ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 1369 - (show annotations)
Tue Oct 8 15:06:46 2013 UTC (6 years, 4 months ago) by ph10
File size: 131181 byte(s)
Error occurred while calculating annotation data.
Update \8 and \9 handling to match most recent Perl.
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 \eddd character with octal code ddd, or back reference
304 \exhh character with hex code hh
305 \ex{hhh..} character with hex code hhh.. (non-JavaScript mode)
306 \euhhhh character with hex code hhhh (JavaScript mode only)
307 .sp
308 The precise effect of \ecx on ASCII characters is as follows: if x is a lower
309 case letter, it is converted to upper case. Then bit 6 of the character (hex
310 40) is inverted. Thus \ecA to \ecZ become hex 01 to hex 1A (A is 41, Z is 5A),
311 but \ec{ becomes hex 3B ({ is 7B), and \ec; becomes hex 7B (; is 3B). If the
312 data item (byte or 16-bit value) following \ec has a value greater than 127, a
313 compile-time error occurs. This locks out non-ASCII characters in all modes.
314 .P
315 The \ec facility was designed for use with ASCII characters, but with the
316 extension to Unicode it is even less useful than it once was. It is, however,
317 recognized when PCRE is compiled in EBCDIC mode, where data items are always
318 bytes. In this mode, all values are valid after \ec. If the next character is a
319 lower case letter, it is converted to upper case. Then the 0xc0 bits of the
320 byte are inverted. Thus \ecA becomes hex 01, as in ASCII (A is C1), but because
321 the EBCDIC letters are disjoint, \ecZ becomes hex 29 (Z is E9), and other
322 characters also generate different values.
323 .P
324 By default, after \ex, from zero to two hexadecimal digits are read (letters
325 can be in upper or lower case). Any number of hexadecimal digits may appear
326 between \ex{ and }, but the character code is constrained as follows:
327 .sp
328 8-bit non-UTF mode less than 0x100
329 8-bit UTF-8 mode less than 0x10ffff and a valid codepoint
330 16-bit non-UTF mode less than 0x10000
331 16-bit UTF-16 mode less than 0x10ffff and a valid codepoint
332 32-bit non-UTF mode less than 0x80000000
333 32-bit UTF-32 mode less than 0x10ffff and a valid codepoint
334 .sp
335 Invalid Unicode codepoints are the range 0xd800 to 0xdfff (the so-called
336 "surrogate" codepoints), and 0xffef.
337 .P
338 If characters other than hexadecimal digits appear between \ex{ and }, or if
339 there is no terminating }, this form of escape is not recognized. Instead, the
340 initial \ex will be interpreted as a basic hexadecimal escape, with no
341 following digits, giving a character whose value is zero.
342 .P
343 If the PCRE_JAVASCRIPT_COMPAT option is set, the interpretation of \ex is
344 as just described only when it is followed by two hexadecimal digits.
345 Otherwise, it matches a literal "x" character. In JavaScript mode, support for
346 code points greater than 256 is provided by \eu, which must be followed by
347 four hexadecimal digits; otherwise it matches a literal "u" character.
348 Character codes specified by \eu in JavaScript mode are constrained in the same
349 was as those specified by \ex in non-JavaScript mode.
350 .P
351 Characters whose value is less than 256 can be defined by either of the two
352 syntaxes for \ex (or by \eu in JavaScript mode). There is no difference in the
353 way they are handled. For example, \exdc is exactly the same as \ex{dc} (or
354 \eu00dc in JavaScript mode).
355 .P
356 After \e0 up to two further octal digits are read. If there are fewer than two
357 digits, just those that are present are used. Thus the sequence \e0\ex\e07
358 specifies two binary zeros followed by a BEL character (code value 7). Make
359 sure you supply two digits after the initial zero if the pattern character that
360 follows is itself an octal digit.
361 .P
362 The handling of a backslash followed by a digit other than 0 is complicated,
363 and Perl has changed in recent releases, causing PCRE also to change. Outside a
364 character class, PCRE reads the digit and any following digits as a decimal
365 number. If the number is less than 8, or if there have been at least that many
366 previous capturing left parentheses in the expression, the entire sequence is
367 taken as a \fIback reference\fP. A description of how this works is given
368 .\" HTML <a href="#backreferences">
369 .\" </a>
370 later,
371 .\"
372 following the discussion of
373 .\" HTML <a href="#subpattern">
374 .\" </a>
375 parenthesized subpatterns.
376 .\"
377 .P
378 Inside a character class, or if the decimal number following \e is greater than
379 7 and there have not been that many capturing subpatterns, PCRE handles \e8 and
380 \e9 as the literal characters "8" and "9", and otherwise re-reads up to three
381 octal digits following the backslash, using them to generate a data character.
382 Any subsequent digits stand for themselves. The value of the character is
383 constrained in the same way as characters specified in hexadecimal. For
384 example:
385 .sp
386 \e040 is another way of writing an ASCII space
387 .\" JOIN
388 \e40 is the same, provided there are fewer than 40
389 previous capturing subpatterns
390 \e7 is always a back reference
391 .\" JOIN
392 \e11 might be a back reference, or another way of
393 writing a tab
394 \e011 is always a tab
395 \e0113 is a tab followed by the character "3"
396 .\" JOIN
397 \e113 might be a back reference, otherwise the
398 character with octal code 113
399 .\" JOIN
400 \e377 might be a back reference, otherwise
401 the value 255 (decimal)
402 .\" JOIN
403 \e81 is either a back reference, or the two
404 characters "8" and "1"
405 .sp
406 Note that octal values of 100 or greater must not be introduced by a leading
407 zero, because no more than three octal digits are ever read.
408 .P
409 All the sequences that define a single character value can be used both inside
410 and outside character classes. In addition, inside a character class, \eb is
411 interpreted as the backspace character (hex 08).
412 .P
413 \eN is not allowed in a character class. \eB, \eR, and \eX are not special
414 inside a character class. Like other unrecognized escape sequences, they are
415 treated as the literal characters "B", "R", and "X" by default, but cause an
416 error if the PCRE_EXTRA option is set. Outside a character class, these
417 sequences have different meanings.
418 .
419 .
420 .SS "Unsupported escape sequences"
421 .rs
422 .sp
423 In Perl, the sequences \el, \eL, \eu, and \eU are recognized by its string
424 handler and used to modify the case of following characters. By default, PCRE
425 does not support these escape sequences. However, if the PCRE_JAVASCRIPT_COMPAT
426 option is set, \eU matches a "U" character, and \eu can be used to define a
427 character by code point, as described in the previous section.
428 .
429 .
430 .SS "Absolute and relative back references"
431 .rs
432 .sp
433 The sequence \eg followed by an unsigned or a negative number, optionally
434 enclosed in braces, is an absolute or relative back reference. A named back
435 reference can be coded as \eg{name}. Back references are discussed
436 .\" HTML <a href="#backreferences">
437 .\" </a>
438 later,
439 .\"
440 following the discussion of
441 .\" HTML <a href="#subpattern">
442 .\" </a>
443 parenthesized subpatterns.
444 .\"
445 .
446 .
447 .SS "Absolute and relative subroutine calls"
448 .rs
449 .sp
450 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
451 a number enclosed either in angle brackets or single quotes, is an alternative
452 syntax for referencing a subpattern as a "subroutine". Details are discussed
453 .\" HTML <a href="#onigurumasubroutines">
454 .\" </a>
455 later.
456 .\"
457 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
458 synonymous. The former is a back reference; the latter is a
459 .\" HTML <a href="#subpatternsassubroutines">
460 .\" </a>
461 subroutine
462 .\"
463 call.
464 .
465 .
466 .\" HTML <a name="genericchartypes"></a>
467 .SS "Generic character types"
468 .rs
469 .sp
470 Another use of backslash is for specifying generic character types:
471 .sp
472 \ed any decimal digit
473 \eD any character that is not a decimal digit
474 \eh any horizontal white space character
475 \eH any character that is not a horizontal white space character
476 \es any white space character
477 \eS any character that is not a white space character
478 \ev any vertical white space character
479 \eV any character that is not a vertical white space character
480 \ew any "word" character
481 \eW any "non-word" character
482 .sp
483 There is also the single sequence \eN, which matches a non-newline character.
484 This is the same as
485 .\" HTML <a href="#fullstopdot">
486 .\" </a>
487 the "." metacharacter
488 .\"
489 when PCRE_DOTALL is not set. Perl also uses \eN to match characters by name;
490 PCRE does not support this.
491 .P
492 Each pair of lower and upper case escape sequences partitions the complete set
493 of characters into two disjoint sets. Any given character matches one, and only
494 one, of each pair. The sequences can appear both inside and outside character
495 classes. They each match one character of the appropriate type. If the current
496 matching point is at the end of the subject string, all of them fail, because
497 there is no character to match.
498 .P
499 For compatibility with Perl, \es did not used to match the VT character (code
500 11), which made it different from the the POSIX "space" class. However, Perl
501 added VT at release 5.18, and PCRE followed suit at release 8.34. The \es
502 characters are now HT (9), LF (10), VT (11), FF (12), CR (13), and space (32).
503 .P
504 A "word" character is an underscore or any character that is a letter or digit.
505 By default, the definition of letters and digits is controlled by PCRE's
506 low-valued character tables, and may vary if locale-specific matching is taking
507 place (see
508 .\" HTML <a href="pcreapi.html#localesupport">
509 .\" </a>
510 "Locale support"
511 .\"
512 in the
513 .\" HREF
514 \fBpcreapi\fP
515 .\"
516 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
517 or "french" in Windows, some character codes greater than 128 are used for
518 accented letters, and these are then matched by \ew. The use of locales with
519 Unicode is discouraged.
520 .P
521 By default, in a UTF mode, characters with values greater than 128 never match
522 \ed, \es, or \ew, and always match \eD, \eS, and \eW. These sequences retain
523 their original meanings from before UTF support was available, mainly for
524 efficiency reasons. However, if PCRE is compiled with Unicode property support,
525 and the PCRE_UCP option is set, the behaviour is changed so that Unicode
526 properties are used to determine character types, as follows:
527 .sp
528 \ed any character that \ep{Nd} matches (decimal digit)
529 \es any character that \ep{Z} matches, plus HT, LF, FF, CR
530 \ew any character that \ep{L} or \ep{N} matches, plus underscore
531 .sp
532 The upper case escapes match the inverse sets of characters. Note that \ed
533 matches only decimal digits, whereas \ew matches any Unicode digit, as well as
534 any Unicode letter, and underscore. Note also that PCRE_UCP affects \eb, and
535 \eB because they are defined in terms of \ew and \eW. Matching these sequences
536 is noticeably slower when PCRE_UCP is set.
537 .P
538 The sequences \eh, \eH, \ev, and \eV are features that were added to Perl at
539 release 5.10. In contrast to the other sequences, which match only ASCII
540 characters by default, these always match certain high-valued codepoints,
541 whether or not PCRE_UCP is set. The horizontal space characters are:
542 .sp
543 U+0009 Horizontal tab (HT)
544 U+0020 Space
545 U+00A0 Non-break space
546 U+1680 Ogham space mark
547 U+180E Mongolian vowel separator
548 U+2000 En quad
549 U+2001 Em quad
550 U+2002 En space
551 U+2003 Em space
552 U+2004 Three-per-em space
553 U+2005 Four-per-em space
554 U+2006 Six-per-em space
555 U+2007 Figure space
556 U+2008 Punctuation space
557 U+2009 Thin space
558 U+200A Hair space
559 U+202F Narrow no-break space
560 U+205F Medium mathematical space
561 U+3000 Ideographic space
562 .sp
563 The vertical space characters are:
564 .sp
565 U+000A Linefeed (LF)
566 U+000B Vertical tab (VT)
567 U+000C Form feed (FF)
568 U+000D Carriage return (CR)
569 U+0085 Next line (NEL)
570 U+2028 Line separator
571 U+2029 Paragraph separator
572 .sp
573 In 8-bit, non-UTF-8 mode, only the characters with codepoints less than 256 are
574 relevant.
575 .
576 .
577 .\" HTML <a name="newlineseq"></a>
578 .SS "Newline sequences"
579 .rs
580 .sp
581 Outside a character class, by default, the escape sequence \eR matches any
582 Unicode newline sequence. In 8-bit non-UTF-8 mode \eR is equivalent to the
583 following:
584 .sp
585 (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
586 .sp
587 This is an example of an "atomic group", details of which are given
588 .\" HTML <a href="#atomicgroup">
589 .\" </a>
590 below.
591 .\"
592 This particular group matches either the two-character sequence CR followed by
593 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
594 U+000B), FF (form feed, U+000C), CR (carriage return, U+000D), or NEL (next
595 line, U+0085). The two-character sequence is treated as a single unit that
596 cannot be split.
597 .P
598 In other modes, two additional characters whose codepoints are greater than 255
599 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
600 Unicode character property support is not needed for these characters to be
601 recognized.
602 .P
603 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
604 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
605 either at compile time or when the pattern is matched. (BSR is an abbrevation
606 for "backslash R".) This can be made the default when PCRE is built; if this is
607 the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
608 It is also possible to specify these settings by starting a pattern string with
609 one of the following sequences:
610 .sp
611 (*BSR_ANYCRLF) CR, LF, or CRLF only
612 (*BSR_UNICODE) any Unicode newline sequence
613 .sp
614 These override the default and the options given to the compiling function, but
615 they can themselves be overridden by options given to a matching function. Note
616 that these special settings, which are not Perl-compatible, are recognized only
617 at the very start of a pattern, and that they must be in upper case. If more
618 than one of them is present, the last one is used. They can be combined with a
619 change of newline convention; for example, a pattern can start with:
620 .sp
622 .sp
623 They can also be combined with the (*UTF8), (*UTF16), (*UTF32), (*UTF) or
624 (*UCP) special sequences. Inside a character class, \eR is treated as an
625 unrecognized escape sequence, and so matches the letter "R" by default, but
626 causes an error if PCRE_EXTRA is set.
627 .
628 .
629 .\" HTML <a name="uniextseq"></a>
630 .SS Unicode character properties
631 .rs
632 .sp
633 When PCRE is built with Unicode character property support, three additional
634 escape sequences that match characters with specific properties are available.
635 When in 8-bit non-UTF-8 mode, these sequences are of course limited to testing
636 characters whose codepoints are less than 256, but they do work in this mode.
