cause an error if the PCRE_EXTRA option is set.
1274 .P
1275 A circumflex can conveniently be used with the upper case character types to
1276 specify a more restricted set of characters than the matching lower case type.
1277 For example, the class [^\eW_] matches any letter or digit, but not underscore,
1278 whereas [\ew] includes underscore. A positive character class should be read as
1279 "something OR something OR ..." and a negative class as "NOT something AND NOT
1280 something AND NOT ...".
1281 .P
1282 The only metacharacters that are recognized in character classes are backslash,
1283 hyphen (only where it can be interpreted as specifying a range), circumflex
1284 (only at the start), opening square bracket (only when it can be interpreted as
1285 introducing a POSIX class name - see the next section), and the terminating
1286 closing square bracket. However, escaping other non-alphanumeric characters
1287 does no harm.
1288 .
1289 .
1291 .rs
1292 .sp
1293 Perl supports the POSIX notation for character classes. This uses names
1294 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1295 this notation. For example,
1296 .sp
1297 [01[:alpha:]%]
1298 .sp
1299 matches "0", "1", any alphabetic character, or "%". The supported class names
1300 are:
1301 .sp
1302 alnum letters and digits
1303 alpha letters
1304 ascii character codes 0 - 127
1305 blank space or tab only
1306 cntrl control characters
1307 digit decimal digits (same as \ed)
1308 graph printing characters, excluding space
1309 lower lower case letters
1310 print printing characters, including space
1311 punct printing characters, excluding letters and digits and space
1312 space white space (not quite the same as \es)
1313 upper upper case letters
1314 word "word" characters (same as \ew)
1315 xdigit hexadecimal digits
1316 .sp
1317 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1318 space (32). "Space" used to be different to \es, which did not include VT, for
1319 Perl compatibility. However, Perl changed at release 5.18, and PCRE followed at
1320 release 8.34. "Space" and \es now match the same set of characters.
1321 .P
1322 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1323 5.8. Another Perl extension is negation, which is indicated by a ^ character
1324 after the colon. For example,
1325 .sp
1326 [12[:^digit:]]
1327 .sp
1328 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1329 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1330 supported, and an error is given if they are encountered.
1331 .P
1332 By default, in UTF modes, characters with values greater than 128 do not match
1333 any of the POSIX character classes. However, if the PCRE_UCP option is passed
1334 to \fBpcre_compile()\fP, some of the classes are changed so that Unicode
1335 character properties are used. This is achieved by replacing the POSIX classes
1336 by other sequences, as follows:
1337 .sp
1338 [:alnum:] becomes \ep{Xan}
1339 [:alpha:] becomes \ep{L}
1340 [:blank:] becomes \eh
1341 [:digit:] becomes \ep{Nd}
1342 [:lower:] becomes \ep{Ll}
1343 [:space:] becomes \ep{Xps}
1344 [:upper:] becomes \ep{Lu}
1345 [:word:] becomes \ep{Xwd}
1346 .sp
1347 Negated versions, such as [:^alpha:] use \eP instead of \ep. The other POSIX
1348 classes are unchanged, and match only characters with code points less than
1349 128.
1350 .
1351 .
1353 .rs
1354 .sp
1355 Vertical bar characters are used to separate alternative patterns. For example,
1356 the pattern
1357 .sp
1358 gilbert|sullivan
1359 .sp
1360 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1361 and an empty alternative is permitted (matching the empty string). The matching
1362 process tries each alternative in turn, from left to right, and the first one
1363 that succeeds is used. If the alternatives are within a subpattern
1364 .\" HTML <a href="#subpattern">
1365 .\" </a>
1366 (defined below),
1367 .\"
1368 "succeeds" means matching the rest of the main pattern as well as the
1369 alternative in the subpattern.
1370 .
1371 .
1373 .rs
1374 .sp
1375 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1376 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1377 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1378 The option letters are
1379 .sp
1380 i for PCRE_CASELESS
1382 s for PCRE_DOTALL
1383 x for PCRE_EXTENDED
1384 .sp
1385 For example, (?im) sets caseless, multiline matching. It is also possible to
1386 unset these options by preceding the letter with a hyphen, and a combined
1387 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1388 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1389 permitted. If a letter appears both before and after the hyphen, the option is
1390 unset.
1391 .P
1392 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1393 changed in the same way as the Perl-compatible options by using the characters
1394 J, U and X respectively.
1395 .P
1396 When one of these option changes occurs at top level (that is, not inside
1397 subpattern parentheses), the change applies to the remainder of the pattern
1398 that follows. If the change is placed right at the start of a pattern, PCRE
1399 extracts it into the global options (and it will therefore show up in data
1400 extracted by the \fBpcre_fullinfo()\fP function).
1401 .P
1402 An option change within a subpattern (see below for a description of
1403 subpatterns) affects only that part of the subpattern that follows it, so
1404 .sp
1405 (a(?i)b)c
1406 .sp
1407 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1408 By this means, options can be made to have different settings in different
1409 parts of the pattern. Any changes made in one alternative do carry on
1410 into subsequent branches within the same subpattern. For example,
1411 .sp
1412 (a(?i)b|c)
1413 .sp
1414 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1415 branch is abandoned before the option setting. This is because the effects of
1416 option settings happen at compile time. There would be some very weird
1417 behaviour otherwise.
1418 .P
1419 \fBNote:\fP There are other PCRE-specific options that can be set by the
1420 application when the compiling or matching functions are called. In some cases
1421 the pattern can contain special leading sequences such as (*CRLF) to override
1422 what the application has set or what has been defaulted. Details are given in
1423 the section entitled
1424 .\" HTML <a href="#newlineseq">
1425 .\" </a>
1426 "Newline sequences"
1427 .\"
1428 above. There are also the (*UTF8), (*UTF16),(*UTF32), and (*UCP) leading
1429 sequences that can be used to set UTF and Unicode property modes; they are
1430 equivalent to setting the PCRE_UTF8, PCRE_UTF16, PCRE_UTF32 and the PCRE_UCP
1431 options, respectively. The (*UTF) sequence is a generic version that can be
1432 used with any of the libraries. However, the application can set the
1433 PCRE_NEVER_UTF option, which locks out the use of the (*UTF) sequences.
1434 .
1435 .
1436 .\" HTML <a name="subpattern"></a>
1438 .rs
1439 .sp
1440 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1441 Turning part of a pattern into a subpattern does two things:
1442 .sp
1443 1. It localizes a set of alternatives. For example, the pattern
1444 .sp
1445 cat(aract|erpillar|)
1446 .sp
1447 matches "cataract", "caterpillar", or "cat". Without the parentheses, it would
1448 match "cataract", "erpillar" or an empty string.
1449 .sp
1450 2. It sets up the subpattern as a capturing subpattern. This means that, when
1451 the whole pattern matches, that portion of the subject string that matched the
1452 subpattern is passed back to the caller via the \fIovector\fP argument of the
1453 matching function. (This applies only to the traditional matching functions;
1454 the DFA matching functions do not support capturing.)
1455 .P
1456 Opening parentheses are counted from left to right (starting from 1) to obtain
1457 numbers for the capturing subpatterns. For example, if the string "the red
1458 king" is matched against the pattern
1459 .sp
1460 the ((red|white) (king|queen))
1461 .sp
1462 the captured substrings are "red king", "red", and "king", and are numbered 1,
1463 2, and 3, respectively.
1464 .P
1465 The fact that plain parentheses fulfil two functions is not always helpful.
1466 There are often times when a grouping subpattern is required without a
1467 capturing requirement. If an opening parenthesis is followed by a question mark
1468 and a colon, the subpattern does not do any capturing, and is not counted when
1469 computing the number of any subsequent capturing subpatterns. For example, if
1470 the string "the white queen" is matched against the pattern
1471 .sp
1472 the ((?:red|white) (king|queen))
1473 .sp
1474 the captured substrings are "white queen" and "queen", and are numbered 1 and
1475 2. The maximum number of capturing subpatterns is 65535.
1476 .P
1477 As a convenient shorthand, if any option settings are required at the start of
1478 a non-capturing subpattern, the option letters may appear between the "?" and
1479 the ":". Thus the two patterns
1480 .sp
1481 (?i:saturday|sunday)
1482 (?:(?i)saturday|sunday)
1483 .sp
1484 match exactly the same set of strings. Because alternative branches are tried
1485 from left to right, and options are not reset until the end of the subpattern
1486 is reached, an option setting in one branch does affect subsequent branches, so
1487 the above patterns match "SUNDAY" as well as "Saturday".
1488 .
1489 .
1490 .\" HTML <a name="dupsubpatternnumber"></a>
1492 .rs
1493 .sp
1494 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1495 the same numbers for its capturing parentheses. Such a subpattern starts with
1496 (?| and is itself a non-capturing subpattern. For example, consider this
1497 pattern:
1498 .sp
1499 (?|(Sat)ur|(Sun))day
1500 .sp
1501 Because the two alternatives are inside a (?| group, both sets of capturing
1502 parentheses are numbered one. Thus, when the pattern matches, you can look
1503 at captured substring number one, whichever alternative matched. This construct
1504 is useful when you want to capture part, but not all, of one of a number of
1505 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1506 number is reset at the start of each branch. The numbers of any capturing
1507 parentheses that follow the subpattern start after the highest number used in
1508 any branch. The following example is taken from the Perl documentation. The
1509 numbers underneath show in which buffer the captured content will be stored.
1510 .sp
1511 # before ---------------branch-reset----------- after
1512 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1513 # 1 2 2 3 2 3 4
1514 .sp
1515 A back reference to a numbered subpattern uses the most recent value that is
1516 set for that number by any subpattern. The following pattern matches "abcabc"
1517 or "defdef":
1518 .sp
1519 /(?|(abc)|(def))\e1/
1520 .sp
1521 In contrast, a subroutine call to a numbered subpattern always refers to the
1522 first one in the pattern with the given number. The following pattern matches
1523 "abcabc" or "defabc":
1524 .sp
1525 /(?|(abc)|(def))(?1)/
1526 .sp
1527 If a
1528 .\" HTML <a href="#conditions">
1529 .\" </a>
1530 condition test
1531 .\"
1532 for a subpattern's having matched refers to a non-unique number, the test is
1533 true if any of the subpatterns of that number have matched.
1534 .P
1535 An alternative approach to using this "branch reset" feature is to use
1536 duplicate named subpatterns, as described in the next section.
1537 .
1538 .
1540 .rs
1541 .sp
1542 Identifying capturing parentheses by number is simple, but it can be very hard
1543 to keep track of the numbers in complicated regular expressions. Furthermore,
1544 if an expression is modified, the numbers may change. To help with this
1545 difficulty, PCRE supports the naming of subpatterns. This feature was not
1546 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1547 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1548 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1549 have different names, but PCRE does not.
1550 .P
1551 In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1552 (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1553 parentheses from other parts of the pattern, such as
1554 .\" HTML <a href="#backreferences">
1555 .\" </a>
1556 back references,
1557 .\"
1558 .\" HTML <a href="#recursion">
1559 .\" </a>
1560 recursion,
1561 .\"
1562 and
1563 .\" HTML <a href="#conditions">
1564 .\" </a>
1565 conditions,
1566 .\"
1567 can be made by name as well as by number.
1568 .P
1569 Names consist of up to 32 alphanumeric characters and underscores. Named
1570 capturing parentheses are still allocated numbers as well as names, exactly as
1571 if the names were not present. The PCRE API provides function calls for
1572 extracting the name-to-number translation table from a compiled pattern. There
1573 is also a convenience function for extracting a captured substring by name.
1574 .P
1575 By default, a name must be unique within a pattern, but it is possible to relax
1576 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1577 names are also always permitted for subpatterns with the same number, set up as
1578 described in the previous section.) Duplicate names can be useful for patterns
1579 where only one instance of the named parentheses can match. Suppose you want to
1580 match the name of a weekday, either as a 3-letter abbreviation or as the full
1581 name, and in both cases you want to extract the abbreviation. This pattern
1582 (ignoring the line breaks) does the job:
1583 .sp
1584 (?<DN>Mon|Fri|Sun)(?:day)?|
1585 (?<DN>Tue)(?:sday)?|
1586 (?<DN>Wed)(?:nesday)?|
1587 (?<DN>Thu)(?:rsday)?|
1588 (?<DN>Sat)(?:urday)?
1589 .sp
1590 There are five capturing substrings, but only one is ever set after a match.
1591 (An alternative way of solving this problem is to use a "branch reset"
1592 subpattern, as described in the previous section.)
1593 .P
1594 The convenience function for extracting the data by name returns the substring
1595 for the first (and in this example, the only) subpattern of that name that
1596 matched. This saves searching to find which numbered subpattern it was.
1597 .P
1598 If you make a back reference to a non-unique named subpattern from elsewhere in
1599 the pattern, the subpatterns to which the name refers are checked in the order
1600 in which they appear in the overall pattern. The first one that is set is used
1601 for the reference. For example, this pattern matches both "foofoo" and
1602 "barbar" but not "foobar" or "barfoo":
1603 .sp
1604 (?:(?<n>foo)|(?<n>bar))\k<n>
1605 .sp
1606 .P
1607 If you make a subroutine call to a non-unique named subpattern, the one that
1608 corresponds to the first occurrence of the name is used. In the absence of
1609 duplicate numbers (see the previous section) this is the one with the lowest
1610 number.
1611 .P
1612 If you use a named reference in a condition
1613 test (see the
1614 .\"
1615 .\" HTML <a href="#conditions">
1616 .\" </a>
1617 section about conditions
1618 .\"
1619 below), either to check whether a subpattern has matched, or to check for
1620 recursion, all subpatterns with the same name are tested. If the condition is
1621 true for any one of them, the overall condition is true. This is the same
1622 behaviour as testing by number. For further details of the interfaces for
1623 handling named subpatterns, see the
1624 .\" HREF
1625 \fBpcreapi\fP
1626 .\"
1627 documentation.
1628 .P
1629 \fBWarning:\fP You cannot use different names to distinguish between two
1630 subpatterns with the same number because PCRE uses only the numbers when
1631 matching. For this reason, an error is given at compile time if different names
1632 are given to subpatterns with the same number. However, you can always give the
1633 same name to subpatterns with the same number, even when PCRE_DUPNAMES is not
1634 set.
1635 .
1636 .
1638 .rs
1639 .sp
1640 Repetition is specified by quantifiers, which can follow any of the following
1641 items:
1642 .sp
1643 a literal data character
1644 the dot metacharacter
1645 the \eC escape sequence
1646 the \eX escape sequence
1647 the \eR escape sequence
1648 an escape such as \ed or \epL that matches a single character
1649 a character class
1650 a back reference (see next section)
1651 a parenthesized subpattern (including assertions)
1652 a subroutine call to a subpattern (recursive or otherwise)
1653 .sp
1654 The general repetition quantifier specifies a minimum and maximum number of
1655 permitted matches, by giving the two numbers in curly brackets (braces),
1656 separated by a comma. The numbers must be less than 65536, and the first must
1657 be less than or equal to the second. For example:
1658 .sp
1659 z{2,4}
1660 .sp
1661 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1662 character. If the second number is omitted, but the comma is present, there is
1663 no upper limit; if the second number and the comma are both omitted, the
1664 quantifier specifies an exact number of required matches. Thus
1665 .sp
1666 [aeiou]{3,}
1667 .sp
1668 matches at least 3 successive vowels, but may match many more, while
1669 .sp
1670 \ed{8}
1671 .sp
1672 matches exactly 8 digits. An opening curly bracket that appears in a position
1673 where a quantifier is not allowed, or one that does not match the syntax of a
1674 quantifier, is taken as a literal character. For example, {,6} is not a
1675 quantifier, but a literal string of four characters.
1676 .P
1677 In UTF modes, quantifiers apply to characters rather than to individual data
1678 units. Thus, for example, \ex{100}{2} matches two characters, each of
1679 which is represented by a two-byte sequence in a UTF-8 string. Similarly,
1680 \eX{3} matches three Unicode extended grapheme clusters, each of which may be
1681 several data units long (and they may be of different lengths).
1682 .P
1683 The quantifier {0} is permitted, causing the expression to behave as if the
1684 previous item and the quantifier were not present. This may be useful for
1685 subpatterns that are referenced as
1686 .\" HTML <a href="#subpatternsassubroutines">
1687 .\" </a>
1688 subroutines
1689 .\"
1690 from elsewhere in the pattern (but see also the section entitled
1691 .\" HTML <a href="#subdefine">
1692 .\" </a>
1693 "Defining subpatterns for use by reference only"
1694 .\"
1695 below). Items other than subpatterns that have a {0} quantifier are omitted
1696 from the compiled pattern.
1697 .P
1698 For convenience, the three most common quantifiers have single-character
1699 abbreviations:
1700 .sp
1701 * is equivalent to {0,}
1702 + is equivalent to {1,}
1703 ? is equivalent to {0,1}
1704 .sp
1705 It is possible to construct infinite loops by following a subpattern that can
1706 match no characters with a quantifier that has no upper limit, for example:
1707 .sp
1708 (a?)*
1709 .sp
1710 Earlier versions of Perl and PCRE used to give an error at compile time for
1711 such patterns. However, because there are cases where this can be useful, such
1712 patterns are now accepted, but if any repetition of the subpattern does in fact
1713 match no characters, the loop is forcibly broken.
1714 .P
1715 By default, the quantifiers are "greedy", that is, they match as much as
1716 possible (up to the maximum number of permitted times), without causing the
1717 rest of the pattern to fail. The classic example of where this gives problems
1718 is in trying to match comments in C programs. These appear between /* and */
1719 and within the comment, individual * and / characters may appear. An attempt to
1720 match C comments by applying the pattern
1721 .sp
1722 /\e*.*\e*/
1723 .sp
1724 to the string
1725 .sp
1726 /* first comment */ not comment /* second comment */
1727 .sp
1728 fails, because it matches the entire string owing to the greediness of the .*
1729 item.
1730 .P
1731 However, if a quantifier is followed by a question mark, it ceases to be
1732 greedy, and instead matches the minimum number of times possible, so the
1733 pattern
1734 .sp
1735 /\e*.*?\e*/
1736 .sp
1737 does the right thing with the C comments. The meaning of the various
1738 quantifiers is not otherwise changed, just the preferred number of matches.
1739 Do not confuse this use of question mark with its use as a quantifier in its
1740 own right. Because it has two uses, it can sometimes appear doubled, as in
1741 .sp
1742 \ed??\ed
1743 .sp
1744 which matches one digit by preference, but can match two if that is the only
1745 way the rest of the pattern matches.
1746 .P
1747 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1748 the quantifiers are not greedy by default, but individual ones can be made
1749 greedy by following them with a question mark. In other words, it inverts the
1750 default behaviour.
1751 .P
1752 When a parenthesized subpattern is quantified with a minimum repeat count that
1753 is greater than 1 or with a limited maximum, more memory is required for the
1754 compiled pattern, in proportion to the size of the minimum or maximum.
1755 .P
1756 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1757 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1758 implicitly anchored, because whatever follows will be tried against every
1759 character position in the subject string, so there is no point in retrying the
1760 overall match at any position after the first. PCRE normally treats such a
1761 pattern as though it were preceded by \eA.
1762 .P
1763 In cases where it is known that the subject string contains no newlines, it is
1764 worth setting PCRE_DOTALL in order to obtain this optimization, or
1765 alternatively using ^ to indicate anchoring explicitly.