637 The extra escape sequences are:
638 .sp
639 \ep{\fIxx\fP} a character with the \fIxx\fP property
640 \eP{\fIxx\fP} a character without the \fIxx\fP property
641 \eX a Unicode extended grapheme cluster
642 .sp
643 The property names represented by \fIxx\fP above are limited to the Unicode
644 script names, the general category properties, "Any", which matches any
645 character (including newline), and some special PCRE properties (described
646 in the
647 .\" HTML <a href="#extraprops">
648 .\" </a>
649 next section).
650 .\"
651 Other Perl properties such as "InMusicalSymbols" are not currently supported by
652 PCRE. Note that \eP{Any} does not match any characters, so always causes a
653 match failure.
654 .P
655 Sets of Unicode characters are defined as belonging to certain scripts. A
656 character from one of these sets can be matched using a script name. For
657 example:
658 .sp
659 \ep{Greek}
660 \eP{Han}
661 .sp
662 Those that are not part of an identified script are lumped together as
663 "Common". The current list of scripts is:
664 .P
665 Arabic,
666 Armenian,
667 Avestan,
668 Balinese,
669 Bamum,
670 Batak,
671 Bengali,
672 Bopomofo,
673 Brahmi,
674 Braille,
675 Buginese,
676 Buhid,
677 Canadian_Aboriginal,
678 Carian,
679 Chakma,
680 Cham,
681 Cherokee,
682 Common,
683 Coptic,
684 Cuneiform,
685 Cypriot,
686 Cyrillic,
687 Deseret,
688 Devanagari,
689 Egyptian_Hieroglyphs,
690 Ethiopic,
691 Georgian,
692 Glagolitic,
693 Gothic,
694 Greek,
695 Gujarati,
696 Gurmukhi,
697 Han,
698 Hangul,
699 Hanunoo,
700 Hebrew,
701 Hiragana,
702 Imperial_Aramaic,
703 Inherited,
704 Inscriptional_Pahlavi,
705 Inscriptional_Parthian,
706 Javanese,
707 Kaithi,
708 Kannada,
709 Katakana,
710 Kayah_Li,
711 Kharoshthi,
712 Khmer,
713 Lao,
714 Latin,
715 Lepcha,
716 Limbu,
717 Linear_B,
718 Lisu,
719 Lycian,
720 Lydian,
721 Malayalam,
722 Mandaic,
723 Meetei_Mayek,
724 Meroitic_Cursive,
725 Meroitic_Hieroglyphs,
726 Miao,
727 Mongolian,
728 Myanmar,
729 New_Tai_Lue,
730 Nko,
731 Ogham,
732 Old_Italic,
733 Old_Persian,
734 Old_South_Arabian,
735 Old_Turkic,
736 Ol_Chiki,
737 Oriya,
738 Osmanya,
739 Phags_Pa,
740 Phoenician,
741 Rejang,
742 Runic,
743 Samaritan,
744 Saurashtra,
745 Sharada,
746 Shavian,
747 Sinhala,
748 Sora_Sompeng,
749 Sundanese,
750 Syloti_Nagri,
751 Syriac,
752 Tagalog,
753 Tagbanwa,
754 Tai_Le,
755 Tai_Tham,
756 Tai_Viet,
757 Takri,
758 Tamil,
759 Telugu,
760 Thaana,
761 Thai,
762 Tibetan,
763 Tifinagh,
764 Ugaritic,
765 Vai,
766 Yi.
767 .P
768 Each character has exactly one Unicode general category property, specified by
769 a two-letter abbreviation. For compatibility with Perl, negation can be
770 specified by including a circumflex between the opening brace and the property
771 name. For example, \ep{^Lu} is the same as \eP{Lu}.
772 .P
773 If only one letter is specified with \ep or \eP, it includes all the general
774 category properties that start with that letter. In this case, in the absence
775 of negation, the curly brackets in the escape sequence are optional; these two
776 examples have the same effect:
777 .sp
778 \ep{L}
779 \epL
780 .sp
781 The following general category property codes are supported:
782 .sp
783 C Other
784 Cc Control
785 Cf Format
786 Cn Unassigned
787 Co Private use
788 Cs Surrogate
789 .sp
790 L Letter
791 Ll Lower case letter
792 Lm Modifier letter
793 Lo Other letter
794 Lt Title case letter
795 Lu Upper case letter
796 .sp
797 M Mark
798 Mc Spacing mark
799 Me Enclosing mark
800 Mn Non-spacing mark
801 .sp
802 N Number
803 Nd Decimal number
804 Nl Letter number
805 No Other number
806 .sp
807 P Punctuation
808 Pc Connector punctuation
809 Pd Dash punctuation
810 Pe Close punctuation
811 Pf Final punctuation
812 Pi Initial punctuation
813 Po Other punctuation
814 Ps Open punctuation
815 .sp
816 S Symbol
817 Sc Currency symbol
818 Sk Modifier symbol
819 Sm Mathematical symbol
820 So Other symbol
821 .sp
822 Z Separator
823 Zl Line separator
824 Zp Paragraph separator
825 Zs Space separator
826 .sp
827 The special property L& is also supported: it matches a character that has
828 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
829 a modifier or "other".
830 .P
831 The Cs (Surrogate) property applies only to characters in the range U+D800 to
832 U+DFFF. Such characters are not valid in Unicode strings and so
833 cannot be tested by PCRE, unless UTF validity checking has been turned off
834 (see the discussion of PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK and
835 PCRE_NO_UTF32_CHECK in the
836 .\" HREF
837 \fBpcreapi\fP
838 .\"
839 page). Perl does not support the Cs property.
840 .P
841 The long synonyms for property names that Perl supports (such as \ep{Letter})
842 are not supported by PCRE, nor is it permitted to prefix any of these
843 properties with "Is".
844 .P
845 No character that is in the Unicode table has the Cn (unassigned) property.
846 Instead, this property is assumed for any code point that is not in the
847 Unicode table.
848 .P
849 Specifying caseless matching does not affect these escape sequences. For
850 example, \ep{Lu} always matches only upper case letters. This is different from
851 the behaviour of current versions of Perl.
852 .P
853 Matching characters by Unicode property is not fast, because PCRE has to do a
854 multistage table lookup in order to find a character's property. That is why
855 the traditional escape sequences such as \ed and \ew do not use Unicode
856 properties in PCRE by default, though you can make them do so by setting the
857 PCRE_UCP option or by starting the pattern with (*UCP).
858 .
859 .
860 .SS Extended grapheme clusters
861 .rs
862 .sp
863 The \eX escape matches any number of Unicode characters that form an "extended
864 grapheme cluster", and treats the sequence as an atomic group
865 .\" HTML <a href="#atomicgroup">
866 .\" </a>
867 (see below).
868 .\"
869 Up to and including release 8.31, PCRE matched an earlier, simpler definition
870 that was equivalent to
871 .sp
872 (?>\ePM\epM*)
873 .sp
874 That is, it matched a character without the "mark" property, followed by zero
875 or more characters with the "mark" property. Characters with the "mark"
876 property are typically non-spacing accents that affect the preceding character.
877 .P
878 This simple definition was extended in Unicode to include more complicated
879 kinds of composite character by giving each character a grapheme breaking
880 property, and creating rules that use these properties to define the boundaries
881 of extended grapheme clusters. In releases of PCRE later than 8.31, \eX matches
882 one of these clusters.
883 .P
884 \eX always matches at least one character. Then it decides whether to add
885 additional characters according to the following rules for ending a cluster:
886 .P
887 1. End at the end of the subject string.
888 .P
889 2. Do not end between CR and LF; otherwise end after any control character.
890 .P
891 3. Do not break Hangul (a Korean script) syllable sequences. Hangul characters
892 are of five types: L, V, T, LV, and LVT. An L character may be followed by an
893 L, V, LV, or LVT character; an LV or V character may be followed by a V or T
894 character; an LVT or T character may be follwed only by a T character.
895 .P
896 4. Do not end before extending characters or spacing marks. Characters with
897 the "mark" property always have the "extend" grapheme breaking property.
898 .P
899 5. Do not end after prepend characters.
900 .P
901 6. Otherwise, end the cluster.
902 .
903 .
904 .\" HTML <a name="extraprops"></a>
905 .SS PCRE's additional properties
906 .rs
907 .sp
908 As well as the standard Unicode properties described above, PCRE supports four
909 more that make it possible to convert traditional escape sequences such as \ew
910 and \es and POSIX character classes to use Unicode properties. PCRE uses these
911 non-standard, non-Perl properties internally when PCRE_UCP is set. However,
912 they may also be used explicitly. These properties are:
913 .sp
914 Xan Any alphanumeric character
915 Xps Any POSIX space character
916 Xsp Any Perl space character
917 Xwd Any Perl "word" character
918 .sp
919 Xan matches characters that have either the L (letter) or the N (number)
920 property. Xps matches the characters tab, linefeed, vertical tab, form feed, or
921 carriage return, and any other character that has the Z (separator) property.
922 Xsp is the same as Xps, except that vertical tab is excluded. Xwd matches the
923 same characters as Xan, plus underscore.
924 .P
925 There is another non-standard property, Xuc, which matches any character that
926 can be represented by a Universal Character Name in C++ and other programming
927 languages. These are the characters $, @, ` (grave accent), and all characters
928 with Unicode code points greater than or equal to U+00A0, except for the
929 surrogates U+D800 to U+DFFF. Note that most base (ASCII) characters are
930 excluded. (Universal Character Names are of the form \euHHHH or \eUHHHHHHHH
931 where H is a hexadecimal digit. Note that the Xuc property does not match these
932 sequences but the characters that they represent.)
933 .
934 .
935 .\" HTML <a name="resetmatchstart"></a>
936 .SS "Resetting the match start"
937 .rs
938 .sp
939 The escape sequence \eK causes any previously matched characters not to be
940 included in the final matched sequence. For example, the pattern:
941 .sp
942 foo\eKbar
943 .sp
944 matches "foobar", but reports that it has matched "bar". This feature is
945 similar to a lookbehind assertion
946 .\" HTML <a href="#lookbehind">
947 .\" </a>
948 (described below).
949 .\"
950 However, in this case, the part of the subject before the real match does not
951 have to be of fixed length, as lookbehind assertions do. The use of \eK does
952 not interfere with the setting of
953 .\" HTML <a href="#subpattern">
954 .\" </a>
955 captured substrings.
956 .\"
957 For example, when the pattern
958 .sp
959 (foo)\eKbar
960 .sp
961 matches "foobar", the first substring is still set to "foo".
962 .P
963 Perl documents that the use of \eK within assertions is "not well defined". In
964 PCRE, \eK is acted upon when it occurs inside positive assertions, but is
965 ignored in negative assertions.
966 .
967 .
968 .\" HTML <a name="smallassertions"></a>
969 .SS "Simple assertions"
970 .rs
971 .sp
972 The final use of backslash is for certain simple assertions. An assertion
973 specifies a condition that has to be met at a particular point in a match,
974 without consuming any characters from the subject string. The use of
975 subpatterns for more complicated assertions is described
976 .\" HTML <a href="#bigassertions">
977 .\" </a>
978 below.
979 .\"
980 The backslashed assertions are:
981 .sp
982 \eb matches at a word boundary
983 \eB matches when not at a word boundary
984 \eA matches at the start of the subject
985 \eZ matches at the end of the subject
986 also matches before a newline at the end of the subject
987 \ez matches only at the end of the subject
988 \eG matches at the first matching position in the subject
989 .sp
990 Inside a character class, \eb has a different meaning; it matches the backspace
991 character. If any other of these assertions appears in a character class, by
992 default it matches the corresponding literal character (for example, \eB
993 matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
994 escape sequence" error is generated instead.
995 .P
996 A word boundary is a position in the subject string where the current character
997 and the previous character do not both match \ew or \eW (i.e. one matches
998 \ew and the other matches \eW), or the start or end of the string if the
999 first or last character matches \ew, respectively. In a UTF mode, the meanings
1000 of \ew and \eW can be changed by setting the PCRE_UCP option. When this is
1001 done, it also affects \eb and \eB. Neither PCRE nor Perl has a separate "start
1002 of word" or "end of word" metasequence. However, whatever follows \eb normally
1003 determines which it is. For example, the fragment \eba matches "a" at the start
1004 of a word.
1005 .P
1006 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
1007 dollar (described in the next section) in that they only ever match at the very
1008 start and end of the subject string, whatever options are set. Thus, they are
1009 independent of multiline mode. These three assertions are not affected by the
1010 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
1011 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
1012 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
1013 at a point other than the beginning of the subject, \eA can never match. The
1014 difference between \eZ and \ez is that \eZ matches before a newline at the end
1015 of the string as well as at the very end, whereas \ez matches only at the end.
1016 .P
1017 The \eG assertion is true only when the current matching position is at the
1018 start point of the match, as specified by the \fIstartoffset\fP argument of
1019 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
1020 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
1021 arguments, you can mimic Perl's /g option, and it is in this kind of
1022 implementation where \eG can be useful.
1023 .P
1024 Note, however, that PCRE's interpretation of \eG, as the start of the current
1025 match, is subtly different from Perl's, which defines it as the end of the
1026 previous match. In Perl, these can be different when the previously matched
1027 string was empty. Because PCRE does just one match at a time, it cannot
1028 reproduce this behaviour.
1029 .P
1030 If all the alternatives of a pattern begin with \eG, the expression is anchored
1031 to the starting match position, and the "anchored" flag is set in the compiled
1032 regular expression.
1033 .
1034 .
1036 .rs
1037 .sp
1038 The circumflex and dollar metacharacters are zero-width assertions. That is,
1039 they test for a particular condition being true without consuming any
1040 characters from the subject string.
1041 .P
1042 Outside a character class, in the default matching mode, the circumflex
1043 character is an assertion that is true only if the current matching point is at
1044 the start of the subject string. If the \fIstartoffset\fP argument of
1045 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
1046 option is unset. Inside a character class, circumflex has an entirely different
1047 meaning
1048 .\" HTML <a href="#characterclass">
1049 .\" </a>
1050 (see below).
1051 .\"
1052 .P
1053 Circumflex need not be the first character of the pattern if a number of
1054 alternatives are involved, but it should be the first thing in each alternative
1055 in which it appears if the pattern is ever to match that branch. If all
1056 possible alternatives start with a circumflex, that is, if the pattern is
1057 constrained to match only at the start of the subject, it is said to be an
1058 "anchored" pattern. (There are also other constructs that can cause a pattern
1059 to be anchored.)
1060 .P
1061 The dollar character is an assertion that is true only if the current matching
1062 point is at the end of the subject string, or immediately before a newline at
1063 the end of the string (by default). Note, however, that it does not actually
1064 match the newline. Dollar need not be the last character of the pattern if a
1065 number of alternatives are involved, but it should be the last item in any
1066 branch in which it appears. Dollar has no special meaning in a character class.
1067 .P
1068 The meaning of dollar can be changed so that it matches only at the very end of
1069 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
1070 does not affect the \eZ assertion.
1071 .P
1072 The meanings of the circumflex and dollar characters are changed if the
1073 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
1074 immediately after internal newlines as well as at the start of the subject
1075 string. It does not match after a newline that ends the string. A dollar
1076 matches before any newlines in the string, as well as at the very end, when
1077 PCRE_MULTILINE is set. When newline is specified as the two-character
1078 sequence CRLF, isolated CR and LF characters do not indicate newlines.
1079 .P
1080 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
1081 \en represents a newline) in multiline mode, but not otherwise. Consequently,
1082 patterns that are anchored in single line mode because all branches start with
1083 ^ are not anchored in multiline mode, and a match for circumflex is possible
1084 when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
1085 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
1086 .P
1087 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
1088 end of the subject in both modes, and if all branches of a pattern start with
1089 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
1090 .
1091 .
1092 .\" HTML <a name="fullstopdot"></a>
1094 .rs
1095 .sp
1096 Outside a character class, a dot in the pattern matches any one character in
1097 the subject string except (by default) a character that signifies the end of a
1098 line.
1099 .P
1100 When a line ending is defined as a single character, dot never matches that
1101 character; when the two-character sequence CRLF is used, dot does not match CR
1102 if it is immediately followed by LF, but otherwise it matches all characters
1103 (including isolated CRs and LFs). When any Unicode line endings are being
1104 recognized, dot does not match CR or LF or any of the other line ending
1105 characters.
1106 .P
1107 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
1108 option is set, a dot matches any one character, without exception. If the
1109 two-character sequence CRLF is present in the subject string, it takes two dots
1110 to match it.
1111 .P
1112 The handling of dot is entirely independent of the handling of circumflex and
1113 dollar, the only relationship being that they both involve newlines. Dot has no
1114 special meaning in a character class.
1115 .P
1116 The escape sequence \eN behaves like a dot, except that it is not affected by
1117 the PCRE_DOTALL option. In other words, it matches any character except one
1118 that signifies the end of a line. Perl also uses \eN to match characters by
1119 name; PCRE does not support this.
1120 .
1121 .
1123 .rs
1124 .sp
1125 Outside a character class, the escape sequence \eC matches any one data unit,
1126 whether or not a UTF mode is set. In the 8-bit library, one data unit is one
1127 byte; in the 16-bit library it is a 16-bit unit; in the 32-bit library it is
1128 a 32-bit unit. Unlike a dot, \eC always
1129 matches line-ending characters. The feature is provided in Perl in order to
1130 match individual bytes in UTF-8 mode, but it is unclear how it can usefully be
1131 used. Because \eC breaks up characters into individual data units, matching one
1132 unit with \eC in a UTF mode means that the rest of the string may start with a
1133 malformed UTF character. This has undefined results, because PCRE assumes that
1134 it is dealing with valid UTF strings (and by default it checks this at the
1135 start of processing unless the PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK or
1136 PCRE_NO_UTF32_CHECK option is used).
1137 .P
1138 PCRE does not allow \eC to appear in lookbehind assertions
1139 .\" HTML <a href="#lookbehind">
1140 .\" </a>
1141 (described below)
1142 .\"
1143 in a UTF mode, because this would make it impossible to calculate the length of
1144 the lookbehind.
1145 .P
1146 In general, the \eC escape sequence is best avoided. However, one
1147 way of using it that avoids the problem of malformed UTF characters is to use a
1148 lookahead to check the length of the next character, as in this pattern, which
1149 could be used with a UTF-8 string (ignore white space and line breaks):
1150 .sp
1151 (?| (?=[\ex00-\ex7f])(\eC) |
1152 (?=[\ex80-\ex{7ff}])(\eC)(\eC) |
1153 (?=[\ex{800}-\ex{ffff}])(\eC)(\eC)(\eC) |
1154 (?=[\ex{10000}-\ex{1fffff}])(\eC)(\eC)(\eC)(\eC))
1155 .sp
1156 A group that starts with (?| resets the capturing parentheses numbers in each
1157 alternative (see
1158 .\" HTML <a href="#dupsubpatternnumber">
1159 .\" </a>
1160 "Duplicate Subpattern Numbers"
1161 .\"
1162 below). The assertions at the start of each branch check the next UTF-8
1163 character for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The
1164 character's individual bytes are then captured by the appropriate number of
1165 groups.
1166 .
1167 .
1168 .\" HTML <a name="characterclass"></a>
1170 .rs
1171 .sp
1172 An opening square bracket introduces a character class, terminated by a closing
1173 square bracket. A closing square bracket on its own is not special by default.
1174 However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
1175 bracket causes a compile-time error. If a closing square bracket is required as
1176 a member of the class, it should be the first data character in the class
1177 (after an initial circumflex, if present) or escaped with a backslash.
1178 .P
1179 A character class matches a single character in the subject. In a UTF mode, the
1180 character may be more than one data unit long. A matched character must be in
1181 the set of characters defined by the class, unless the first character in the
1182 class definition is a circumflex, in which case the subject character must not
1183 be in the set defined by the class. If a circumflex is actually required as a
1184 member of the class, ensure it is not the first character, or escape it with a
1185 backslash.
1186 .P
1187 For example, the character class [aeiou] matches any lower case vowel, while
1188 [^aeiou] matches any character that is not a lower case vowel. Note that a
1189 circumflex is just a convenient notation for specifying the characters that
1190 are in the class by enumerating those that are not. A class that starts with a
1191 circumflex is not an assertion; it still consumes a character from the subject
1192 string, and therefore it fails if the current pointer is at the end of the
1193 string.
1194 .P
1195 In UTF-8 (UTF-16, UTF-32) mode, characters with values greater than 255 (0xffff)
1196 can be included in a class as a literal string of data units, or by using the
1197 \ex{ escaping mechanism.
1198 .P
1199 When caseless matching is set, any letters in a class represent both their
1200 upper case and lower case versions, so for example, a caseless [aeiou] matches
1201 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1202 caseful version would. In a UTF mode, PCRE always understands the concept of
1203 case for characters whose values are less than 128, so caseless matching is
1204 always possible. For characters with higher values, the concept of case is
1205 supported if PCRE is compiled with Unicode property support, but not otherwise.
1206 If you want to use caseless matching in a UTF mode for characters 128 and
1207 above, you must ensure that PCRE is compiled with Unicode property support as
1208 well as with UTF support.
1209 .P
1210 Characters that might indicate line breaks are never treated in any special way
1211 when matching character classes, whatever line-ending sequence is in use, and
1212 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
1213 such as [^a] always matches one of these characters.
1214 .P
1215 The minus (hyphen) character can be used to specify a range of characters in a
1216 character class. For example, [d-m] matches any letter between d and m,
1217 inclusive. If a minus character is required in a class, it must be escaped with
1218 a backslash or appear in a position where it cannot be interpreted as
1219 indicating a range, typically as the first or last character in the class.
1220 .P
1221 It is not possible to have the literal character "]" as the end character of a
1222 range. A pattern such as [W-]46] is interpreted as a class of two characters
1223 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1224 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1225 the end of range, so [W-\e]46] is interpreted as a class containing a range
1226 followed by two other characters. The octal or hexadecimal representation of
1227 "]" can also be used to end a range.
1228 .P
1229 Ranges operate in the collating sequence of character values. They can also be
1230 used for characters specified numerically, for example [\e000-\e037]. Ranges
1231 can include any characters that are valid for the current mode.
1232 .P
1233 If a range that includes letters is used when caseless matching is set, it
1234 matches the letters in either case. For example, [W-c] is equivalent to
1235 [][\e\e^_`wxyzabc], matched caselessly, and in a non-UTF mode, if character
1236 tables for a French locale are in use, [\exc8-\excb] matches accented E
1237 characters in both cases. In UTF modes, PCRE supports the concept of case for
1238 characters with values greater than 128 only when it is compiled with Unicode
1239 property support.
1240 .P
1241 The character escape sequences \ed, \eD, \eh, \eH, \ep, \eP, \es, \eS, \ev,
1242 \eV, \ew, and \eW may appear in a character class, and add the characters that
1243 they match to the class. For example, [\edABCDEF] matches any hexadecimal
1244 digit. In UTF modes, the PCRE_UCP option affects the meanings of \ed, \es, \ew
1245 and their upper case partners, just as it does when they appear outside a
1246 character class, as described in the section entitled
1247 .\" HTML <a href="#genericchartypes">
1248 .\" </a>
1249 "Generic character types"
1250 .\"
1251 above. The escape sequence \eb has a different meaning inside a character
1252 class; it matches the backspace character. The sequences \eB, \eN, \eR, and \eX
1253 are not special inside a character class. Like any other unrecognized escape
1254 sequences, they are treated as the literal characters "B", "N", "R", and "X" by
1255 default, but cause an error if the PCRE_EXTRA option is set.
1256 .P
1257 A circumflex can conveniently be used with the upper case character types to
1258 specify a more restricted set of characters than the matching lower case type.
1259 For example, the class [^\eW_] matches any letter or digit, but not underscore,
1260 whereas [\ew] includes underscore. A positive character class should be read as
1261 "something OR something OR ..." and a negative class as "NOT something AND NOT
1262 something AND NOT ...".
1263 .P
1264 The only metacharacters that are recognized in character classes are backslash,
1265 hyphen (only where it can be interpreted as specifying a range), circumflex
1266 (only at the start), opening square bracket (only when it can be interpreted as
1267 introducing a POSIX class name - see the next section), and the terminating
1268 closing square bracket. However, escaping other non-alphanumeric characters
1269 does no harm.
1270 .
1271 .
1273 .rs
1274 .sp
1275 Perl supports the POSIX notation for character classes. This uses names
1276 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1277 this notation. For example,
1278 .sp
1279 [01[:alpha:]%]
1280 .sp
1281 matches "0", "1", any alphabetic character, or "%". The supported class names
1282 are:
1283 .sp
1284 alnum letters and digits
1285 alpha letters
1286 ascii character codes 0 - 127
1287 blank space or tab only
1288 cntrl control characters
1289 digit decimal digits (same as \ed)
1290 graph printing characters, excluding space
1291 lower lower case letters
1292 print printing characters, including space
1293 punct printing characters, excluding letters and digits and space
1294 space white space (not quite the same as \es)
1295 upper upper case letters
1296 word "word" characters (same as \ew)
1297 xdigit hexadecimal digits
1298 .sp
1299 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1300 space (32). "Space" used to be different to \es, which did not include VT, for
1301 Perl compatibility. However, Perl changed at release 5.18, and PCRE followed at
1302 release 8.34. "Space" and \es now match the same set of characters.
1303 .P
1304 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1305 5.8. Another Perl extension is negation, which is indicated by a ^ character
1306 after the colon. For example,
1307 .sp
1308 [12[:^digit:]]
1309 .sp
1310 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1311 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1312 supported, and an error is given if they are encountered.
1313 .P
1314 By default, in UTF modes, characters with values greater than 128 do not match
1315 any of the POSIX character classes. However, if the PCRE_UCP option is passed
1316 to \fBpcre_compile()\fP, some of the classes are changed so that Unicode
1317 character properties are used. This is achieved by replacing the POSIX classes
1318 by other sequences, as follows:
1319 .sp
1320 [:alnum:] becomes \ep{Xan}
1321 [:alpha:] becomes \ep{L}
1322 [:blank:] becomes \eh
1323 [:digit:] becomes \ep{Nd}
1324 [:lower:] becomes \ep{Ll}
1325 [:space:] becomes \ep{Xps}
1326 [:upper:] becomes \ep{Lu}
1327 [:word:] becomes \ep{Xwd}
1328 .sp
1329 Negated versions, such as [:^alpha:] use \eP instead of \ep. The other POSIX
1330 classes are unchanged, and match only characters with code points less than
1331 128.
1332 .
1333 .
1335 .rs
1336 .sp
1337 Vertical bar characters are used to separate alternative patterns. For example,
1338 the pattern
1339 .sp
1340 gilbert|sullivan
1341 .sp
1342 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1343 and an empty alternative is permitted (matching the empty string). The matching
1344 process tries each alternative in turn, from left to right, and the first one
1345 that succeeds is used. If the alternatives are within a subpattern
1346 .\" HTML <a href="#subpattern">
1347 .\" </a>
1348 (defined below),
1349 .\"
1350 "succeeds" means matching the rest of the main pattern as well as the
1351 alternative in the subpattern.
1352 .
1353 .
1355 .rs
1356 .sp
1357 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1358 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1359 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1360 The option letters are
1361 .sp
1362 i for PCRE_CASELESS
1364 s for PCRE_DOTALL
1365 x for PCRE_EXTENDED
1366 .sp
1367 For example, (?im) sets caseless, multiline matching. It is also possible to
1368 unset these options by preceding the letter with a hyphen, and a combined
1369 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1370 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1371 permitted. If a letter appears both before and after the hyphen, the option is
1372 unset.
1373 .P
1374 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1375 changed in the same way as the Perl-compatible options by using the characters
1376 J, U and X respectively.
1377 .P
1378 When one of these option changes occurs at top level (that is, not inside
1379 subpattern parentheses), the change applies to the remainder of the pattern
1380 that follows. If the change is placed right at the start of a pattern, PCRE
1381 extracts it into the global options (and it will therefore show up in data
1382 extracted by the \fBpcre_fullinfo()\fP function).
1383 .P
1384 An option change within a subpattern (see below for a description of
1385 subpatterns) affects only that part of the subpattern that follows it, so
1386 .sp
1387 (a(?i)b)c
1388 .sp
1389 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1390 By this means, options can be made to have different settings in different
1391 parts of the pattern. Any changes made in one alternative do carry on
1392 into subsequent branches within the same subpattern. For example,
1393 .sp
1394 (a(?i)b|c)
1395 .sp
1396 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1397 branch is abandoned before the option setting. This is because the effects of
1398 option settings happen at compile time. There would be some very weird
1399 behaviour otherwise.
1400 .P
1401 \fBNote:\fP There are other PCRE-specific options that can be set by the
1402 application when the compiling or matching functions are called. In some cases
1403 the pattern can contain special leading sequences such as (*CRLF) to override
1404 what the application has set or what has been defaulted. Details are given in
1405 the section entitled
1406 .\" HTML <a href="#newlineseq">
1407 .\" </a>
1408 "Newline sequences"
1409 .\"
1410 above. There are also the (*UTF8), (*UTF16),(*UTF32), and (*UCP) leading
1411 sequences that can be used to set UTF and Unicode property modes; they are
1412 equivalent to setting the PCRE_UTF8, PCRE_UTF16, PCRE_UTF32 and the PCRE_UCP
1413 options, respectively. The (*UTF) sequence is a generic version that can be
1414 used with any of the libraries. However, the application can set the
1415 PCRE_NEVER_UTF option, which locks out the use of the (*UTF) sequences.
1416 .