1766 .P
1767 However, there are some cases where the optimization cannot be used. When .*
1768 is inside capturing parentheses that are the subject of a back reference
1769 elsewhere in the pattern, a match at the start may fail where a later one
1770 succeeds. Consider, for example:
1771 .sp
1772 (.*)abc\e1
1773 .sp
1774 If the subject is "xyz123abc123" the match point is the fourth character. For
1775 this reason, such a pattern is not implicitly anchored.
1776 .P
1777 Another case where implicit anchoring is not applied is when the leading .* is
1778 inside an atomic group. Once again, a match at the start may fail where a later
1779 one succeeds. Consider this pattern:
1780 .sp
1781 (?>.*?a)b
1782 .sp
1783 It matches "ab" in the subject "aab". The use of the backtracking control verbs
1784 (*PRUNE) and (*SKIP) also disable this optimization.
1785 .P
1786 When a capturing subpattern is repeated, the value captured is the substring
1787 that matched the final iteration. For example, after
1788 .sp
1789 (tweedle[dume]{3}\es*)+
1790 .sp
1791 has matched "tweedledum tweedledee" the value of the captured substring is
1792 "tweedledee". However, if there are nested capturing subpatterns, the
1793 corresponding captured values may have been set in previous iterations. For
1794 example, after
1795 .sp
1796 /(a|(b))+/
1797 .sp
1798 matches "aba" the value of the second captured substring is "b".
1799 .
1800 .
1801 .\" HTML <a name="atomicgroup"></a>
1803 .rs
1804 .sp
1805 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1806 repetition, failure of what follows normally causes the repeated item to be
1807 re-evaluated to see if a different number of repeats allows the rest of the
1808 pattern to match. Sometimes it is useful to prevent this, either to change the
1809 nature of the match, or to cause it fail earlier than it otherwise might, when
1810 the author of the pattern knows there is no point in carrying on.
1811 .P
1812 Consider, for example, the pattern \ed+foo when applied to the subject line
1813 .sp
1814 123456bar
1815 .sp
1816 After matching all 6 digits and then failing to match "foo", the normal
1817 action of the matcher is to try again with only 5 digits matching the \ed+
1818 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1819 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1820 that once a subpattern has matched, it is not to be re-evaluated in this way.
1821 .P
1822 If we use atomic grouping for the previous example, the matcher gives up
1823 immediately on failing to match "foo" the first time. The notation is a kind of
1824 special parenthesis, starting with (?> as in this example:
1825 .sp
1826 (?>\ed+)foo
1827 .sp
1828 This kind of parenthesis "locks up" the part of the pattern it contains once
1829 it has matched, and a failure further into the pattern is prevented from
1830 backtracking into it. Backtracking past it to previous items, however, works as
1831 normal.
1832 .P
1833 An alternative description is that a subpattern of this type matches the string
1834 of characters that an identical standalone pattern would match, if anchored at
1835 the current point in the subject string.
1836 .P
1837 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1838 the above example can be thought of as a maximizing repeat that must swallow
1839 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1840 number of digits they match in order to make the rest of the pattern match,
1841 (?>\ed+) can only match an entire sequence of digits.
1842 .P
1843 Atomic groups in general can of course contain arbitrarily complicated
1844 subpatterns, and can be nested. However, when the subpattern for an atomic
1845 group is just a single repeated item, as in the example above, a simpler
1846 notation, called a "possessive quantifier" can be used. This consists of an
1847 additional + character following a quantifier. Using this notation, the
1848 previous example can be rewritten as
1849 .sp
1850 \ed++foo
1851 .sp
1852 Note that a possessive quantifier can be used with an entire group, for
1853 example:
1854 .sp
1855 (abc|xyz){2,3}+
1856 .sp
1857 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1858 option is ignored. They are a convenient notation for the simpler forms of
1859 atomic group. However, there is no difference in the meaning of a possessive
1860 quantifier and the equivalent atomic group, though there may be a performance
1861 difference; possessive quantifiers should be slightly faster.
1862 .P
1863 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1864 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1865 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1866 package, and PCRE copied it from there. It ultimately found its way into Perl
1867 at release 5.10.
1868 .P
1869 PCRE has an optimization that automatically "possessifies" certain simple
1870 pattern constructs. For example, the sequence A+B is treated as A++B because
1871 there is no point in backtracking into a sequence of A's when B must follow.
1872 .P
1873 When a pattern contains an unlimited repeat inside a subpattern that can itself
1874 be repeated an unlimited number of times, the use of an atomic group is the
1875 only way to avoid some failing matches taking a very long time indeed. The
1876 pattern
1877 .sp
1878 (\eD+|<\ed+>)*[!?]
1879 .sp
1880 matches an unlimited number of substrings that either consist of non-digits, or
1881 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1882 quickly. However, if it is applied to
1883 .sp
1884 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1885 .sp
1886 it takes a long time before reporting failure. This is because the string can
1887 be divided between the internal \eD+ repeat and the external * repeat in a
1888 large number of ways, and all have to be tried. (The example uses [!?] rather
1889 than a single character at the end, because both PCRE and Perl have an
1890 optimization that allows for fast failure when a single character is used. They
1891 remember the last single character that is required for a match, and fail early
1892 if it is not present in the string.) If the pattern is changed so that it uses
1893 an atomic group, like this:
1894 .sp
1895 ((?>\eD+)|<\ed+>)*[!?]
1896 .sp
1897 sequences of non-digits cannot be broken, and failure happens quickly.
1898 .
1899 .
1900 .\" HTML <a name="backreferences"></a>
1902 .rs
1903 .sp
1904 Outside a character class, a backslash followed by a digit greater than 0 (and
1905 possibly further digits) is a back reference to a capturing subpattern earlier
1906 (that is, to its left) in the pattern, provided there have been that many
1907 previous capturing left parentheses.
1908 .P
1909 However, if the decimal number following the backslash is less than 10, it is
1910 always taken as a back reference, and causes an error only if there are not
1911 that many capturing left parentheses in the entire pattern. In other words, the
1912 parentheses that are referenced need not be to the left of the reference for
1913 numbers less than 10. A "forward back reference" of this type can make sense
1914 when a repetition is involved and the subpattern to the right has participated
1915 in an earlier iteration.
1916 .P
1917 It is not possible to have a numerical "forward back reference" to a subpattern
1918 whose number is 10 or more using this syntax because a sequence such as \e50 is
1919 interpreted as a character defined in octal. See the subsection entitled
1920 "Non-printing characters"
1921 .\" HTML <a href="#digitsafterbackslash">
1922 .\" </a>
1923 above
1924 .\"
1925 for further details of the handling of digits following a backslash. There is
1926 no such problem when named parentheses are used. A back reference to any
1927 subpattern is possible using named parentheses (see below).
1928 .P
1929 Another way of avoiding the ambiguity inherent in the use of digits following a
1930 backslash is to use the \eg escape sequence. This escape must be followed by an
1931 unsigned number or a negative number, optionally enclosed in braces. These
1932 examples are all identical:
1933 .sp
1934 (ring), \e1
1935 (ring), \eg1
1936 (ring), \eg{1}
1937 .sp
1938 An unsigned number specifies an absolute reference without the ambiguity that
1939 is present in the older syntax. It is also useful when literal digits follow
1940 the reference. A negative number is a relative reference. Consider this
1941 example:
1942 .sp
1943 (abc(def)ghi)\eg{-1}
1944 .sp
1945 The sequence \eg{-1} is a reference to the most recently started capturing
1946 subpattern before \eg, that is, is it equivalent to \e2 in this example.
1947 Similarly, \eg{-2} would be equivalent to \e1. The use of relative references
1948 can be helpful in long patterns, and also in patterns that are created by
1949 joining together fragments that contain references within themselves.
1950 .P
1951 A back reference matches whatever actually matched the capturing subpattern in
1952 the current subject string, rather than anything matching the subpattern
1953 itself (see
1954 .\" HTML <a href="#subpatternsassubroutines">
1955 .\" </a>
1956 "Subpatterns as subroutines"
1957 .\"
1958 below for a way of doing that). So the pattern
1959 .sp
1960 (sens|respons)e and \e1ibility
1961 .sp
1962 matches "sense and sensibility" and "response and responsibility", but not
1963 "sense and responsibility". If caseful matching is in force at the time of the
1964 back reference, the case of letters is relevant. For example,
1965 .sp
1966 ((?i)rah)\es+\e1
1967 .sp
1968 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1969 capturing subpattern is matched caselessly.
1970 .P
1971 There are several different ways of writing back references to named
1972 subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1973 \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1974 back reference syntax, in which \eg can be used for both numeric and named
1975 references, is also supported. We could rewrite the above example in any of
1976 the following ways:
1977 .sp
1978 (?<p1>(?i)rah)\es+\ek<p1>
1979 (?'p1'(?i)rah)\es+\ek{p1}
1980 (?P<p1>(?i)rah)\es+(?P=p1)
1981 (?<p1>(?i)rah)\es+\eg{p1}
1982 .sp
1983 A subpattern that is referenced by name may appear in the pattern before or
1984 after the reference.
1985 .P
1986 There may be more than one back reference to the same subpattern. If a
1987 subpattern has not actually been used in a particular match, any back
1988 references to it always fail by default. For example, the pattern
1989 .sp
1990 (a|(bc))\e2
1991 .sp
1992 always fails if it starts to match "a" rather than "bc". However, if the
1993 PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1994 unset value matches an empty string.
1995 .P
1996 Because there may be many capturing parentheses in a pattern, all digits
1997 following a backslash are taken as part of a potential back reference number.
1998 If the pattern continues with a digit character, some delimiter must be used to
1999 terminate the back reference. If the PCRE_EXTENDED option is set, this can be
2000 white space. Otherwise, the \eg{ syntax or an empty comment (see
2001 .\" HTML <a href="#comments">
2002 .\" </a>
2003 "Comments"
2004 .\"
2005 below) can be used.
2006 .
2007 .SS "Recursive back references"
2008 .rs
2009 .sp
2010 A back reference that occurs inside the parentheses to which it refers fails
2011 when the subpattern is first used, so, for example, (a\e1) never matches.
2012 However, such references can be useful inside repeated subpatterns. For
2013 example, the pattern
2014 .sp
2015 (a|b\e1)+
2016 .sp
2017 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
2018 the subpattern, the back reference matches the character string corresponding
2019 to the previous iteration. In order for this to work, the pattern must be such
2020 that the first iteration does not need to match the back reference. This can be
2021 done using alternation, as in the example above, or by a quantifier with a
2022 minimum of zero.
2023 .P
2024 Back references of this type cause the group that they reference to be treated
2025 as an
2026 .\" HTML <a href="#atomicgroup">
2027 .\" </a>
2028 atomic group.
2029 .\"
2030 Once the whole group has been matched, a subsequent matching failure cannot
2031 cause backtracking into the middle of the group.
2032 .
2033 .
2034 .\" HTML <a name="bigassertions"></a>
2036 .rs
2037 .sp
2038 An assertion is a test on the characters following or preceding the current
2039 matching point that does not actually consume any characters. The simple
2040 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
2041 .\" HTML <a href="#smallassertions">
2042 .\" </a>
2043 above.
2044 .\"
2045 .P
2046 More complicated assertions are coded as subpatterns. There are two kinds:
2047 those that look ahead of the current position in the subject string, and those
2048 that look behind it. An assertion subpattern is matched in the normal way,
2049 except that it does not cause the current matching position to be changed.
2050 .P
2051 Assertion subpatterns are not capturing subpatterns. If such an assertion
2052 contains capturing subpatterns within it, these are counted for the purposes of
2053 numbering the capturing subpatterns in the whole pattern. However, substring
2054 capturing is carried out only for positive assertions. (Perl sometimes, but not
2055 always, does do capturing in negative assertions.)
2056 .P
2057 For compatibility with Perl, assertion subpatterns may be repeated; though
2058 it makes no sense to assert the same thing several times, the side effect of
2059 capturing parentheses may occasionally be useful. In practice, there only three
2060 cases:
2061 .sp
2062 (1) If the quantifier is {0}, the assertion is never obeyed during matching.
2063 However, it may contain internal capturing parenthesized groups that are called
2064 from elsewhere via the
2065 .\" HTML <a href="#subpatternsassubroutines">
2066 .\" </a>
2067 subroutine mechanism.
2068 .\"
2069 .sp
2070 (2) If quantifier is {0,n} where n is greater than zero, it is treated as if it
2071 were {0,1}. At run time, the rest of the pattern match is tried with and
2072 without the assertion, the order depending on the greediness of the quantifier.
2073 .sp
2074 (3) If the minimum repetition is greater than zero, the quantifier is ignored.
2075 The assertion is obeyed just once when encountered during matching.
2076 .
2077 .
2078 .SS "Lookahead assertions"
2079 .rs
2080 .sp
2081 Lookahead assertions start with (?= for positive assertions and (?! for
2082 negative assertions. For example,
2083 .sp
2084 \ew+(?=;)
2085 .sp
2086 matches a word followed by a semicolon, but does not include the semicolon in
2087 the match, and
2088 .sp
2089 foo(?!bar)
2090 .sp
2091 matches any occurrence of "foo" that is not followed by "bar". Note that the
2092 apparently similar pattern
2093 .sp
2094 (?!foo)bar
2095 .sp
2096 does not find an occurrence of "bar" that is preceded by something other than
2097 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
2098 (?!foo) is always true when the next three characters are "bar". A
2099 lookbehind assertion is needed to achieve the other effect.
2100 .P
2101 If you want to force a matching failure at some point in a pattern, the most
2102 convenient way to do it is with (?!) because an empty string always matches, so
2103 an assertion that requires there not to be an empty string must always fail.
2104 The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).
2105 .
2106 .
2107 .\" HTML <a name="lookbehind"></a>
2108 .SS "Lookbehind assertions"
2109 .rs
2110 .sp
2111 Lookbehind assertions start with (?<= for positive assertions and (?<! for
2112 negative assertions. For example,
2113 .sp
2114 (?<!foo)bar
2115 .sp
2116 does find an occurrence of "bar" that is not preceded by "foo". The contents of
2117 a lookbehind assertion are restricted such that all the strings it matches must
2118 have a fixed length. However, if there are several top-level alternatives, they
2119 do not all have to have the same fixed length. Thus
2120 .sp
2121 (?<=bullock|donkey)
2122 .sp
2123 is permitted, but
2124 .sp
2125 (?<!dogs?|cats?)
2126 .sp
2127 causes an error at compile time. Branches that match different length strings
2128 are permitted only at the top level of a lookbehind assertion. This is an
2129 extension compared with Perl, which requires all branches to match the same
2130 length of string. An assertion such as
2131 .sp
2132 (?<=ab(c|de))
2133 .sp
2134 is not permitted, because its single top-level branch can match two different
2135 lengths, but it is acceptable to PCRE if rewritten to use two top-level
2136 branches:
2137 .sp
2138 (?<=abc|abde)
2139 .sp
2140 In some cases, the escape sequence \eK
2141 .\" HTML <a href="#resetmatchstart">
2142 .\" </a>
2143 (see above)
2144 .\"
2145 can be used instead of a lookbehind assertion to get round the fixed-length
2146 restriction.
2147 .P
2148 The implementation of lookbehind assertions is, for each alternative, to
2149 temporarily move the current position back by the fixed length and then try to
2150 match. If there are insufficient characters before the current position, the
2151 assertion fails.
2152 .P
2153 In a UTF mode, PCRE does not allow the \eC escape (which matches a single data
2154 unit even in a UTF mode) to appear in lookbehind assertions, because it makes
2155 it impossible to calculate the length of the lookbehind. The \eX and \eR
2156 escapes, which can match different numbers of data units, are also not
2157 permitted.
2158 .P
2159 .\" HTML <a href="#subpatternsassubroutines">
2160 .\" </a>
2161 "Subroutine"
2162 .\"
2163 calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
2164 as the subpattern matches a fixed-length string.
2165 .\" HTML <a href="#recursion">
2166 .\" </a>
2167 Recursion,
2168 .\"
2169 however, is not supported.
2170 .P
2171 Possessive quantifiers can be used in conjunction with lookbehind assertions to
2172 specify efficient matching of fixed-length strings at the end of subject
2173 strings. Consider a simple pattern such as
2174 .sp
2175 abcd$
2176 .sp
2177 when applied to a long string that does not match. Because matching proceeds
2178 from left to right, PCRE will look for each "a" in the subject and then see if
2179 what follows matches the rest of the pattern. If the pattern is specified as
2180 .sp
2181 ^.*abcd$
2182 .sp
2183 the initial .* matches the entire string at first, but when this fails (because
2184 there is no following "a"), it backtracks to match all but the last character,
2185 then all but the last two characters, and so on. Once again the search for "a"
2186 covers the entire string, from right to left, so we are no better off. However,
2187 if the pattern is written as
2188 .sp
2189 ^.*+(?<=abcd)
2190 .sp
2191 there can be no backtracking for the .*+ item; it can match only the entire
2192 string. The subsequent lookbehind assertion does a single test on the last four
2193 characters. If it fails, the match fails immediately. For long strings, this
2194 approach makes a significant difference to the processing time.
2195 .
2196 .
2197 .SS "Using multiple assertions"
2198 .rs
2199 .sp
2200 Several assertions (of any sort) may occur in succession. For example,
2201 .sp
2202 (?<=\ed{3})(?<!999)foo
2203 .sp
2204 matches "foo" preceded by three digits that are not "999". Notice that each of
2205 the assertions is applied independently at the same point in the subject
2206 string. First there is a check that the previous three characters are all
2207 digits, and then there is a check that the same three characters are not "999".
2208 This pattern does \fInot\fP match "foo" preceded by six characters, the first
2209 of which are digits and the last three of which are not "999". For example, it
2210 doesn't match "123abcfoo". A pattern to do that is
2211 .sp
2212 (?<=\ed{3}...)(?<!999)foo
2213 .sp
2214 This time the first assertion looks at the preceding six characters, checking
2215 that the first three are digits, and then the second assertion checks that the
2216 preceding three characters are not "999".
2217 .P
2218 Assertions can be nested in any combination. For example,
2219 .sp
2220 (?<=(?<!foo)bar)baz
2221 .sp
2222 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
2223 preceded by "foo", while
2224 .sp
2225 (?<=\ed{3}(?!999)...)foo
2226 .sp
2227 is another pattern that matches "foo" preceded by three digits and any three
2228 characters that are not "999".
2229 .
2230 .
2231 .\" HTML <a name="conditions"></a>
2233 .rs
2234 .sp
2235 It is possible to cause the matching process to obey a subpattern
2236 conditionally or to choose between two alternative subpatterns, depending on
2237 the result of an assertion, or whether a specific capturing subpattern has
2238 already been matched. The two possible forms of conditional subpattern are:
2239 .sp
2240 (?(condition)yes-pattern)
2241 (?(condition)yes-pattern|no-pattern)
2242 .sp
2243 If the condition is satisfied, the yes-pattern is used; otherwise the
2244 no-pattern (if present) is used. If there are more than two alternatives in the
2245 subpattern, a compile-time error occurs. Each of the two alternatives may
2246 itself contain nested subpatterns of any form, including conditional
2247 subpatterns; the restriction to two alternatives applies only at the level of
2248 the condition. This pattern fragment is an example where the alternatives are
2249 complex:
2250 .sp
2251 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
2252 .sp
2253 .P
2254 There are four kinds of condition: references to subpatterns, references to
2255 recursion, a pseudo-condition called DEFINE, and assertions.
2256 .