1417 .
1418 .\" HTML <a name="subpattern"></a>
1420 .rs
1421 .sp
1422 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1423 Turning part of a pattern into a subpattern does two things:
1424 .sp
1425 1. It localizes a set of alternatives. For example, the pattern
1426 .sp
1427 cat(aract|erpillar|)
1428 .sp
1429 matches "cataract", "caterpillar", or "cat". Without the parentheses, it would
1430 match "cataract", "erpillar" or an empty string.
1431 .sp
1432 2. It sets up the subpattern as a capturing subpattern. This means that, when
1433 the whole pattern matches, that portion of the subject string that matched the
1434 subpattern is passed back to the caller via the \fIovector\fP argument of the
1435 matching function. (This applies only to the traditional matching functions;
1436 the DFA matching functions do not support capturing.)
1437 .P
1438 Opening parentheses are counted from left to right (starting from 1) to obtain
1439 numbers for the capturing subpatterns. For example, if the string "the red
1440 king" is matched against the pattern
1441 .sp
1442 the ((red|white) (king|queen))
1443 .sp
1444 the captured substrings are "red king", "red", and "king", and are numbered 1,
1445 2, and 3, respectively.
1446 .P
1447 The fact that plain parentheses fulfil two functions is not always helpful.
1448 There are often times when a grouping subpattern is required without a
1449 capturing requirement. If an opening parenthesis is followed by a question mark
1450 and a colon, the subpattern does not do any capturing, and is not counted when
1451 computing the number of any subsequent capturing subpatterns. For example, if
1452 the string "the white queen" is matched against the pattern
1453 .sp
1454 the ((?:red|white) (king|queen))
1455 .sp
1456 the captured substrings are "white queen" and "queen", and are numbered 1 and
1457 2. The maximum number of capturing subpatterns is 65535.
1458 .P
1459 As a convenient shorthand, if any option settings are required at the start of
1460 a non-capturing subpattern, the option letters may appear between the "?" and
1461 the ":". Thus the two patterns
1462 .sp
1463 (?i:saturday|sunday)
1464 (?:(?i)saturday|sunday)
1465 .sp
1466 match exactly the same set of strings. Because alternative branches are tried
1467 from left to right, and options are not reset until the end of the subpattern
1468 is reached, an option setting in one branch does affect subsequent branches, so
1469 the above patterns match "SUNDAY" as well as "Saturday".
1470 .
1471 .
1472 .\" HTML <a name="dupsubpatternnumber"></a>
1474 .rs
1475 .sp
1476 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1477 the same numbers for its capturing parentheses. Such a subpattern starts with
1478 (?| and is itself a non-capturing subpattern. For example, consider this
1479 pattern:
1480 .sp
1481 (?|(Sat)ur|(Sun))day
1482 .sp
1483 Because the two alternatives are inside a (?| group, both sets of capturing
1484 parentheses are numbered one. Thus, when the pattern matches, you can look
1485 at captured substring number one, whichever alternative matched. This construct
1486 is useful when you want to capture part, but not all, of one of a number of
1487 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1488 number is reset at the start of each branch. The numbers of any capturing
1489 parentheses that follow the subpattern start after the highest number used in
1490 any branch. The following example is taken from the Perl documentation. The
1491 numbers underneath show in which buffer the captured content will be stored.
1492 .sp
1493 # before ---------------branch-reset----------- after
1494 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1495 # 1 2 2 3 2 3 4
1496 .sp
1497 A back reference to a numbered subpattern uses the most recent value that is
1498 set for that number by any subpattern. The following pattern matches "abcabc"
1499 or "defdef":
1500 .sp
1501 /(?|(abc)|(def))\e1/
1502 .sp
1503 In contrast, a subroutine call to a numbered subpattern always refers to the
1504 first one in the pattern with the given number. The following pattern matches
1505 "abcabc" or "defabc":
1506 .sp
1507 /(?|(abc)|(def))(?1)/
1508 .sp
1509 If a
1510 .\" HTML <a href="#conditions">
1511 .\" </a>
1512 condition test
1513 .\"
1514 for a subpattern's having matched refers to a non-unique number, the test is
1515 true if any of the subpatterns of that number have matched.
1516 .P
1517 An alternative approach to using this "branch reset" feature is to use
1518 duplicate named subpatterns, as described in the next section.
1519 .
1520 .
1522 .rs
1523 .sp
1524 Identifying capturing parentheses by number is simple, but it can be very hard
1525 to keep track of the numbers in complicated regular expressions. Furthermore,
1526 if an expression is modified, the numbers may change. To help with this
1527 difficulty, PCRE supports the naming of subpatterns. This feature was not
1528 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1529 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1530 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1531 have different names, but PCRE does not.
1532 .P
1533 In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1534 (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1535 parentheses from other parts of the pattern, such as
1536 .\" HTML <a href="#backreferences">
1537 .\" </a>
1538 back references,
1539 .\"
1540 .\" HTML <a href="#recursion">
1541 .\" </a>
1542 recursion,
1543 .\"
1544 and
1545 .\" HTML <a href="#conditions">
1546 .\" </a>
1547 conditions,
1548 .\"
1549 can be made by name as well as by number.
1550 .P
1551 Names consist of up to 32 alphanumeric characters and underscores. Named
1552 capturing parentheses are still allocated numbers as well as names, exactly as
1553 if the names were not present. The PCRE API provides function calls for
1554 extracting the name-to-number translation table from a compiled pattern. There
1555 is also a convenience function for extracting a captured substring by name.
1556 .P
1557 By default, a name must be unique within a pattern, but it is possible to relax
1558 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1559 names are also always permitted for subpatterns with the same number, set up as
1560 described in the previous section.) Duplicate names can be useful for patterns
1561 where only one instance of the named parentheses can match. Suppose you want to
1562 match the name of a weekday, either as a 3-letter abbreviation or as the full
1563 name, and in both cases you want to extract the abbreviation. This pattern
1564 (ignoring the line breaks) does the job:
1565 .sp
1566 (?<DN>Mon|Fri|Sun)(?:day)?|
1567 (?<DN>Tue)(?:sday)?|
1568 (?<DN>Wed)(?:nesday)?|
1569 (?<DN>Thu)(?:rsday)?|
1570 (?<DN>Sat)(?:urday)?
1571 .sp
1572 There are five capturing substrings, but only one is ever set after a match.
1573 (An alternative way of solving this problem is to use a "branch reset"
1574 subpattern, as described in the previous section.)
1575 .P
1576 The convenience function for extracting the data by name returns the substring
1577 for the first (and in this example, the only) subpattern of that name that
1578 matched. This saves searching to find which numbered subpattern it was.
1579 .P
1580 If you make a back reference to a non-unique named subpattern from elsewhere in
1581 the pattern, the subpatterns to which the name refers are checked in the order
1582 in which they appear in the overall pattern. The first one that is set is used
1583 for the reference. For example, this pattern matches both "foofoo" and
1584 "barbar" but not "foobar" or "barfoo":
1585 .sp
1586 (?:(?<n>foo)|(?<n>bar))\k<n>
1587 .sp
1588 .P
1589 If you make a subroutine call to a non-unique named subpattern, the one that
1590 corresponds to the first occurrence of the name is used. In the absence of
1591 duplicate numbers (see the previous section) this is the one with the lowest
1592 number.
1593 .P
1594 If you use a named reference in a condition
1595 test (see the
1596 .\"
1597 .\" HTML <a href="#conditions">
1598 .\" </a>
1599 section about conditions
1600 .\"
1601 below), either to check whether a subpattern has matched, or to check for
1602 recursion, all subpatterns with the same name are tested. If the condition is
1603 true for any one of them, the overall condition is true. This is the same
1604 behaviour as testing by number. For further details of the interfaces for
1605 handling named subpatterns, see the
1606 .\" HREF
1607 \fBpcreapi\fP
1608 .\"
1609 documentation.
1610 .P
1611 \fBWarning:\fP You cannot use different names to distinguish between two
1612 subpatterns with the same number because PCRE uses only the numbers when
1613 matching. For this reason, an error is given at compile time if different names
1614 are given to subpatterns with the same number. However, you can always give the
1615 same name to subpatterns with the same number, even when PCRE_DUPNAMES is not
1616 set.
1617 .
1618 .
1620 .rs
1621 .sp
1622 Repetition is specified by quantifiers, which can follow any of the following
1623 items:
1624 .sp
1625 a literal data character
1626 the dot metacharacter
1627 the \eC escape sequence
1628 the \eX escape sequence
1629 the \eR escape sequence
1630 an escape such as \ed or \epL that matches a single character
1631 a character class
1632 a back reference (see next section)
1633 a parenthesized subpattern (including assertions)
1634 a subroutine call to a subpattern (recursive or otherwise)
1635 .sp
1636 The general repetition quantifier specifies a minimum and maximum number of
1637 permitted matches, by giving the two numbers in curly brackets (braces),
1638 separated by a comma. The numbers must be less than 65536, and the first must
1639 be less than or equal to the second. For example:
1640 .sp
1641 z{2,4}
1642 .sp
1643 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1644 character. If the second number is omitted, but the comma is present, there is
1645 no upper limit; if the second number and the comma are both omitted, the
1646 quantifier specifies an exact number of required matches. Thus
1647 .sp
1648 [aeiou]{3,}
1649 .sp
1650 matches at least 3 successive vowels, but may match many more, while
1651 .sp
1652 \ed{8}
1653 .sp
1654 matches exactly 8 digits. An opening curly bracket that appears in a position
1655 where a quantifier is not allowed, or one that does not match the syntax of a
1656 quantifier, is taken as a literal character. For example, {,6} is not a
1657 quantifier, but a literal string of four characters.
1658 .P
1659 In UTF modes, quantifiers apply to characters rather than to individual data
1660 units. Thus, for example, \ex{100}{2} matches two characters, each of
1661 which is represented by a two-byte sequence in a UTF-8 string. Similarly,
1662 \eX{3} matches three Unicode extended grapheme clusters, each of which may be
1663 several data units long (and they may be of different lengths).
1664 .P
1665 The quantifier {0} is permitted, causing the expression to behave as if the
1666 previous item and the quantifier were not present. This may be useful for
1667 subpatterns that are referenced as
1668 .\" HTML <a href="#subpatternsassubroutines">
1669 .\" </a>
1670 subroutines
1671 .\"
1672 from elsewhere in the pattern (but see also the section entitled
1673 .\" HTML <a href="#subdefine">
1674 .\" </a>
1675 "Defining subpatterns for use by reference only"
1676 .\"
1677 below). Items other than subpatterns that have a {0} quantifier are omitted
1678 from the compiled pattern.
1679 .P
1680 For convenience, the three most common quantifiers have single-character
1681 abbreviations:
1682 .sp
1683 * is equivalent to {0,}
1684 + is equivalent to {1,}
1685 ? is equivalent to {0,1}
1686 .sp
1687 It is possible to construct infinite loops by following a subpattern that can
1688 match no characters with a quantifier that has no upper limit, for example:
1689 .sp
1690 (a?)*
1691 .sp
1692 Earlier versions of Perl and PCRE used to give an error at compile time for
1693 such patterns. However, because there are cases where this can be useful, such
1694 patterns are now accepted, but if any repetition of the subpattern does in fact
1695 match no characters, the loop is forcibly broken.
1696 .P
1697 By default, the quantifiers are "greedy", that is, they match as much as
1698 possible (up to the maximum number of permitted times), without causing the
1699 rest of the pattern to fail. The classic example of where this gives problems
1700 is in trying to match comments in C programs. These appear between /* and */
1701 and within the comment, individual * and / characters may appear. An attempt to
1702 match C comments by applying the pattern
1703 .sp
1704 /\e*.*\e*/
1705 .sp
1706 to the string
1707 .sp
1708 /* first comment */ not comment /* second comment */
1709 .sp
1710 fails, because it matches the entire string owing to the greediness of the .*
1711 item.
1712 .P
1713 However, if a quantifier is followed by a question mark, it ceases to be
1714 greedy, and instead matches the minimum number of times possible, so the
1715 pattern
1716 .sp
1717 /\e*.*?\e*/
1718 .sp
1719 does the right thing with the C comments. The meaning of the various
1720 quantifiers is not otherwise changed, just the preferred number of matches.
1721 Do not confuse this use of question mark with its use as a quantifier in its
1722 own right. Because it has two uses, it can sometimes appear doubled, as in
1723 .sp
1724 \ed??\ed
1725 .sp
1726 which matches one digit by preference, but can match two if that is the only
1727 way the rest of the pattern matches.
1728 .P
1729 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1730 the quantifiers are not greedy by default, but individual ones can be made
1731 greedy by following them with a question mark. In other words, it inverts the
1732 default behaviour.
1733 .P
1734 When a parenthesized subpattern is quantified with a minimum repeat count that
1735 is greater than 1 or with a limited maximum, more memory is required for the
1736 compiled pattern, in proportion to the size of the minimum or maximum.
1737 .P
1738 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1739 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1740 implicitly anchored, because whatever follows will be tried against every
1741 character position in the subject string, so there is no point in retrying the
1742 overall match at any position after the first. PCRE normally treats such a
1743 pattern as though it were preceded by \eA.
1744 .P
1745 In cases where it is known that the subject string contains no newlines, it is
1746 worth setting PCRE_DOTALL in order to obtain this optimization, or
1747 alternatively using ^ to indicate anchoring explicitly.
1748 .P
1749 However, there are some cases where the optimization cannot be used. When .*
1750 is inside capturing parentheses that are the subject of a back reference
1751 elsewhere in the pattern, a match at the start may fail where a later one
1752 succeeds. Consider, for example:
1753 .sp
1754 (.*)abc\e1
1755 .sp
1756 If the subject is "xyz123abc123" the match point is the fourth character. For
1757 this reason, such a pattern is not implicitly anchored.
1758 .P
1759 Another case where implicit anchoring is not applied is when the leading .* is
1760 inside an atomic group. Once again, a match at the start may fail where a later
1761 one succeeds. Consider this pattern:
1762 .sp
1763 (?>.*?a)b
1764 .sp
1765 It matches "ab" in the subject "aab". The use of the backtracking control verbs
1766 (*PRUNE) and (*SKIP) also disable this optimization.