2257 .SS "Checking for a used subpattern by number"
2258 .rs
2259 .sp
2260 If the text between the parentheses consists of a sequence of digits, the
2261 condition is true if a capturing subpattern of that number has previously
2262 matched. If there is more than one capturing subpattern with the same number
2263 (see the earlier
2264 .\"
2265 .\" HTML <a href="#recursion">
2266 .\" </a>
2267 section about duplicate subpattern numbers),
2268 .\"
2269 the condition is true if any of them have matched. An alternative notation is
2270 to precede the digits with a plus or minus sign. In this case, the subpattern
2271 number is relative rather than absolute. The most recently opened parentheses
2272 can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside
2273 loops it can also make sense to refer to subsequent groups. The next
2274 parentheses to be opened can be referenced as (?(+1), and so on. (The value
2275 zero in any of these forms is not used; it provokes a compile-time error.)
2276 .P
2277 Consider the following pattern, which contains non-significant white space to
2278 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
2279 three parts for ease of discussion:
2280 .sp
2281 ( \e( )? [^()]+ (?(1) \e) )
2282 .sp
2283 The first part matches an optional opening parenthesis, and if that
2284 character is present, sets it as the first captured substring. The second part
2285 matches one or more characters that are not parentheses. The third part is a
2286 conditional subpattern that tests whether or not the first set of parentheses
2287 matched. If they did, that is, if subject started with an opening parenthesis,
2288 the condition is true, and so the yes-pattern is executed and a closing
2289 parenthesis is required. Otherwise, since no-pattern is not present, the
2290 subpattern matches nothing. In other words, this pattern matches a sequence of
2291 non-parentheses, optionally enclosed in parentheses.
2292 .P
2293 If you were embedding this pattern in a larger one, you could use a relative
2294 reference:
2295 .sp
2296 ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
2297 .sp
2298 This makes the fragment independent of the parentheses in the larger pattern.
2299 .
2300 .SS "Checking for a used subpattern by name"
2301 .rs
2302 .sp
2303 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
2304 subpattern by name. For compatibility with earlier versions of PCRE, which had
2305 this facility before Perl, the syntax (?(name)...) is also recognized. However,
2306 there is a possible ambiguity with this syntax, because subpattern names may
2307 consist entirely of digits. PCRE looks first for a named subpattern; if it
2308 cannot find one and the name consists entirely of digits, PCRE looks for a
2309 subpattern of that number, which must be greater than zero. Using subpattern
2310 names that consist entirely of digits is not recommended.
2311 .P
2312 Rewriting the above example to use a named subpattern gives this:
2313 .sp
2314 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
2315 .sp
2316 If the name used in a condition of this kind is a duplicate, the test is
2317 applied to all subpatterns of the same name, and is true if any one of them has
2318 matched.
2319 .
2320 .SS "Checking for pattern recursion"
2321 .rs
2322 .sp
2323 If the condition is the string (R), and there is no subpattern with the name R,
2324 the condition is true if a recursive call to the whole pattern or any
2325 subpattern has been made. If digits or a name preceded by ampersand follow the
2326 letter R, for example:
2327 .sp
2328 (?(R3)...) or (?(R&name)...)
2329 .sp
2330 the condition is true if the most recent recursion is into a subpattern whose
2331 number or name is given. This condition does not check the entire recursion
2332 stack. If the name used in a condition of this kind is a duplicate, the test is
2333 applied to all subpatterns of the same name, and is true if any one of them is
2334 the most recent recursion.
2335 .P
2336 At "top level", all these recursion test conditions are false.
2337 .\" HTML <a href="#recursion">
2338 .\" </a>
2339 The syntax for recursive patterns
2340 .\"
2341 is described below.
2342 .
2343 .\" HTML <a name="subdefine"></a>
2344 .SS "Defining subpatterns for use by reference only"
2345 .rs
2346 .sp
2347 If the condition is the string (DEFINE), and there is no subpattern with the
2348 name DEFINE, the condition is always false. In this case, there may be only one
2349 alternative in the subpattern. It is always skipped if control reaches this
2350 point in the pattern; the idea of DEFINE is that it can be used to define
2351 subroutines that can be referenced from elsewhere. (The use of
2352 .\" HTML <a href="#subpatternsassubroutines">
2353 .\" </a>
2354 subroutines
2355 .\"
2356 is described below.) For example, a pattern to match an IPv4 address such as
2357 "" could be written like this (ignore white space and line
2358 breaks):
2359 .sp
2360 (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
2361 \eb (?&byte) (\e.(?&byte)){3} \eb
2362 .sp
2363 The first part of the pattern is a DEFINE group inside which a another group
2364 named "byte" is defined. This matches an individual component of an IPv4
2365 address (a number less than 256). When matching takes place, this part of the
2366 pattern is skipped because DEFINE acts like a false condition. The rest of the
2367 pattern uses references to the named group to match the four dot-separated
2368 components of an IPv4 address, insisting on a word boundary at each end.
2369 .
2370 .SS "Assertion conditions"
2371 .rs
2372 .sp
2373 If the condition is not in any of the above formats, it must be an assertion.
2374 This may be a positive or negative lookahead or lookbehind assertion. Consider
2375 this pattern, again containing non-significant white space, and with the two
2376 alternatives on the second line:
2377 .sp
2378 (?(?=[^a-z]*[a-z])
2379 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
2380 .sp
2381 The condition is a positive lookahead assertion that matches an optional
2382 sequence of non-letters followed by a letter. In other words, it tests for the
2383 presence of at least one letter in the subject. If a letter is found, the
2384 subject is matched against the first alternative; otherwise it is matched
2385 against the second. This pattern matches strings in one of the two forms
2386 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2387 .
2388 .
2389 .\" HTML <a name="comments"></a>
2391 .rs
2392 .sp
2393 There are two ways of including comments in patterns that are processed by
2394 PCRE. In both cases, the start of the comment must not be in a character class,
2395 nor in the middle of any other sequence of related characters such as (?: or a
2396 subpattern name or number. The characters that make up a comment play no part
2397 in the pattern matching.
2398 .P
2399 The sequence (?# marks the start of a comment that continues up to the next
2400 closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED
2401 option is set, an unescaped # character also introduces a comment, which in
2402 this case continues to immediately after the next newline character or
2403 character sequence in the pattern. Which characters are interpreted as newlines
2404 is controlled by the options passed to a compiling function or by a special
2405 sequence at the start of the pattern, as described in the section entitled
2406 .\" HTML <a href="#newlines">
2407 .\" </a>
2408 "Newline conventions"
2409 .\"
2410 above. Note that the end of this type of comment is a literal newline sequence
2411 in the pattern; escape sequences that happen to represent a newline do not
2412 count. For example, consider this pattern when PCRE_EXTENDED is set, and the
2413 default newline convention is in force:
2414 .sp
2415 abc #comment \en still comment
2416 .sp
2417 On encountering the # character, \fBpcre_compile()\fP skips along, looking for
2418 a newline in the pattern. The sequence \en is still literal at this stage, so
2419 it does not terminate the comment. Only an actual character with the code value
2420 0x0a (the default newline) does so.
2421 .
2422 .
2423 .\" HTML <a name="recursion"></a>
2425 .rs
2426 .sp
2427 Consider the problem of matching a string in parentheses, allowing for
2428 unlimited nested parentheses. Without the use of recursion, the best that can
2429 be done is to use a pattern that matches up to some fixed depth of nesting. It
2430 is not possible to handle an arbitrary nesting depth.
2431 .P
2432 For some time, Perl has provided a facility that allows regular expressions to
2433 recurse (amongst other things). It does this by interpolating Perl code in the
2434 expression at run time, and the code can refer to the expression itself. A Perl
2435 pattern using code interpolation to solve the parentheses problem can be
2436 created like this:
2437 .sp
2438 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
2439 .sp
2440 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2441 recursively to the pattern in which it appears.
2442 .P
2443 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2444 supports special syntax for recursion of the entire pattern, and also for
2445 individual subpattern recursion. After its introduction in PCRE and Python,
2446 this kind of recursion was subsequently introduced into Perl at release 5.10.
2447 .P
2448 A special item that consists of (? followed by a number greater than zero and a
2449 closing parenthesis is a recursive subroutine call of the subpattern of the
2450 given number, provided that it occurs inside that subpattern. (If not, it is a
2451 .\" HTML <a href="#subpatternsassubroutines">
2452 .\" </a>
2453 non-recursive subroutine
2454 .\"
2455 call, which is described in the next section.) The special item (?R) or (?0) is
2456 a recursive call of the entire regular expression.
2457 .P
2458 This PCRE pattern solves the nested parentheses problem (assume the
2459 PCRE_EXTENDED option is set so that white space is ignored):
2460 .sp
2461 \e( ( [^()]++ | (?R) )* \e)
2462 .sp
2463 First it matches an opening parenthesis. Then it matches any number of
2464 substrings which can either be a sequence of non-parentheses, or a recursive
2465 match of the pattern itself (that is, a correctly parenthesized substring).
2466 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2467 to avoid backtracking into sequences of non-parentheses.
2468 .P
2469 If this were part of a larger pattern, you would not want to recurse the entire
2470 pattern, so instead you could use this:
2471 .sp
2472 ( \e( ( [^()]++ | (?1) )* \e) )
2473 .sp
2474 We have put the pattern into parentheses, and caused the recursion to refer to
2475 them instead of the whole pattern.
2476 .P
2477 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2478 is made easier by the use of relative references. Instead of (?1) in the
2479 pattern above you can write (?-2) to refer to the second most recently opened
2480 parentheses preceding the recursion. In other words, a negative number counts
2481 capturing parentheses leftwards from the point at which it is encountered.
2482 .P
2483 It is also possible to refer to subsequently opened parentheses, by writing
2484 references such as (?+2). However, these cannot be recursive because the
2485 reference is not inside the parentheses that are referenced. They are always
2486 .\" HTML <a href="#subpatternsassubroutines">
2487 .\" </a>
2488 non-recursive subroutine
2489 .\"
2490 calls, as described in the next section.
2491 .P
2492 An alternative approach is to use named parentheses instead. The Perl syntax
2493 for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
2494 could rewrite the above example as follows:
2495 .sp
2496 (?<pn> \e( ( [^()]++ | (?&pn) )* \e) )
2497 .sp
2498 If there is more than one subpattern with the same name, the earliest one is
2499 used.
2500 .P
2501 This particular example pattern that we have been looking at contains nested
2502 unlimited repeats, and so the use of a possessive quantifier for matching
2503 strings of non-parentheses is important when applying the pattern to strings
2504 that do not match. For example, when this pattern is applied to
2505 .sp
2506 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2507 .sp
2508 it yields "no match" quickly. However, if a possessive quantifier is not used,
2509 the match runs for a very long time indeed because there are so many different
2510 ways the + and * repeats can carve up the subject, and all have to be tested
2511 before failure can be reported.
2512 .P
2513 At the end of a match, the values of capturing parentheses are those from
2514 the outermost level. If you want to obtain intermediate values, a callout
2515 function can be used (see below and the
2516 .\" HREF
2517 \fBpcrecallout\fP
2518 .\"
2519 documentation). If the pattern above is matched against
2520 .sp
2521 (ab(cd)ef)
2522 .sp
2523 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2524 the last value taken on at the top level. If a capturing subpattern is not
2525 matched at the top level, its final captured value is unset, even if it was
2526 (temporarily) set at a deeper level during the matching process.
2527 .P
2528 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2529 obtain extra memory to store data during a recursion, which it does by using
2530 \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no memory can
2531 be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2532 .P
2533 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2534 Consider this pattern, which matches text in angle brackets, allowing for
2535 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2536 recursing), whereas any characters are permitted at the outer level.
2537 .sp
2538 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
2539 .sp
2540 In this pattern, (?(R) is the start of a conditional subpattern, with two
2541 different alternatives for the recursive and non-recursive cases. The (?R) item
2542 is the actual recursive call.
2543 .
2544 .
2545 .\" HTML <a name="recursiondifference"></a>
2546 .SS "Differences in recursion processing between PCRE and Perl"
2547 .rs
2548 .sp
2549 Recursion processing in PCRE differs from Perl in two important ways. In PCRE
2550 (like Python, but unlike Perl), a recursive subpattern call is always treated
2551 as an atomic group. That is, once it has matched some of the subject string, it
2552 is never re-entered, even if it contains untried alternatives and there is a
2553 subsequent matching failure. This can be illustrated by the following pattern,
2554 which purports to match a palindromic string that contains an odd number of
2555 characters (for example, "a", "aba", "abcba", "abcdcba"):
2556 .sp
2557 ^(.|(.)(?1)\e2)$
2558 .sp
2559 The idea is that it either matches a single character, or two identical
2560 characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2561 it does not if the pattern is longer than three characters. Consider the
2562 subject string "abcba":
2563 .P
2564 At the top level, the first character is matched, but as it is not at the end
2565 of the string, the first alternative fails; the second alternative is taken
2566 and the recursion kicks in. The recursive call to subpattern 1 successfully
2567 matches the next character ("b"). (Note that the beginning and end of line
2568 tests are not part of the recursion).
2569 .P
2570 Back at the top level, the next character ("c") is compared with what
2571 subpattern 2 matched, which was "a". This fails. Because the recursion is
2572 treated as an atomic group, there are now no backtracking points, and so the
2573 entire match fails. (Perl is able, at this point, to re-enter the recursion and
2574 try the second alternative.) However, if the pattern is written with the
2575 alternatives in the other order, things are different:
2576 .sp
2577 ^((.)(?1)\e2|.)$
2578 .sp
2579 This time, the recursing alternative is tried first, and continues to recurse
2580 until it runs out of characters, at which point the recursion fails. But this
2581 time we do have another alternative to try at the higher level. That is the big
2582 difference: in the previous case the remaining alternative is at a deeper
2583 recursion level, which PCRE cannot use.
2584 .P
2585 To change the pattern so that it matches all palindromic strings, not just
2586 those with an odd number of characters, it is tempting to change the pattern to
2587 this:
2588 .sp
2589 ^((.)(?1)\e2|.?)$
2590 .sp
2591 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2592 deeper recursion has matched a single character, it cannot be entered again in
2593 order to match an empty string. The solution is to separate the two cases, and
2594 write out the odd and even cases as alternatives at the higher level:
2595 .sp
2596 ^(?:((.)(?1)\e2|)|((.)(?3)\e4|.))
2597 .sp
2598 If you want to match typical palindromic phrases, the pattern has to ignore all
2599 non-word characters, which can be done like this:
2600 .sp
2601 ^\eW*+(?:((.)\eW*+(?1)\eW*+\e2|)|((.)\eW*+(?3)\eW*+\e4|\eW*+.\eW*+))\eW*+$
2602 .sp
2603 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2604 man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2605 the use of the possessive quantifier *+ to avoid backtracking into sequences of
2606 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2607 more) to match typical phrases, and Perl takes so long that you think it has
2608 gone into a loop.
2609 .P
2610 \fBWARNING\fP: The palindrome-matching patterns above work only if the subject
2611 string does not start with a palindrome that is shorter than the entire string.
2612 For example, although "abcba" is correctly matched, if the subject is "ababa",
2613 PCRE finds the palindrome "aba" at the start, then fails at top level because
2614 the end of the string does not follow. Once again, it cannot jump back into the
2615 recursion to try other alternatives, so the entire match fails.
2616 .P
2617 The second way in which PCRE and Perl differ in their recursion processing is
2618 in the handling of captured values. In Perl, when a subpattern is called
2619 recursively or as a subpattern (see the next section), it has no access to any
2620 values that were captured outside the recursion, whereas in PCRE these values
2621 can be referenced. Consider this pattern:
2622 .sp
2623 ^(.)(\e1|a(?2))
2624 .sp
2625 In PCRE, this pattern matches "bab". The first capturing parentheses match "b",
2626 then in the second group, when the back reference \e1 fails to match "b", the
2627 second alternative matches "a" and then recurses. In the recursion, \e1 does
2628 now match "b" and so the whole match succeeds. In Perl, the pattern fails to
2629 match because inside the recursive call \e1 cannot access the externally set
2630 value.
2631 .
2632 .
2633 .\" HTML <a name="subpatternsassubroutines"></a>
2635 .rs
2636 .sp
2637 If the syntax for a recursive subpattern call (either by number or by
2638 name) is used outside the parentheses to which it refers, it operates like a
2639 subroutine in a programming language. The called subpattern may be defined
2640 before or after the reference. A numbered reference can be absolute or
2641 relative, as in these examples:
2642 .sp
2643 (...(absolute)...)...(?2)...
2644 (...(relative)...)...(?-1)...
2645 (...(?+1)...(relative)...
2646 .sp
2647 An earlier example pointed out that the pattern
2648 .sp
2649 (sens|respons)e and \e1ibility
2650 .sp
2651 matches "sense and sensibility" and "response and responsibility", but not
2652 "sense and responsibility". If instead the pattern
2653 .sp
2654 (sens|respons)e and (?1)ibility
2655 .sp
2656 is used, it does match "sense and responsibility" as well as the other two
2657 strings. Another example is given in the discussion of DEFINE above.
2658 .P
2659 All subroutine calls, whether recursive or not, are always treated as atomic
2660 groups. That is, once a subroutine has matched some of the subject string, it
2661 is never re-entered, even if it contains untried alternatives and there is a
2662 subsequent matching failure. Any capturing parentheses that are set during the
2663 subroutine call revert to their previous values afterwards.
2664 .P
2665 Processing options such as case-independence are fixed when a subpattern is
2666 defined, so if it is used as a subroutine, such options cannot be changed for
2667 different calls. For example, consider this pattern:
2668 .sp
2669 (abc)(?i:(?-1))
2670 .sp
2671 It matches "abcabc". It does not match "abcABC" because the change of
2672 processing option does not affect the called subpattern.
2673 .
2674 .
2675 .\" HTML <a name="onigurumasubroutines"></a>
2677 .rs
2678 .sp
2679 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
2680 a number enclosed either in angle brackets or single quotes, is an alternative
2681 syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2682 are two of the examples used above, rewritten using this syntax:
2683 .sp
2684 (?<pn> \e( ( (?>[^()]+) | \eg<pn> )* \e) )
2685 (sens|respons)e and \eg'1'ibility
2686 .sp
2687 PCRE supports an extension to Oniguruma: if a number is preceded by a
2688 plus or a minus sign it is taken as a relative reference. For example:
2689 .sp
2690 (abc)(?i:\eg<-1>)
2691 .sp
2692 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
2693 synonymous. The former is a back reference; the latter is a subroutine call.
2694 .
2695 .
2697 .rs
2698 .sp
2699 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2700 code to be obeyed in the middle of matching a regular expression. This makes it
2701 possible, amongst other things, to extract different substrings that match the
2702 same pair of parentheses when there is a repetition.
2703 .P
2704 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2705 code. The feature is called "callout". The caller of PCRE provides an external
2706 function by putting its entry point in the global variable \fIpcre_callout\fP
2707 (8-bit library) or \fIpcre[16|32]_callout\fP (16-bit or 32-bit library).
2708 By default, this variable contains NULL, which disables all calling out.
2709 .P
2710 Within a regular expression, (?C) indicates the points at which the external
2711 function is to be called. If you want to identify different callout points, you
2712 can put a number less than 256 after the letter C. The default value is zero.
2713 For example, this pattern has two callout points:
2714 .sp
2715 (?C1)abc(?C2)def
2716 .sp
2717 If the PCRE_AUTO_CALLOUT flag is passed to a compiling function, callouts are
2718 automatically installed before each item in the pattern. They are all numbered
2719 255. If there is a conditional group in the pattern whose condition is an
2720 assertion, an additional callout is inserted just before the condition. An
2721 explicit callout may also be set at this position, as in this example:
2722 .sp
2723 (?(?C9)(?=a)abc|def)
2724 .sp
2725 Note that this applies only to assertion conditions, not to other types of
2726 condition.