1767 .P
1768 When a capturing subpattern is repeated, the value captured is the substring
1769 that matched the final iteration. For example, after
1770 .sp
1771 (tweedle[dume]{3}\es*)+
1772 .sp
1773 has matched "tweedledum tweedledee" the value of the captured substring is
1774 "tweedledee". However, if there are nested capturing subpatterns, the
1775 corresponding captured values may have been set in previous iterations. For
1776 example, after
1777 .sp
1778 /(a|(b))+/
1779 .sp
1780 matches "aba" the value of the second captured substring is "b".
1781 .
1782 .
1783 .\" HTML <a name="atomicgroup"></a>
1785 .rs
1786 .sp
1787 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1788 repetition, failure of what follows normally causes the repeated item to be
1789 re-evaluated to see if a different number of repeats allows the rest of the
1790 pattern to match. Sometimes it is useful to prevent this, either to change the
1791 nature of the match, or to cause it fail earlier than it otherwise might, when
1792 the author of the pattern knows there is no point in carrying on.
1793 .P
1794 Consider, for example, the pattern \ed+foo when applied to the subject line
1795 .sp
1796 123456bar
1797 .sp
1798 After matching all 6 digits and then failing to match "foo", the normal
1799 action of the matcher is to try again with only 5 digits matching the \ed+
1800 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1801 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1802 that once a subpattern has matched, it is not to be re-evaluated in this way.
1803 .P
1804 If we use atomic grouping for the previous example, the matcher gives up
1805 immediately on failing to match "foo" the first time. The notation is a kind of
1806 special parenthesis, starting with (?> as in this example:
1807 .sp
1808 (?>\ed+)foo
1809 .sp
1810 This kind of parenthesis "locks up" the part of the pattern it contains once
1811 it has matched, and a failure further into the pattern is prevented from
1812 backtracking into it. Backtracking past it to previous items, however, works as
1813 normal.
1814 .P
1815 An alternative description is that a subpattern of this type matches the string
1816 of characters that an identical standalone pattern would match, if anchored at
1817 the current point in the subject string.
1818 .P
1819 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1820 the above example can be thought of as a maximizing repeat that must swallow
1821 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1822 number of digits they match in order to make the rest of the pattern match,
1823 (?>\ed+) can only match an entire sequence of digits.
1824 .P
1825 Atomic groups in general can of course contain arbitrarily complicated
1826 subpatterns, and can be nested. However, when the subpattern for an atomic
1827 group is just a single repeated item, as in the example above, a simpler
1828 notation, called a "possessive quantifier" can be used. This consists of an
1829 additional + character following a quantifier. Using this notation, the
1830 previous example can be rewritten as
1831 .sp
1832 \ed++foo
1833 .sp
1834 Note that a possessive quantifier can be used with an entire group, for
1835 example:
1836 .sp
1837 (abc|xyz){2,3}+
1838 .sp
1839 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1840 option is ignored. They are a convenient notation for the simpler forms of
1841 atomic group. However, there is no difference in the meaning of a possessive
1842 quantifier and the equivalent atomic group, though there may be a performance
1843 difference; possessive quantifiers should be slightly faster.
1844 .P
1845 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1846 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1847 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1848 package, and PCRE copied it from there. It ultimately found its way into Perl
1849 at release 5.10.
1850 .P
1851 PCRE has an optimization that automatically "possessifies" certain simple
1852 pattern constructs. For example, the sequence A+B is treated as A++B because
1853 there is no point in backtracking into a sequence of A's when B must follow.
1854 .P
1855 When a pattern contains an unlimited repeat inside a subpattern that can itself
1856 be repeated an unlimited number of times, the use of an atomic group is the
1857 only way to avoid some failing matches taking a very long time indeed. The
1858 pattern
1859 .sp
1860 (\eD+|<\ed+>)*[!?]
1861 .sp
1862 matches an unlimited number of substrings that either consist of non-digits, or
1863 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1864 quickly. However, if it is applied to
1865 .sp
1866 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1867 .sp
1868 it takes a long time before reporting failure. This is because the string can
1869 be divided between the internal \eD+ repeat and the external * repeat in a
1870 large number of ways, and all have to be tried. (The example uses [!?] rather
1871 than a single character at the end, because both PCRE and Perl have an
1872 optimization that allows for fast failure when a single character is used. They
1873 remember the last single character that is required for a match, and fail early
1874 if it is not present in the string.) If the pattern is changed so that it uses
1875 an atomic group, like this:
1876 .sp
1877 ((?>\eD+)|<\ed+>)*[!?]
1878 .sp
1879 sequences of non-digits cannot be broken, and failure happens quickly.
1880 .
1881 .
1882 .\" HTML <a name="backreferences"></a>
1884 .rs
1885 .sp
1886 Outside a character class, a backslash followed by a digit greater than 0 (and
1887 possibly further digits) is a back reference to a capturing subpattern earlier
1888 (that is, to its left) in the pattern, provided there have been that many
1889 previous capturing left parentheses.
1890 .P
1891 However, if the decimal number following the backslash is less than 10, it is
1892 always taken as a back reference, and causes an error only if there are not
1893 that many capturing left parentheses in the entire pattern. In other words, the
1894 parentheses that are referenced need not be to the left of the reference for
1895 numbers less than 10. A "forward back reference" of this type can make sense
1896 when a repetition is involved and the subpattern to the right has participated
1897 in an earlier iteration.
1898 .P
1899 It is not possible to have a numerical "forward back reference" to a subpattern
1900 whose number is 10 or more using this syntax because a sequence such as \e50 is
1901 interpreted as a character defined in octal. See the subsection entitled
1902 "Non-printing characters"
1903 .\" HTML <a href="#digitsafterbackslash">
1904 .\" </a>
1905 above
1906 .\"
1907 for further details of the handling of digits following a backslash. There is
1908 no such problem when named parentheses are used. A back reference to any
1909 subpattern is possible using named parentheses (see below).
1910 .P
1911 Another way of avoiding the ambiguity inherent in the use of digits following a
1912 backslash is to use the \eg escape sequence. This escape must be followed by an
1913 unsigned number or a negative number, optionally enclosed in braces. These
1914 examples are all identical:
1915 .sp
1916 (ring), \e1
1917 (ring), \eg1
1918 (ring), \eg{1}
1919 .sp
1920 An unsigned number specifies an absolute reference without the ambiguity that
1921 is present in the older syntax. It is also useful when literal digits follow
1922 the reference. A negative number is a relative reference. Consider this
1923 example:
1924 .sp
1925 (abc(def)ghi)\eg{-1}
1926 .sp
1927 The sequence \eg{-1} is a reference to the most recently started capturing
1928 subpattern before \eg, that is, is it equivalent to \e2 in this example.
1929 Similarly, \eg{-2} would be equivalent to \e1. The use of relative references
1930 can be helpful in long patterns, and also in patterns that are created by
1931 joining together fragments that contain references within themselves.
1932 .P
1933 A back reference matches whatever actually matched the capturing subpattern in
1934 the current subject string, rather than anything matching the subpattern
1935 itself (see
1936 .\" HTML <a href="#subpatternsassubroutines">
1937 .\" </a>
1938 "Subpatterns as subroutines"
1939 .\"
1940 below for a way of doing that). So the pattern
1941 .sp
1942 (sens|respons)e and \e1ibility
1943 .sp
1944 matches "sense and sensibility" and "response and responsibility", but not
1945 "sense and responsibility". If caseful matching is in force at the time of the
1946 back reference, the case of letters is relevant. For example,
1947 .sp
1948 ((?i)rah)\es+\e1
1949 .sp
1950 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1951 capturing subpattern is matched caselessly.
1952 .P
1953 There are several different ways of writing back references to named
1954 subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1955 \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1956 back reference syntax, in which \eg can be used for both numeric and named
1957 references, is also supported. We could rewrite the above example in any of
1958 the following ways:
1959 .sp
1960 (?<p1>(?i)rah)\es+\ek<p1>
1961 (?'p1'(?i)rah)\es+\ek{p1}
1962 (?P<p1>(?i)rah)\es+(?P=p1)
1963 (?<p1>(?i)rah)\es+\eg{p1}
1964 .sp
1965 A subpattern that is referenced by name may appear in the pattern before or
1966 after the reference.
1967 .P
1968 There may be more than one back reference to the same subpattern. If a
1969 subpattern has not actually been used in a particular match, any back
1970 references to it always fail by default. For example, the pattern
1971 .sp
1972 (a|(bc))\e2
1973 .sp
1974 always fails if it starts to match "a" rather than "bc". However, if the
1975 PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1976 unset value matches an empty string.
1977 .P
1978 Because there may be many capturing parentheses in a pattern, all digits
1979 following a backslash are taken as part of a potential back reference number.
1980 If the pattern continues with a digit character, some delimiter must be used to
1981 terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1982 white space. Otherwise, the \eg{ syntax or an empty comment (see
1983 .\" HTML <a href="#comments">
1984 .\" </a>
1985 "Comments"
1986 .\"
1987 below) can be used.
1988 .
1989 .SS "Recursive back references"
1990 .rs
1991 .sp
1992 A back reference that occurs inside the parentheses to which it refers fails
1993 when the subpattern is first used, so, for example, (a\e1) never matches.
1994 However, such references can be useful inside repeated subpatterns. For
1995 example, the pattern
1996 .sp
1997 (a|b\e1)+
1998 .sp
1999 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
2000 the subpattern, the back reference matches the character string corresponding
2001 to the previous iteration. In order for this to work, the pattern must be such
2002 that the first iteration does not need to match the back reference. This can be
2003 done using alternation, as in the example above, or by a quantifier with a
2004 minimum of zero.
2005 .P
2006 Back references of this type cause the group that they reference to be treated
2007 as an
2008 .\" HTML <a href="#atomicgroup">
2009 .\" </a>
2010 atomic group.
2011 .\"
2012 Once the whole group has been matched, a subsequent matching failure cannot
2013 cause backtracking into the middle of the group.
2014 .
2015 .
2016 .\" HTML <a name="bigassertions"></a>
2018 .rs
2019 .sp
2020 An assertion is a test on the characters following or preceding the current
2021 matching point that does not actually consume any characters. The simple
2022 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
2023 .\" HTML <a href="#smallassertions">
2024 .\" </a>
2025 above.
2026 .\"
2027 .P
2028 More complicated assertions are coded as subpatterns. There are two kinds:
2029 those that look ahead of the current position in the subject string, and those
2030 that look behind it. An assertion subpattern is matched in the normal way,
2031 except that it does not cause the current matching position to be changed.
2032 .P
2033 Assertion subpatterns are not capturing subpatterns. If such an assertion
2034 contains capturing subpatterns within it, these are counted for the purposes of
2035 numbering the capturing subpatterns in the whole pattern. However, substring
2036 capturing is carried out only for positive assertions. (Perl sometimes, but not
2037 always, does do capturing in negative assertions.)
2038 .P
2039 For compatibility with Perl, assertion subpatterns may be repeated; though
2040 it makes no sense to assert the same thing several times, the side effect of
2041 capturing parentheses may occasionally be useful. In practice, there only three
2042 cases:
2043 .sp
2044 (1) If the quantifier is {0}, the assertion is never obeyed during matching.
2045 However, it may contain internal capturing parenthesized groups that are called
2046 from elsewhere via the
2047 .\" HTML <a href="#subpatternsassubroutines">
2048 .\" </a>
2049 subroutine mechanism.
2050 .\"
2051 .sp
2052 (2) If quantifier is {0,n} where n is greater than zero, it is treated as if it
2053 were {0,1}. At run time, the rest of the pattern match is tried with and
2054 without the assertion, the order depending on the greediness of the quantifier.
2055 .sp
2056 (3) If the minimum repetition is greater than zero, the quantifier is ignored.
2057 The assertion is obeyed just once when encountered during matching.
2058 .
2059 .
2060 .SS "Lookahead assertions"
2061 .rs
2062 .sp
2063 Lookahead assertions start with (?= for positive assertions and (?! for
2064 negative assertions. For example,
2065 .sp
2066 \ew+(?=;)
2067 .sp
2068 matches a word followed by a semicolon, but does not include the semicolon in
2069 the match, and
2070 .sp
2071 foo(?!bar)
2072 .sp
2073 matches any occurrence of "foo" that is not followed by "bar". Note that the
2074 apparently similar pattern
2075 .sp
2076 (?!foo)bar
2077 .sp
2078 does not find an occurrence of "bar" that is preceded by something other than
2079 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
2080 (?!foo) is always true when the next three characters are "bar". A
2081 lookbehind assertion is needed to achieve the other effect.
2082 .P
2083 If you want to force a matching failure at some point in a pattern, the most
2084 convenient way to do it is with (?!) because an empty string always matches, so
2085 an assertion that requires there not to be an empty string must always fail.
2086 The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).
2087 .
2088 .
2089 .\" HTML <a name="lookbehind"></a>
2090 .SS "Lookbehind assertions"
2091 .rs
2092 .sp
2093 Lookbehind assertions start with (?<= for positive assertions and (?<! for
2094 negative assertions. For example,
2095 .sp
2096 (?<!foo)bar
2097 .sp
2098 does find an occurrence of "bar" that is not preceded by "foo". The contents of
2099 a lookbehind assertion are restricted such that all the strings it matches must
2100 have a fixed length. However, if there are several top-level alternatives, they
2101 do not all have to have the same fixed length. Thus
2102 .sp
2103 (?<=bullock|donkey)
2104 .sp
2105 is permitted, but
2106 .sp
2107 (?<!dogs?|cats?)
2108 .sp
2109 causes an error at compile time. Branches that match different length strings
2110 are permitted only at the top level of a lookbehind assertion. This is an
2111 extension compared with Perl, which requires all branches to match the same
2112 length of string. An assertion such as
2113 .sp
2114 (?<=ab(c|de))
2115 .sp
2116 is not permitted, because its single top-level branch can match two different
2117 lengths, but it is acceptable to PCRE if rewritten to use two top-level
2118 branches:
2119 .sp
2120 (?<=abc|abde)
2121 .sp
2122 In some cases, the escape sequence \eK
2123 .\" HTML <a href="#resetmatchstart">
2124 .\" </a>
2125 (see above)
2126 .\"
2127 can be used instead of a lookbehind assertion to get round the fixed-length
2128 restriction.
2129 .P
2130 The implementation of lookbehind assertions is, for each alternative, to
2131 temporarily move the current position back by the fixed length and then try to
2132 match. If there are insufficient characters before the current position, the
2133 assertion fails.
2134 .P
2135 In a UTF mode, PCRE does not allow the \eC escape (which matches a single data
2136 unit even in a UTF mode) to appear in lookbehind assertions, because it makes
2137 it impossible to calculate the length of the lookbehind. The \eX and \eR
2138 escapes, which can match different numbers of data units, are also not
2139 permitted.
2140 .P
2141 .\" HTML <a href="#subpatternsassubroutines">
2142 .\" </a>
2143 "Subroutine"
2144 .\"
2145 calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
2146 as the subpattern matches a fixed-length string.
2147 .\" HTML <a href="#recursion">
2148 .\" </a>
2149 Recursion,
2150 .\"
2151 however, is not supported.
2152 .P
2153 Possessive quantifiers can be used in conjunction with lookbehind assertions to
2154 specify efficient matching of fixed-length strings at the end of subject
2155 strings. Consider a simple pattern such as
2156 .sp
2157 abcd$
2158 .sp
2159 when applied to a long string that does not match. Because matching proceeds
2160 from left to right, PCRE will look for each "a" in the subject and then see if
2161 what follows matches the rest of the pattern. If the pattern is specified as
2162 .sp
2163 ^.*abcd$
2164 .sp
2165 the initial .* matches the entire string at first, but when this fails (because
2166 there is no following "a"), it backtracks to match all but the last character,
2167 then all but the last two characters, and so on. Once again the search for "a"
2168 covers the entire string, from right to left, so we are no better off. However,
2169 if the pattern is written as
2170 .sp
2171 ^.*+(?<=abcd)
2172 .sp
2173 there can be no backtracking for the .*+ item; it can match only the entire
2174 string. The subsequent lookbehind assertion does a single test on the last four
2175 characters. If it fails, the match fails immediately. For long strings, this
2176 approach makes a significant difference to the processing time.
2177 .
2178 .
2179 .SS "Using multiple assertions"
2180 .rs
2181 .sp
2182 Several assertions (of any sort) may occur in succession. For example,
2183 .sp
2184 (?<=\ed{3})(?<!999)foo
2185 .sp
2186 matches "foo" preceded by three digits that are not "999". Notice that each of
2187 the assertions is applied independently at the same point in the subject
2188 string. First there is a check that the previous three characters are all
2189 digits, and then there is a check that the same three characters are not "999".
2190 This pattern does \fInot\fP match "foo" preceded by six characters, the first
2191 of which are digits and the last three of which are not "999". For example, it
2192 doesn't match "123abcfoo". A pattern to do that is
2193 .sp
2194 (?<=\ed{3}...)(?<!999)foo
2195 .sp
2196 This time the first assertion looks at the preceding six characters, checking
2197 that the first three are digits, and then the second assertion checks that the
2198 preceding three characters are not "999".
2199 .P
2200 Assertions can be nested in any combination. For example,
2201 .sp
2202 (?<=(?<!foo)bar)baz
2203 .sp
2204 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
2205 preceded by "foo", while
2206 .sp
2207 (?<=\ed{3}(?!999)...)foo
2208 .sp
2209 is another pattern that matches "foo" preceded by three digits and any three
2210 characters that are not "999".
2211 .
2212 .
2213 .\" HTML <a name="conditions"></a>
2215 .rs
2216 .sp
2217 It is possible to cause the matching process to obey a subpattern
2218 conditionally or to choose between two alternative subpatterns, depending on
2219 the result of an assertion, or whether a specific capturing subpattern has
2220 already been matched. The two possible forms of conditional subpattern are:
2221 .sp
2222 (?(condition)yes-pattern)
2223 (?(condition)yes-pattern|no-pattern)
2224 .sp
2225 If the condition is satisfied, the yes-pattern is used; otherwise the
2226 no-pattern (if present) is used. If there are more than two alternatives in the
2227 subpattern, a compile-time error occurs. Each of the two alternatives may
2228 itself contain nested subpatterns of any form, including conditional
2229 subpatterns; the restriction to two alternatives applies only at the level of
2230 the condition. This pattern fragment is an example where the alternatives are
2231 complex:
2232 .sp
2233 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
2234 .sp
2235 .P
2236 There are four kinds of condition: references to subpatterns, references to
2237 recursion, a pseudo-condition called DEFINE, and assertions.
2238 .
2239 .SS "Checking for a used subpattern by number"
2240 .rs
2241 .sp
2242 If the text between the parentheses consists of a sequence of digits, the
2243 condition is true if a capturing subpattern of that number has previously
2244 matched. If there is more than one capturing subpattern with the same number
2245 (see the earlier
2246 .\"
2247 .\" HTML <a href="#recursion">
2248 .\" </a>
2249 section about duplicate subpattern numbers),
2250 .\"
2251 the condition is true if any of them have matched. An alternative notation is
2252 to precede the digits with a plus or minus sign. In this case, the subpattern
2253 number is relative rather than absolute. The most recently opened parentheses
2254 can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside
2255 loops it can also make sense to refer to subsequent groups. The next
2256 parentheses to be opened can be referenced as (?(+1), and so on. (The value
2257 zero in any of these forms is not used; it provokes a compile-time error.)
2258 .P
2259 Consider the following pattern, which contains non-significant white space to
2260 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
2261 three parts for ease of discussion:
2262 .sp
2263 ( \e( )? [^()]+ (?(1) \e) )
2264 .sp
2265 The first part matches an optional opening parenthesis, and if that
2266 character is present, sets it as the first captured substring. The second part
2267 matches one or more characters that are not parentheses. The third part is a
2268 conditional subpattern that tests whether or not the first set of parentheses
2269 matched. If they did, that is, if subject started with an opening parenthesis,
2270 the condition is true, and so the yes-pattern is executed and a closing
2271 parenthesis is required. Otherwise, since no-pattern is not present, the
2272 subpattern matches nothing. In other words, this pattern matches a sequence of
2273 non-parentheses, optionally enclosed in parentheses.
2274 .P
2275 If you were embedding this pattern in a larger one, you could use a relative
2276 reference:
2277 .sp
2278 ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
2279 .sp
2280 This makes the fragment independent of the parentheses in the larger pattern.
2281 .
2282 .SS "Checking for a used subpattern by name"
2283 .rs
2284 .sp
2285 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
2286 subpattern by name. For compatibility with earlier versions of PCRE, which had
2287 this facility before Perl, the syntax (?(name)...) is also recognized. However,
2288 there is a possible ambiguity with this syntax, because subpattern names may
2289 consist entirely of digits. PCRE looks first for a named subpattern; if it
2290 cannot find one and the name consists entirely of digits, PCRE looks for a
2291 subpattern of that number, which must be greater than zero. Using subpattern
2292 names that consist entirely of digits is not recommended.
2293 .P
2294 Rewriting the above example to use a named subpattern gives this:
2295 .sp
2296 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
2297 .sp
2298 If the name used in a condition of this kind is a duplicate, the test is
2299 applied to all subpatterns of the same name, and is true if any one of them has
2300 matched.
2301 .
2302 .SS "Checking for pattern recursion"
2303 .rs
2304 .sp
2305 If the condition is the string (R), and there is no subpattern with the name R,
2306 the condition is true if a recursive call to the whole pattern or any
2307 subpattern has been made. If digits or a name preceded by ampersand follow the
2308 letter R, for example:
2309 .sp
2310 (?(R3)...) or (?(R&name)...)
2311 .sp
2312 the condition is true if the most recent recursion is into a subpattern whose
2313 number or name is given. This condition does not check the entire recursion
2314 stack. If the name used in a condition of this kind is a duplicate, the test is
2315 applied to all subpatterns of the same name, and is true if any one of them is
2316 the most recent recursion.
2317 .P
2318 At "top level", all these recursion test conditions are false.
2319 .\" HTML <a href="#recursion">
2320 .\" </a>
2321 The syntax for recursive patterns
2322 .\"
2323 is described below.
2324 .
2325 .\" HTML <a name="subdefine"></a>
2326 .SS "Defining subpatterns for use by reference only"
2327 .rs
2328 .sp
2329 If the condition is the string (DEFINE), and there is no subpattern with the
2330 name DEFINE, the condition is always false. In this case, there may be only one
2331 alternative in the subpattern. It is always skipped if control reaches this
2332 point in the pattern; the idea of DEFINE is that it can be used to define
2333 subroutines that can be referenced from elsewhere. (The use of
2334 .\" HTML <a href="#subpatternsassubroutines">
2335 .\" </a>
2336 subroutines
2337 .\"
2338 is described below.) For example, a pattern to match an IPv4 address such as
2339 "" could be written like this (ignore white space and line
2340 breaks):
2341 .sp
2342 (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
2343 \eb (?&byte) (\e.(?&byte)){3} \eb
2344 .sp
2345 The first part of the pattern is a DEFINE group inside which a another group
2346 named "byte" is defined. This matches an individual component of an IPv4
2347 address (a number less than 256). When matching takes place, this part of the
2348 pattern is skipped because DEFINE acts like a false condition. The rest of the
2349 pattern uses references to the named group to match the four dot-separated
2350 components of an IPv4 address, insisting on a word boundary at each end.
2351 .
2352 .SS "Assertion conditions"
2353 .rs
2354 .sp
2355 If the condition is not in any of the above formats, it must be an assertion.
2356 This may be a positive or negative lookahead or lookbehind assertion. Consider
2357 this pattern, again containing non-significant white space, and with the two
2358 alternatives on the second line:
2359 .sp
2360 (?(?=[^a-z]*[a-z])
2361 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
2362 .sp
2363 The condition is a positive lookahead assertion that matches an optional
2364 sequence of non-letters followed by a letter. In other words, it tests for the
2365 presence of at least one letter in the subject. If a letter is found, the
2366 subject is matched against the first alternative; otherwise it is matched
2367 against the second. This pattern matches strings in one of the two forms
2368 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2369 .
2370 .
2371 .\" HTML <a name="comments"></a>
2373 .rs
2374 .sp
2375 There are two ways of including comments in patterns that are processed by
2376 PCRE. In both cases, the start of the comment must not be in a character class,
2377 nor in the middle of any other sequence of related characters such as (?: or a
2378 subpattern name or number. The characters that make up a comment play no part
2379 in the pattern matching.
2380 .P
2381 The sequence (?# marks the start of a comment that continues up to the next
2382 closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED
2383 option is set, an unescaped # character also introduces a comment, which in
2384 this case continues to immediately after the next newline character or
2385 character sequence in the pattern. Which characters are interpreted as newlines
2386 is controlled by the options passed to a compiling function or by a special
2387 sequence at the start of the pattern, as described in the section entitled
2388 .\" HTML <a href="#newlines">
2389 .\" </a>
2390 "Newline conventions"
2391 .\"
2392 above. Note that the end of this type of comment is a literal newline sequence
2393 in the pattern; escape sequences that happen to represent a newline do not
2394 count. For example, consider this pattern when PCRE_EXTENDED is set, and the
2395 default newline convention is in force:
2396 .sp
2397 abc #comment \en still comment
2398 .sp
2399 On encountering the # character, \fBpcre_compile()\fP skips along, looking for
2400 a newline in the pattern. The sequence \en is still literal at this stage, so
2401 it does not terminate the comment. Only an actual character with the code value
2402 0x0a (the default newline) does so.
2403 .
2404 .
2405 .\" HTML <a name="recursion"></a>
2407 .rs
2408 .sp
2409 Consider the problem of matching a string in parentheses, allowing for
2410 unlimited nested parentheses. Without the use of recursion, the best that can
2411 be done is to use a pattern that matches up to some fixed depth of nesting. It
2412 is not possible to handle an arbitrary nesting depth.
2413 .P
2414 For some time, Perl has provided a facility that allows regular expressions to
2415 recurse (amongst other things). It does this by interpolating Perl code in the
2416 expression at run time, and the code can refer to the expression itself. A Perl
2417 pattern using code interpolation to solve the parentheses problem can be
2418 created like this:
2419 .sp
2420 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
2421 .sp
2422 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2423 recursively to the pattern in which it appears.
2424 .P
2425 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2426 supports special syntax for recursion of the entire pattern, and also for
2427 individual subpattern recursion. After its introduction in PCRE and Python,
2428 this kind of recursion was subsequently introduced into Perl at release 5.10.
2429 .P
2430 A special item that consists of (? followed by a number greater than zero and a
2431 closing parenthesis is a recursive subroutine call of the subpattern of the
2432 given number, provided that it occurs inside that subpattern. (If not, it is a
2433 .\" HTML <a href="#subpatternsassubroutines">
2434 .\" </a>
2435 non-recursive subroutine
2436 .\"
2437 call, which is described in the next section.) The special item (?R) or (?0) is
2438 a recursive call of the entire regular expression.
2439 .P
2440 This PCRE pattern solves the nested parentheses problem (assume the
2441 PCRE_EXTENDED option is set so that white space is ignored):
2442 .sp
2443 \e( ( [^()]++ | (?R) )* \e)
2444 .sp
2445 First it matches an opening parenthesis. Then it matches any number of
2446 substrings which can either be a sequence of non-parentheses, or a recursive
2447 match of the pattern itself (that is, a correctly parenthesized substring).
2448 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2449 to avoid backtracking into sequences of non-parentheses.
2450 .P
2451 If this were part of a larger pattern, you would not want to recurse the entire
2452 pattern, so instead you could use this:
2453 .sp
2454 ( \e( ( [^()]++ | (?1) )* \e) )
2455 .sp
2456 We have put the pattern into parentheses, and caused the recursion to refer to
2457 them instead of the whole pattern.
2458 .P
2459 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2460 is made easier by the use of relative references. Instead of (?1) in the
2461 pattern above you can write (?-2) to refer to the second most recently opened
2462 parentheses preceding the recursion. In other words, a negative number counts
2463 capturing parentheses leftwards from the point at which it is encountered.
2464 .P
2465 It is also possible to refer to subsequently opened parentheses, by writing
2466 references such as (?+2). However, these cannot be recursive because the
2467 reference is not inside the parentheses that are referenced. They are always
2468 .\" HTML <a href="#subpatternsassubroutines">
2469 .\" </a>
2470 non-recursive subroutine
2471 .\"
2472 calls, as described in the next section.
2473 .P
2474 An alternative approach is to use named parentheses instead. The Perl syntax
2475 for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
2476 could rewrite the above example as follows:
2477 .sp
2478 (?<pn> \e( ( [^()]++ | (?&pn) )* \e) )
2479 .sp
2480 If there is more than one subpattern with the same name, the earliest one is
2481 used.