2727 .P
2728 During matching, when PCRE reaches a callout point, the external function is
2729 called. It is provided with the number of the callout, the position in the
2730 pattern, and, optionally, one item of data originally supplied by the caller of
2731 the matching function. The callout function may cause matching to proceed, to
2732 backtrack, or to fail altogether. A complete description of the interface to
2733 the callout function is given in the
2734 .\" HREF
2735 \fBpcrecallout\fP
2736 .\"
2737 documentation.
2738 .
2739 .
2740 .\" HTML <a name="backtrackcontrol"></a>
2742 .rs
2743 .sp
2744 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2745 are still described in the Perl documentation as "experimental and subject to
2746 change or removal in a future version of Perl". It goes on to say: "Their usage
2747 in production code should be noted to avoid problems during upgrades." The same
2748 remarks apply to the PCRE features described in this section.
2749 .P
2750 The new verbs make use of what was previously invalid syntax: an opening
2751 parenthesis followed by an asterisk. They are generally of the form
2752 (*VERB) or (*VERB:NAME). Some may take either form, possibly behaving
2753 differently depending on whether or not a name is present. A name is any
2754 sequence of characters that does not include a closing parenthesis. The maximum
2755 length of name is 255 in the 8-bit library and 65535 in the 16-bit and 32-bit
2756 libraries. If the name is empty, that is, if the closing parenthesis
2757 immediately follows the colon, the effect is as if the colon were not there.
2758 Any number of these verbs may occur in a pattern.
2759 .P
2760 Since these verbs are specifically related to backtracking, most of them can be
2761 used only when the pattern is to be matched using one of the traditional
2762 matching functions, because these use a backtracking algorithm. With the
2763 exception of (*FAIL), which behaves like a failing negative assertion, the
2764 backtracking control verbs cause an error if encountered by a DFA matching
2765 function.
2766 .P
2767 The behaviour of these verbs in
2768 .\" HTML <a href="#btrepeat">
2769 .\" </a>
2770 repeated groups,
2771 .\"
2772 .\" HTML <a href="#btassert">
2773 .\" </a>
2774 assertions,
2775 .\"
2776 and in
2777 .\" HTML <a href="#btsub">
2778 .\" </a>
2779 subpatterns called as subroutines
2780 .\"
2781 (whether or not recursively) is documented below.
2782 .
2783 .
2784 .\" HTML <a name="nooptimize"></a>
2785 .SS "Optimizations that affect backtracking verbs"
2786 .rs
2787 .sp
2788 PCRE contains some optimizations that are used to speed up matching by running
2789 some checks at the start of each match attempt. For example, it may know the
2790 minimum length of matching subject, or that a particular character must be
2791 present. When one of these optimizations bypasses the running of a match, any
2792 included backtracking verbs will not, of course, be processed. You can suppress
2793 the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2794 when calling \fBpcre_compile()\fP or \fBpcre_exec()\fP, or by starting the
2795 pattern with (*NO_START_OPT). There is more discussion of this option in the
2796 section entitled
2797 .\" HTML <a href="pcreapi.html#execoptions">
2798 .\" </a>
2799 "Option bits for \fBpcre_exec()\fP"
2800 .\"
2801 in the
2802 .\" HREF
2803 \fBpcreapi\fP
2804 .\"
2805 documentation.
2806 .P
2807 Experiments with Perl suggest that it too has similar optimizations, sometimes
2808 leading to anomalous results.
2809 .
2810 .
2811 .SS "Verbs that act immediately"
2812 .rs
2813 .sp
2814 The following verbs act as soon as they are encountered. They may not be
2815 followed by a name.
2816 .sp
2817 (*ACCEPT)
2818 .sp
2819 This verb causes the match to end successfully, skipping the remainder of the
2820 pattern. However, when it is inside a subpattern that is called as a
2821 subroutine, only that subpattern is ended successfully. Matching then continues
2822 at the outer level. If (*ACCEPT) in triggered in a positive assertion, the
2823 assertion succeeds; in a negative assertion, the assertion fails.
2824 .P
2825 If (*ACCEPT) is inside capturing parentheses, the data so far is captured. For
2826 example:
2827 .sp
2828 A((?:A|B(*ACCEPT)|C)D)
2829 .sp
2830 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2831 the outer parentheses.
2832 .sp
2833 (*FAIL) or (*F)
2834 .sp
2835 This verb causes a matching failure, forcing backtracking to occur. It is
2836 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2837 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2838 Perl features that are not present in PCRE. The nearest equivalent is the
2839 callout feature, as for example in this pattern:
2840 .sp
2841 a+(?C)(*FAIL)
2842 .sp
2843 A match with the string "aaaa" always fails, but the callout is taken before
2844 each backtrack happens (in this example, 10 times).
2845 .
2846 .
2847 .SS "Recording which path was taken"
2848 .rs
2849 .sp
2850 There is one verb whose main purpose is to track how a match was arrived at,
2851 though it also has a secondary use in conjunction with advancing the match
2852 starting point (see (*SKIP) below).
2853 .sp
2854 (*MARK:NAME) or (*:NAME)
2855 .sp
2856 A name is always required with this verb. There may be as many instances of
2857 (*MARK) as you like in a pattern, and their names do not have to be unique.
2858 .P
2859 When a match succeeds, the name of the last-encountered (*MARK:NAME),
2860 (*PRUNE:NAME), or (*THEN:NAME) on the matching path is passed back to the
2861 caller as described in the section entitled
2862 .\" HTML <a href="pcreapi.html#extradata">
2863 .\" </a>
2864 "Extra data for \fBpcre_exec()\fP"
2865 .\"
2866 in the
2867 .\" HREF
2868 \fBpcreapi\fP
2869 .\"
2870 documentation. Here is an example of \fBpcretest\fP output, where the /K
2871 modifier requests the retrieval and outputting of (*MARK) data:
2872 .sp
2873 re> /X(*MARK:A)Y|X(*MARK:B)Z/K
2874 data> XY
2875 0: XY
2876 MK: A
2877 XZ
2878 0: XZ
2879 MK: B
2880 .sp
2881 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2882 indicates which of the two alternatives matched. This is a more efficient way
2883 of obtaining this information than putting each alternative in its own
2884 capturing parentheses.
2885 .P
2886 If a verb with a name is encountered in a positive assertion that is true, the
2887 name is recorded and passed back if it is the last-encountered. This does not
2888 happen for negative assertions or failing positive assertions.
2889 .P
2890 After a partial match or a failed match, the last encountered name in the
2891 entire match process is returned. For example:
2892 .sp
2893 re> /X(*MARK:A)Y|X(*MARK:B)Z/K
2894 data> XP
2895 No match, mark = B
2896 .sp
2897 Note that in this unanchored example the mark is retained from the match
2898 attempt that started at the letter "X" in the subject. Subsequent match
2899 attempts starting at "P" and then with an empty string do not get as far as the
2900 (*MARK) item, but nevertheless do not reset it.
2901 .P
2902 If you are interested in (*MARK) values after failed matches, you should
2903 probably set the PCRE_NO_START_OPTIMIZE option
2904 .\" HTML <a href="#nooptimize">
2905 .\" </a>
2906 (see above)
2907 .\"
2908 to ensure that the match is always attempted.
2909 .
2910 .
2911 .SS "Verbs that act after backtracking"
2912 .rs
2913 .sp
2914 The following verbs do nothing when they are encountered. Matching continues
2915 with what follows, but if there is no subsequent match, causing a backtrack to
2916 the verb, a failure is forced. That is, backtracking cannot pass to the left of
2917 the verb. However, when one of these verbs appears inside an atomic group or an
2918 assertion that is true, its effect is confined to that group, because once the
2919 group has been matched, there is never any backtracking into it. In this
2920 situation, backtracking can "jump back" to the left of the entire atomic group
2921 or assertion. (Remember also, as stated above, that this localization also
2922 applies in subroutine calls.)
2923 .P
2924 These verbs differ in exactly what kind of failure occurs when backtracking
2925 reaches them. The behaviour described below is what happens when the verb is
2926 not in a subroutine or an assertion. Subsequent sections cover these special
2927 cases.
2928 .sp
2929 (*COMMIT)
2930 .sp
2931 This verb, which may not be followed by a name, causes the whole match to fail
2932 outright if there is a later matching failure that causes backtracking to reach
2933 it. Even if the pattern is unanchored, no further attempts to find a match by
2934 advancing the starting point take place. If (*COMMIT) is the only backtracking
2935 verb that is encountered, once it has been passed \fBpcre_exec()\fP is
2936 committed to finding a match at the current starting point, or not at all. For
2937 example:
2938 .sp
2939 a+(*COMMIT)b
2940 .sp
2941 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2942 dynamic anchor, or "I've started, so I must finish." The name of the most
2943 recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2944 match failure.
2945 .P
2946 If there is more than one backtracking verb in a pattern, a different one that
2947 follows (*COMMIT) may be triggered first, so merely passing (*COMMIT) during a
2948 match does not always guarantee that a match must be at this starting point.
2949 .P
2950 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2951 unless PCRE's start-of-match optimizations are turned off, as shown in this
2952 \fBpcretest\fP example:
2953 .sp
2954 re> /(*COMMIT)abc/
2955 data> xyzabc
2956 0: abc
2957 xyzabc\eY
2958 No match
2959 .sp
2960 PCRE knows that any match must start with "a", so the optimization skips along
2961 the subject to "a" before running the first match attempt, which succeeds. When
2962 the optimization is disabled by the \eY escape in the second subject, the match
2963 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2964 starting points.