2482 .P
2483 This particular example pattern that we have been looking at contains nested
2484 unlimited repeats, and so the use of a possessive quantifier for matching
2485 strings of non-parentheses is important when applying the pattern to strings
2486 that do not match. For example, when this pattern is applied to
2487 .sp
2488 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2489 .sp
2490 it yields "no match" quickly. However, if a possessive quantifier is not used,
2491 the match runs for a very long time indeed because there are so many different
2492 ways the + and * repeats can carve up the subject, and all have to be tested
2493 before failure can be reported.
2494 .P
2495 At the end of a match, the values of capturing parentheses are those from
2496 the outermost level. If you want to obtain intermediate values, a callout
2497 function can be used (see below and the
2498 .\" HREF
2499 \fBpcrecallout\fP
2500 .\"
2501 documentation). If the pattern above is matched against
2502 .sp
2503 (ab(cd)ef)
2504 .sp
2505 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2506 the last value taken on at the top level. If a capturing subpattern is not
2507 matched at the top level, its final captured value is unset, even if it was
2508 (temporarily) set at a deeper level during the matching process.
2509 .P
2510 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2511 obtain extra memory to store data during a recursion, which it does by using
2512 \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no memory can
2513 be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2514 .P
2515 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2516 Consider this pattern, which matches text in angle brackets, allowing for
2517 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2518 recursing), whereas any characters are permitted at the outer level.
2519 .sp
2520 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
2521 .sp
2522 In this pattern, (?(R) is the start of a conditional subpattern, with two
2523 different alternatives for the recursive and non-recursive cases. The (?R) item
2524 is the actual recursive call.
2525 .
2526 .
2527 .\" HTML <a name="recursiondifference"></a>
2528 .SS "Differences in recursion processing between PCRE and Perl"
2529 .rs
2530 .sp
2531 Recursion processing in PCRE differs from Perl in two important ways. In PCRE
2532 (like Python, but unlike Perl), a recursive subpattern call is always treated
2533 as an atomic group. That is, once it has matched some of the subject string, it
2534 is never re-entered, even if it contains untried alternatives and there is a
2535 subsequent matching failure. This can be illustrated by the following pattern,
2536 which purports to match a palindromic string that contains an odd number of
2537 characters (for example, "a", "aba", "abcba", "abcdcba"):
2538 .sp
2539 ^(.|(.)(?1)\e2)$
2540 .sp
2541 The idea is that it either matches a single character, or two identical
2542 characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2543 it does not if the pattern is longer than three characters. Consider the
2544 subject string "abcba":
2545 .P
2546 At the top level, the first character is matched, but as it is not at the end
2547 of the string, the first alternative fails; the second alternative is taken
2548 and the recursion kicks in. The recursive call to subpattern 1 successfully
2549 matches the next character ("b"). (Note that the beginning and end of line
2550 tests are not part of the recursion).
2551 .P
2552 Back at the top level, the next character ("c") is compared with what
2553 subpattern 2 matched, which was "a". This fails. Because the recursion is
2554 treated as an atomic group, there are now no backtracking points, and so the
2555 entire match fails. (Perl is able, at this point, to re-enter the recursion and
2556 try the second alternative.) However, if the pattern is written with the
2557 alternatives in the other order, things are different:
2558 .sp
2559 ^((.)(?1)\e2|.)$
2560 .sp
2561 This time, the recursing alternative is tried first, and continues to recurse
2562 until it runs out of characters, at which point the recursion fails. But this
2563 time we do have another alternative to try at the higher level. That is the big
2564 difference: in the previous case the remaining alternative is at a deeper
2565 recursion level, which PCRE cannot use.
2566 .P
2567 To change the pattern so that it matches all palindromic strings, not just
2568 those with an odd number of characters, it is tempting to change the pattern to
2569 this:
2570 .sp
2571 ^((.)(?1)\e2|.?)$
2572 .sp
2573 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2574 deeper recursion has matched a single character, it cannot be entered again in
2575 order to match an empty string. The solution is to separate the two cases, and
2576 write out the odd and even cases as alternatives at the higher level:
2577 .sp
2578 ^(?:((.)(?1)\e2|)|((.)(?3)\e4|.))
2579 .sp
2580 If you want to match typical palindromic phrases, the pattern has to ignore all
2581 non-word characters, which can be done like this:
2582 .sp
2583 ^\eW*+(?:((.)\eW*+(?1)\eW*+\e2|)|((.)\eW*+(?3)\eW*+\e4|\eW*+.\eW*+))\eW*+$
2584 .sp
2585 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2586 man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2587 the use of the possessive quantifier *+ to avoid backtracking into sequences of
2588 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2589 more) to match typical phrases, and Perl takes so long that you think it has
2590 gone into a loop.
2591 .P
2592 \fBWARNING\fP: The palindrome-matching patterns above work only if the subject
2593 string does not start with a palindrome that is shorter than the entire string.
2594 For example, although "abcba" is correctly matched, if the subject is "ababa",
2595 PCRE finds the palindrome "aba" at the start, then fails at top level because
2596 the end of the string does not follow. Once again, it cannot jump back into the
2597 recursion to try other alternatives, so the entire match fails.
2598 .P
2599 The second way in which PCRE and Perl differ in their recursion processing is
2600 in the handling of captured values. In Perl, when a subpattern is called
2601 recursively or as a subpattern (see the next section), it has no access to any
2602 values that were captured outside the recursion, whereas in PCRE these values
2603 can be referenced. Consider this pattern:
2604 .sp
2605 ^(.)(\e1|a(?2))
2606 .sp
2607 In PCRE, this pattern matches "bab". The first capturing parentheses match "b",
2608 then in the second group, when the back reference \e1 fails to match "b", the
2609 second alternative matches "a" and then recurses. In the recursion, \e1 does
2610 now match "b" and so the whole match succeeds. In Perl, the pattern fails to
2611 match because inside the recursive call \e1 cannot access the externally set
2612 value.
2613 .
2614 .
2615 .\" HTML <a name="subpatternsassubroutines"></a>
2617 .rs
2618 .sp
2619 If the syntax for a recursive subpattern call (either by number or by
2620 name) is used outside the parentheses to which it refers, it operates like a
2621 subroutine in a programming language. The called subpattern may be defined
2622 before or after the reference. A numbered reference can be absolute or
2623 relative, as in these examples:
2624 .sp
2625 (...(absolute)...)...(?2)...
2626 (...(relative)...)...(?-1)...
2627 (...(?+1)...(relative)...
2628 .sp
2629 An earlier example pointed out that the pattern
2630 .sp
2631 (sens|respons)e and \e1ibility
2632 .sp
2633 matches "sense and sensibility" and "response and responsibility", but not
2634 "sense and responsibility". If instead the pattern
2635 .sp
2636 (sens|respons)e and (?1)ibility
2637 .sp
2638 is used, it does match "sense and responsibility" as well as the other two
2639 strings. Another example is given in the discussion of DEFINE above.
2640 .P
2641 All subroutine calls, whether recursive or not, are always treated as atomic
2642 groups. That is, once a subroutine has matched some of the subject string, it
2643 is never re-entered, even if it contains untried alternatives and there is a
2644 subsequent matching failure. Any capturing parentheses that are set during the
2645 subroutine call revert to their previous values afterwards.
2646 .P
2647 Processing options such as case-independence are fixed when a subpattern is
2648 defined, so if it is used as a subroutine, such options cannot be changed for
2649 different calls. For example, consider this pattern:
2650 .sp
2651 (abc)(?i:(?-1))
2652 .sp
2653 It matches "abcabc". It does not match "abcABC" because the change of
2654 processing option does not affect the called subpattern.
2655 .
2656 .
2657 .\" HTML <a name="onigurumasubroutines"></a>
2659 .rs
2660 .sp
2661 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
2662 a number enclosed either in angle brackets or single quotes, is an alternative
2663 syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2664 are two of the examples used above, rewritten using this syntax:
2665 .sp
2666 (?<pn> \e( ( (?>[^()]+) | \eg<pn> )* \e) )
2667 (sens|respons)e and \eg'1'ibility
2668 .sp
2669 PCRE supports an extension to Oniguruma: if a number is preceded by a
2670 plus or a minus sign it is taken as a relative reference. For example:
2671 .sp
2672 (abc)(?i:\eg<-1>)
2673 .sp
2674 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
2675 synonymous. The former is a back reference; the latter is a subroutine call.
2676 .
2677 .
2679 .rs
2680 .sp
2681 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2682 code to be obeyed in the middle of matching a regular expression. This makes it
2683 possible, amongst other things, to extract different substrings that match the
2684 same pair of parentheses when there is a repetition.
2685 .P
2686 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2687 code. The feature is called "callout". The caller of PCRE provides an external
2688 function by putting its entry point in the global variable \fIpcre_callout\fP
2689 (8-bit library) or \fIpcre[16|32]_callout\fP (16-bit or 32-bit library).
2690 By default, this variable contains NULL, which disables all calling out.
2691 .P
2692 Within a regular expression, (?C) indicates the points at which the external
2693 function is to be called. If you want to identify different callout points, you
2694 can put a number less than 256 after the letter C. The default value is zero.
2695 For example, this pattern has two callout points:
2696 .sp
2697 (?C1)abc(?C2)def
2698 .sp
2699 If the PCRE_AUTO_CALLOUT flag is passed to a compiling function, callouts are
2700 automatically installed before each item in the pattern. They are all numbered
2701 255. If there is a conditional group in the pattern whose condition is an
2702 assertion, an additional callout is inserted just before the condition. An
2703 explicit callout may also be set at this position, as in this example:
2704 .sp
2705 (?(?C9)(?=a)abc|def)
2706 .sp
2707 Note that this applies only to assertion conditions, not to other types of
2708 condition.
2709 .P
2710 During matching, when PCRE reaches a callout point, the external function is
2711 called. It is provided with the number of the callout, the position in the
2712 pattern, and, optionally, one item of data originally supplied by the caller of
2713 the matching function. The callout function may cause matching to proceed, to
2714 backtrack, or to fail altogether. A complete description of the interface to
2715 the callout function is given in the
2716 .\" HREF
2717 \fBpcrecallout\fP
2718 .\"
2719 documentation.
2720 .
2721 .
2722 .\" HTML <a name="backtrackcontrol"></a>
2724 .rs
2725 .sp
2726 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2727 are still described in the Perl documentation as "experimental and subject to
2728 change or removal in a future version of Perl". It goes on to say: "Their usage
2729 in production code should be noted to avoid problems during upgrades." The same
2730 remarks apply to the PCRE features described in this section.
2731 .P
2732 The new verbs make use of what was previously invalid syntax: an opening
2733 parenthesis followed by an asterisk. They are generally of the form
2734 (*VERB) or (*VERB:NAME). Some may take either form, possibly behaving
2735 differently depending on whether or not a name is present. A name is any
2736 sequence of characters that does not include a closing parenthesis. The maximum
2737 length of name is 255 in the 8-bit library and 65535 in the 16-bit and 32-bit
2738 libraries. If the name is empty, that is, if the closing parenthesis
2739 immediately follows the colon, the effect is as if the colon were not there.
2740 Any number of these verbs may occur in a pattern.
2741 .P
2742 Since these verbs are specifically related to backtracking, most of them can be
2743 used only when the pattern is to be matched using one of the traditional
2744 matching functions, because these use a backtracking algorithm. With the
2745 exception of (*FAIL), which behaves like a failing negative assertion, the
2746 backtracking control verbs cause an error if encountered by a DFA matching
2747 function.
2748 .P
2749 The behaviour of these verbs in
2750 .\" HTML <a href="#btrepeat">
2751 .\" </a>
2752 repeated groups,
2753 .\"
2754 .\" HTML <a href="#btassert">
2755 .\" </a>
2756 assertions,
2757 .\"
2758 and in
2759 .\" HTML <a href="#btsub">
2760 .\" </a>
2761 subpatterns called as subroutines
2762 .\"
2763 (whether or not recursively) is documented below.
2764 .
2765 .
2766 .\" HTML <a name="nooptimize"></a>
2767 .SS "Optimizations that affect backtracking verbs"
2768 .rs
2769 .sp
2770 PCRE contains some optimizations that are used to speed up matching by running
2771 some checks at the start of each match attempt. For example, it may know the
2772 minimum length of matching subject, or that a particular character must be
2773 present. When one of these optimizations bypasses the running of a match, any
2774 included backtracking verbs will not, of course, be processed. You can suppress
2775 the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2776 when calling \fBpcre_compile()\fP or \fBpcre_exec()\fP, or by starting the
2777 pattern with (*NO_START_OPT). There is more discussion of this option in the
2778 section entitled
2779 .\" HTML <a href="pcreapi.html#execoptions">
2780 .\" </a>
2781 "Option bits for \fBpcre_exec()\fP"
2782 .\"
2783 in the
2784 .\" HREF
2785 \fBpcreapi\fP
2786 .\"
2787 documentation.
2788 .P
2789 Experiments with Perl suggest that it too has similar optimizations, sometimes
2790 leading to anomalous results.
2791 .
2792 .
2793 .SS "Verbs that act immediately"
2794 .rs
2795 .sp
2796 The following verbs act as soon as they are encountered. They may not be
2797 followed by a name.
2798 .sp
2799 (*ACCEPT)
2800 .sp
2801 This verb causes the match to end successfully, skipping the remainder of the
2802 pattern. However, when it is inside a subpattern that is called as a
2803 subroutine, only that subpattern is ended successfully. Matching then continues
2804 at the outer level. If (*ACCEPT) in triggered in a positive assertion, the
2805 assertion succeeds; in a negative assertion, the assertion fails.
2806 .P
2807 If (*ACCEPT) is inside capturing parentheses, the data so far is captured. For
2808 example:
2809 .sp
2810 A((?:A|B(*ACCEPT)|C)D)
2811 .sp
2812 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2813 the outer parentheses.
2814 .sp
2815 (*FAIL) or (*F)
2816 .sp
2817 This verb causes a matching failure, forcing backtracking to occur. It is
2818 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2819 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2820 Perl features that are not present in PCRE. The nearest equivalent is the
2821 callout feature, as for example in this pattern:
2822 .sp
2823 a+(?C)(*FAIL)
2824 .sp
2825 A match with the string "aaaa" always fails, but the callout is taken before
2826 each backtrack happens (in this example, 10 times).
2827 .
2828 .
2829 .SS "Recording which path was taken"
2830 .rs
2831 .sp
2832 There is one verb whose main purpose is to track how a match was arrived at,
2833 though it also has a secondary use in conjunction with advancing the match
2834 starting point (see (*SKIP) below).
2835 .sp
2836 (*MARK:NAME) or (*:NAME)
2837 .sp
2838 A name is always required with this verb. There may be as many instances of
2839 (*MARK) as you like in a pattern, and their names do not have to be unique.
2840 .P
2841 When a match succeeds, the name of the last-encountered (*MARK:NAME),
2842 (*PRUNE:NAME), or (*THEN:NAME) on the matching path is passed back to the
2843 caller as described in the section entitled
2844 .\" HTML <a href="pcreapi.html#extradata">
2845 .\" </a>
2846 "Extra data for \fBpcre_exec()\fP"
2847 .\"
2848 in the
2849 .\" HREF
2850 \fBpcreapi\fP
2851 .\"
2852 documentation. Here is an example of \fBpcretest\fP output, where the /K
2853 modifier requests the retrieval and outputting of (*MARK) data:
2854 .sp
2855 re> /X(*MARK:A)Y|X(*MARK:B)Z/K
2856 data> XY
2857 0: XY
2858 MK: A
2859 XZ
2860 0: XZ
2861 MK: B
2862 .sp
2863 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2864 indicates which of the two alternatives matched. This is a more efficient way
2865 of obtaining this information than putting each alternative in its own
2866 capturing parentheses.
2867 .P
2868 If a verb with a name is encountered in a positive assertion that is true, the
2869 name is recorded and passed back if it is the last-encountered. This does not
2870 happen for negative assertions or failing positive assertions.
2871 .P
2872 After a partial match or a failed match, the last encountered name in the
2873 entire match process is returned. For example:
2874 .sp
2875 re> /X(*MARK:A)Y|X(*MARK:B)Z/K
2876 data> XP
2877 No match, mark = B
2878 .sp
2879 Note that in this unanchored example the mark is retained from the match
2880 attempt that started at the letter "X" in the subject. Subsequent match
2881 attempts starting at "P" and then with an empty string do not get as far as the
2882 (*MARK) item, but nevertheless do not reset it.
2883 .P
2884 If you are interested in (*MARK) values after failed matches, you should
2885 probably set the PCRE_NO_START_OPTIMIZE option
2886 .\" HTML <a href="#nooptimize">
2887 .\" </a>
2888 (see above)
2889 .\"
2890 to ensure that the match is always attempted.
2891 .
2892 .
2893 .SS "Verbs that act after backtracking"
2894 .rs
2895 .sp
2896 The following verbs do nothing when they are encountered. Matching continues
2897 with what follows, but if there is no subsequent match, causing a backtrack to
2898 the verb, a failure is forced. That is, backtracking cannot pass to the left of
2899 the verb. However, when one of these verbs appears inside an atomic group or an
2900 assertion that is true, its effect is confined to that group, because once the
2901 group has been matched, there is never any backtracking into it. In this
2902 situation, backtracking can "jump back" to the left of the entire atomic group
2903 or assertion. (Remember also, as stated above, that this localization also
2904 applies in subroutine calls.)
2905 .P
2906 These verbs differ in exactly what kind of failure occurs when backtracking
2907 reaches them. The behaviour described below is what happens when the verb is
2908 not in a subroutine or an assertion. Subsequent sections cover these special
2909 cases.
2910 .sp
2911 (*COMMIT)
2912 .sp
2913 This verb, which may not be followed by a name, causes the whole match to fail
2914 outright if there is a later matching failure that causes backtracking to reach
2915 it. Even if the pattern is unanchored, no further attempts to find a match by
2916 advancing the starting point take place. If (*COMMIT) is the only backtracking
2917 verb that is encountered, once it has been passed \fBpcre_exec()\fP is
2918 committed to finding a match at the current starting point, or not at all. For
2919 example:
2920 .sp
2921 a+(*COMMIT)b
2922 .sp
2923 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2924 dynamic anchor, or "I've started, so I must finish." The name of the most
2925 recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2926 match failure.
2927 .P
2928 If there is more than one backtracking verb in a pattern, a different one that
2929 follows (*COMMIT) may be triggered first, so merely passing (*COMMIT) during a
2930 match does not always guarantee that a match must be at this starting point.
2931 .P
2932 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2933 unless PCRE's start-of-match optimizations are turned off, as shown in this
2934 \fBpcretest\fP example:
2935 .sp
2936 re> /(*COMMIT)abc/
2937 data> xyzabc
2938 0: abc
2939 xyzabc\eY
2940 No match
2941 .sp
2942 PCRE knows that any match must start with "a", so the optimization skips along
2943 the subject to "a" before running the first match attempt, which succeeds. When
2944 the optimization is disabled by the \eY escape in the second subject, the match
2945 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2946 starting points.
2947 .sp
2948 (*PRUNE) or (*PRUNE:NAME)
2949 .sp
2950 This verb causes the match to fail at the current starting position in the
2951 subject if there is a later matching failure that causes backtracking to reach
2952 it. If the pattern is unanchored, the normal "bumpalong" advance to the next
2953 starting character then happens. Backtracking can occur as usual to the left of
2954 (*PRUNE), before it is reached, or when matching to the right of (*PRUNE), but
2955 if there is no match to the right, backtracking cannot cross (*PRUNE). In
2956 simple cases, the use of (*PRUNE) is just an alternative to an atomic group or
2957 possessive quantifier, but there are some uses of (*PRUNE) that cannot be
2958 expressed in any other way. In an anchored pattern (*PRUNE) has the same effect
2959 as (*COMMIT).
2960 .P
2961 The behaviour of (*PRUNE:NAME) is the not the same as (*MARK:NAME)(*PRUNE).
2962 It is like (*MARK:NAME) in that the name is remembered for passing back to the
2963 caller. However, (*SKIP:NAME) searches only for names set with (*MARK).
2964 .sp
2965 (*SKIP)
2966 .sp
2967 This verb, when given without a name, is like (*PRUNE), except that if the
2968 pattern is unanchored, the "bumpalong" advance is not to the next character,
2969 but to the position in the subject where (*SKIP) was encountered. (*SKIP)
2970 signifies that whatever text was matched leading up to it cannot be part of a
2971 successful match. Consider:
2972 .sp
2973 a+(*SKIP)b
2974 .sp
2975 If the subject is "aaaac...", after the first match attempt fails (starting at
2976 the first character in the string), the starting point skips on to start the
2977 next attempt at "c". Note that a possessive quantifer does not have the same
2978 effect as this example; although it would suppress backtracking during the
2979 first match attempt, the second attempt would start at the second character
2980 instead of skipping on to "c".
2981 .sp
2982 (*SKIP:NAME)
2983 .sp
2984 When (*SKIP) has an associated name, its behaviour is modified. When it is
2985 triggered, the previous path through the pattern is searched for the most
2986 recent (*MARK) that has the same name. If one is found, the "bumpalong" advance
2987 is to the subject position that corresponds to that (*MARK) instead of to where
2988 (*SKIP) was encountered. If no (*MARK) with a matching name is found, the
2989 (*SKIP) is ignored.
2990 .P
2991 Note that (*SKIP:NAME) searches only for names set by (*MARK:NAME). It ignores
2992 names that are set by (*PRUNE:NAME) or (*THEN:NAME).
2993 .sp
2994 (*THEN) or (*THEN:NAME)
2995 .sp
2996 This verb causes a skip to the next innermost alternative when backtracking
2997 reaches it. That is, it cancels any further backtracking within the current
2998 alternative. Its name comes from the observation that it can be used for a
2999 pattern-based if-then-else block:
3000 .sp
3001 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
3002 .sp
3003 If the COND1 pattern matches, FOO is tried (and possibly further items after
3004 the end of the group if FOO succeeds); on failure, the matcher skips to the
3005 second alternative and tries COND2, without backtracking into COND1. If that
3006 succeeds and BAR fails, COND3 is tried. If subsequently BAZ fails, there are no
3007 more alternatives, so there is a backtrack to whatever came before the entire
3008 group. If (*THEN) is not inside an alternation, it acts like (*PRUNE).
3009 .P
3010 The behaviour of (*THEN:NAME) is the not the same as (*MARK:NAME)(*THEN).
3011 It is like (*MARK:NAME) in that the name is remembered for passing back to the
3012 caller. However, (*SKIP:NAME) searches only for names set with (*MARK).
3013 .P
3014 A subpattern that does not contain a | character is just a part of the
3015 enclosing alternative; it is not a nested alternation with only one
3016 alternative. The effect of (*THEN) extends beyond such a subpattern to the
3017 enclosing alternative. Consider this pattern, where A, B, etc. are complex
3018 pattern fragments that do not contain any | characters at this level:
3019 .sp
3020 A (B(*THEN)C) | D
3021 .sp
3022 If A and B are matched, but there is a failure in C, matching does not
3023 backtrack into A; instead it moves to the next alternative, that is, D.
3024 However, if the subpattern containing (*THEN) is given an alternative, it
3025 behaves differently:
3026 .sp
3027 A (B(*THEN)C | (*FAIL)) | D
3028 .sp
3029 The effect of (*THEN) is now confined to the inner subpattern. After a failure
3030 in C, matching moves to (*FAIL), which causes the whole subpattern to fail
3031 because there are no more alternatives to try. In this case, matching does now
3032 backtrack into A.
3033 .P
3034 Note that a conditional subpattern is not considered as having two
3035 alternatives, because only one is ever used. In other words, the | character in
3036 a conditional subpattern has a different meaning. Ignoring white space,
3037 consider:
3038 .sp
3039 ^.*? (?(?=a) a | b(*THEN)c )
3040 .sp
3041 If the subject is "ba", this pattern does not match. Because .*? is ungreedy,
3042 it initially matches zero characters. The condition (?=a) then fails, the
3043 character "b" is matched, but "c" is not. At this point, matching does not
3044 backtrack to .*? as might perhaps be expected from the presence of the |
3045 character. The conditional subpattern is part of the single alternative that
3046 comprises the whole pattern, and so the match fails. (If there was a backtrack
3047 into .*?, allowing it to match "b", the match would succeed.)
3048 .P
3049 The verbs just described provide four different "strengths" of control when
3050 subsequent matching fails. (*THEN) is the weakest, carrying on the match at the
3051 next alternative. (*PRUNE) comes next, failing the match at the current
3052 starting position, but allowing an advance to the next character (for an
3053 unanchored pattern). (*SKIP) is similar, except that the advance may be more
3054 than one character. (*COMMIT) is the strongest, causing the entire match to
3055 fail.
3056 .
3057 .
3058 .SS "More than one backtracking verb"
3059 .rs
3060 .sp
3061 If more than one backtracking verb is present in a pattern, the one that is
3062 backtracked onto first acts. For example, consider this pattern, where A, B,
3063 etc. are complex pattern fragments:
3064 .sp
3066 .sp
3067 If A matches but B fails, the backtrack to (*COMMIT) causes the entire match to
3068 fail. However, if A and B match, but C fails, the backtrack to (*THEN) causes
3069 the next alternative (ABD) to be tried. This behaviour is consistent, but is
3070 not always the same as Perl's. It means that if two or more backtracking verbs
3071 appear in succession, all the the last of them has no effect. Consider this
3072 example:
3073 .sp
3074 ...(*COMMIT)(*PRUNE)...
3075 .sp
3076 If there is a matching failure to the right, backtracking onto (*PRUNE) causes
3077 it to be triggered, and its action is taken. There can never be a backtrack
3078 onto (*COMMIT).
3079 .
3080 .
3081 .\" HTML <a name="btrepeat"></a>
3082 .SS "Backtracking verbs in repeated groups"
3083 .rs
3084 .sp
3085 PCRE differs from Perl in its handling of backtracking verbs in repeated
3086 groups. For example, consider:
3087 .sp
3088 /(a(*COMMIT)b)+ac/
3089 .sp
3090 If the subject is "abac", Perl matches, but PCRE fails because the (*COMMIT) in
3091 the second repeat of the group acts.
3092 .
3093 .
3094 .\" HTML <a name="btassert"></a>
3095 .SS "Backtracking verbs in assertions"
3096 .rs
3097 .sp
3098 (*FAIL) in an assertion has its normal effect: it forces an immediate backtrack.
3099 .P
3100 (*ACCEPT) in a positive assertion causes the assertion to succeed without any
3101 further processing. In a negative assertion, (*ACCEPT) causes the assertion to
3102 fail without any further processing.
3103 .P
3104 The other backtracking verbs are not treated specially if they appear in a
3105 positive assertion. In particular, (*THEN) skips to the next alternative in the
3106 innermost enclosing group that has alternations, whether or not this is within
3107 the assertion.
3108 .P
3109 Negative assertions are, however, different, in order to ensure that changing a
3110 positive assertion into a negative assertion changes its result. Backtracking
3111 into (*COMMIT), (*SKIP), or (*PRUNE) causes a negative assertion to be true,
3112 without considering any further alternative branches in the assertion.
3113 Backtracking into (*THEN) causes it to skip to the next enclosing alternative
3114 within the assertion (the normal behaviour), but if the assertion does not have
3115 such an alternative, (*THEN) behaves like (*PRUNE).
3116 .
3117 .
3118 .\" HTML <a name="btsub"></a>
3119 .SS "Backtracking verbs in subroutines"
3120 .rs
3121 .sp
3122 These behaviours occur whether or not the subpattern is called recursively.
3123 Perl's treatment of subroutines is different in some cases.
3124 .P
3125 (*FAIL) in a subpattern called as a subroutine has its normal effect: it forces
3126 an immediate backtrack.
3127 .P
3128 (*ACCEPT) in a subpattern called as a subroutine causes the subroutine match to
3129 succeed without any further processing. Matching then continues after the
3130 subroutine call.
3131 .P
3132 (*COMMIT), (*SKIP), and (*PRUNE) in a subpattern called as a subroutine cause
3133 the subroutine match to fail.
3134 .P
3135 (*THEN) skips to the next alternative in the innermost enclosing group within
3136 the subpattern that has alternatives. If there is no such group within the
3137 subpattern, (*THEN) causes the subroutine match to fail.
3138 .
3139 .
3140 .SH "SEE ALSO"
3141 .rs
3142 .sp
3143 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3),
3144 \fBpcresyntax\fP(3), \fBpcre\fP(3), \fBpcre16(3)\fP, \fBpcre32(3)\fP.
3145 .
3146 .
3148 .rs
3149 .sp
3150 .nf
3151 Philip Hazel
3152 University Computing Service
3153 Cambridge CB2 3QH, England.
3154 .fi
3155 .
3156 .
3158 .rs
3159 .sp
3160 .nf
3161 Last updated: 08 October 2013
3162 Copyright (c) 1997-2013 University of Cambridge.
3163 .fi


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

  ViewVC Help
Powered by ViewVC 1.1.5