2965 .sp
2966 (*PRUNE) or (*PRUNE:NAME)
2967 .sp
2968 This verb causes the match to fail at the current starting position in the
2969 subject if there is a later matching failure that causes backtracking to reach
2970 it. If the pattern is unanchored, the normal "bumpalong" advance to the next
2971 starting character then happens. Backtracking can occur as usual to the left of
2972 (*PRUNE), before it is reached, or when matching to the right of (*PRUNE), but
2973 if there is no match to the right, backtracking cannot cross (*PRUNE). In
2974 simple cases, the use of (*PRUNE) is just an alternative to an atomic group or
2975 possessive quantifier, but there are some uses of (*PRUNE) that cannot be
2976 expressed in any other way. In an anchored pattern (*PRUNE) has the same effect
2977 as (*COMMIT).
2978 .P
2979 The behaviour of (*PRUNE:NAME) is the not the same as (*MARK:NAME)(*PRUNE).
2980 It is like (*MARK:NAME) in that the name is remembered for passing back to the
2981 caller. However, (*SKIP:NAME) searches only for names set with (*MARK).
2982 .sp
2983 (*SKIP)
2984 .sp
2985 This verb, when given without a name, is like (*PRUNE), except that if the
2986 pattern is unanchored, the "bumpalong" advance is not to the next character,
2987 but to the position in the subject where (*SKIP) was encountered. (*SKIP)
2988 signifies that whatever text was matched leading up to it cannot be part of a
2989 successful match. Consider:
2990 .sp
2991 a+(*SKIP)b
2992 .sp
2993 If the subject is "aaaac...", after the first match attempt fails (starting at
2994 the first character in the string), the starting point skips on to start the
2995 next attempt at "c". Note that a possessive quantifer does not have the same
2996 effect as this example; although it would suppress backtracking during the
2997 first match attempt, the second attempt would start at the second character
2998 instead of skipping on to "c".
2999 .sp
3000 (*SKIP:NAME)
3001 .sp
3002 When (*SKIP) has an associated name, its behaviour is modified. When it is
3003 triggered, the previous path through the pattern is searched for the most
3004 recent (*MARK) that has the same name. If one is found, the "bumpalong" advance
3005 is to the subject position that corresponds to that (*MARK) instead of to where
3006 (*SKIP) was encountered. If no (*MARK) with a matching name is found, the
3007 (*SKIP) is ignored.
3008 .P
3009 Note that (*SKIP:NAME) searches only for names set by (*MARK:NAME). It ignores
3010 names that are set by (*PRUNE:NAME) or (*THEN:NAME).
3011 .sp
3012 (*THEN) or (*THEN:NAME)
3013 .sp
3014 This verb causes a skip to the next innermost alternative when backtracking
3015 reaches it. That is, it cancels any further backtracking within the current
3016 alternative. Its name comes from the observation that it can be used for a
3017 pattern-based if-then-else block:
3018 .sp
3019 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
3020 .sp
3021 If the COND1 pattern matches, FOO is tried (and possibly further items after
3022 the end of the group if FOO succeeds); on failure, the matcher skips to the
3023 second alternative and tries COND2, without backtracking into COND1. If that
3024 succeeds and BAR fails, COND3 is tried. If subsequently BAZ fails, there are no
3025 more alternatives, so there is a backtrack to whatever came before the entire
3026 group. If (*THEN) is not inside an alternation, it acts like (*PRUNE).
3027 .P
3028 The behaviour of (*THEN:NAME) is the not the same as (*MARK:NAME)(*THEN).
3029 It is like (*MARK:NAME) in that the name is remembered for passing back to the
3030 caller. However, (*SKIP:NAME) searches only for names set with (*MARK).
3031 .P
3032 A subpattern that does not contain a | character is just a part of the
3033 enclosing alternative; it is not a nested alternation with only one
3034 alternative. The effect of (*THEN) extends beyond such a subpattern to the
3035 enclosing alternative. Consider this pattern, where A, B, etc. are complex
3036 pattern fragments that do not contain any | characters at this level:
3037 .sp
3038 A (B(*THEN)C) | D
3039 .sp
3040 If A and B are matched, but there is a failure in C, matching does not
3041 backtrack into A; instead it moves to the next alternative, that is, D.
3042 However, if the subpattern containing (*THEN) is given an alternative, it
3043 behaves differently:
3044 .sp
3045 A (B(*THEN)C | (*FAIL)) | D
3046 .sp
3047 The effect of (*THEN) is now confined to the inner subpattern. After a failure
3048 in C, matching moves to (*FAIL), which causes the whole subpattern to fail
3049 because there are no more alternatives to try. In this case, matching does now
3050 backtrack into A.
3051 .P
3052 Note that a conditional subpattern is not considered as having two
3053 alternatives, because only one is ever used. In other words, the | character in
3054 a conditional subpattern has a different meaning. Ignoring white space,
3055 consider:
3056 .sp
3057 ^.*? (?(?=a) a | b(*THEN)c )
3058 .sp
3059 If the subject is "ba", this pattern does not match. Because .*? is ungreedy,
3060 it initially matches zero characters. The condition (?=a) then fails, the
3061 character "b" is matched, but "c" is not. At this point, matching does not
3062 backtrack to .*? as might perhaps be expected from the presence of the |
3063 character. The conditional subpattern is part of the single alternative that
3064 comprises the whole pattern, and so the match fails. (If there was a backtrack
3065 into .*?, allowing it to match "b", the match would succeed.)
3066 .P
3067 The verbs just described provide four different "strengths" of control when
3068 subsequent matching fails. (*THEN) is the weakest, carrying on the match at the
3069 next alternative. (*PRUNE) comes next, failing the match at the current
3070 starting position, but allowing an advance to the next character (for an
3071 unanchored pattern). (*SKIP) is similar, except that the advance may be more
3072 than one character. (*COMMIT) is the strongest, causing the entire match to
3073 fail.
3074 .
3075 .
3076 .SS "More than one backtracking verb"
3077 .rs
3078 .sp
3079 If more than one backtracking verb is present in a pattern, the one that is
3080 backtracked onto first acts. For example, consider this pattern, where A, B,
3081 etc. are complex pattern fragments:
3082 .sp
3084 .sp
3085 If A matches but B fails, the backtrack to (*COMMIT) causes the entire match to
3086 fail. However, if A and B match, but C fails, the backtrack to (*THEN) causes
3087 the next alternative (ABD) to be tried. This behaviour is consistent, but is
3088 not always the same as Perl's. It means that if two or more backtracking verbs
3089 appear in succession, all the the last of them has no effect. Consider this
3090 example:
3091 .sp
3092 ...(*COMMIT)(*PRUNE)...
3093 .sp
3094 If there is a matching failure to the right, backtracking onto (*PRUNE) causes
3095 it to be triggered, and its action is taken. There can never be a backtrack
3096 onto (*COMMIT).
3097 .
3098 .
3099 .\" HTML <a name="btrepeat"></a>
3100 .SS "Backtracking verbs in repeated groups"
3101 .rs
3102 .sp
3103 PCRE differs from Perl in its handling of backtracking verbs in repeated
3104 groups. For example, consider:
3105 .sp
3106 /(a(*COMMIT)b)+ac/
3107 .sp
3108 If the subject is "abac", Perl matches, but PCRE fails because the (*COMMIT) in
3109 the second repeat of the group acts.
3110 .
3111 .
3112 .\" HTML <a name="btassert"></a>
3113 .SS "Backtracking verbs in assertions"
3114 .rs
3115 .sp
3116 (*FAIL) in an assertion has its normal effect: it forces an immediate backtrack.
3117 .P
3118 (*ACCEPT) in a positive assertion causes the assertion to succeed without any
3119 further processing. In a negative assertion, (*ACCEPT) causes the assertion to
3120 fail without any further processing.
3121 .P
3122 The other backtracking verbs are not treated specially if they appear in a
3123 positive assertion. In particular, (*THEN) skips to the next alternative in the
3124 innermost enclosing group that has alternations, whether or not this is within
3125 the assertion.
3126 .P
3127 Negative assertions are, however, different, in order to ensure that changing a
3128 positive assertion into a negative assertion changes its result. Backtracking
3129 into (*COMMIT), (*SKIP), or (*PRUNE) causes a negative assertion to be true,
3130 without considering any further alternative branches in the assertion.
3131 Backtracking into (*THEN) causes it to skip to the next enclosing alternative
3132 within the assertion (the normal behaviour), but if the assertion does not have
3133 such an alternative, (*THEN) behaves like (*PRUNE).
3134 .
3135 .
3136 .\" HTML <a name="btsub"></a>
3137 .SS "Backtracking verbs in subroutines"
3138 .rs
3139 .sp
3140 These behaviours occur whether or not the subpattern is called recursively.
3141 Perl's treatment of subroutines is different in some cases.
3142 .P
3143 (*FAIL) in a subpattern called as a subroutine has its normal effect: it forces
3144 an immediate backtrack.
3145 .P
3146 (*ACCEPT) in a subpattern called as a subroutine causes the subroutine match to
3147 succeed without any further processing. Matching then continues after the
3148 subroutine call.
3149 .P
3150 (*COMMIT), (*SKIP), and (*PRUNE) in a subpattern called as a subroutine cause
3151 the subroutine match to fail.
3152 .P
3153 (*THEN) skips to the next alternative in the innermost enclosing group within
3154 the subpattern that has alternatives. If there is no such group within the
3155 subpattern, (*THEN) causes the subroutine match to fail.
3156 .
3157 .
3158 .SH "SEE ALSO"
3159 .rs
3160 .sp
3161 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3),
3162 \fBpcresyntax\fP(3), \fBpcre\fP(3), \fBpcre16(3)\fP, \fBpcre32(3)\fP.
3163 .
3164 .
3166 .rs
3167 .sp
3168 .nf
3169 Philip Hazel
3170 University Computing Service
3171 Cambridge CB2 3QH, England.
3172 .fi
3173 .
3174 .
3176 .rs
3177 .sp
3178 .nf
3179 Last updated: 08 October 2013
3180 Copyright (c) 1997-2013 University of Cambridge.
3181 .fi


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

  ViewVC Help
Powered by ViewVC 1.1.5