ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 1668 - (show annotations)
Fri Oct 14 17:06:22 2016 UTC (8 days, 8 hours ago) by ph10
File size: 136108 byte(s)
Error occurred while calculating annotation data.
Fix documentation error.
1 .TH PCREPATTERN 3 "14 October 2016" "PCRE 8.40"
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 (*UCP).
84 This has the same effect as setting the PCRE_UCP option: it causes sequences
85 such as \ed and \ew to use Unicode properties to determine character types,
86 instead of recognizing only characters with codes less than 128 via a lookup
87 table.
88 .
89 .
90 .SS "Disabling auto-possessification"
91 .rs
92 .sp
93 If a pattern starts with (*NO_AUTO_POSSESS), it has the same effect as setting
94 the PCRE_NO_AUTO_POSSESS option at compile time. This stops PCRE from making
95 quantifiers possessive when what follows cannot match the repeated item. For
96 example, by default a+b is treated as a++b. For more details, see the
97 .\" HREF
98 \fBpcreapi\fP
99 .\"
100 documentation.
101 .
102 .
103 .SS "Disabling start-up optimizations"
104 .rs
105 .sp
106 If a pattern starts with (*NO_START_OPT), it has the same effect as setting the
107 PCRE_NO_START_OPTIMIZE option either at compile or matching time. This disables
108 several optimizations for quickly reaching "no match" results. For more
109 details, see the
110 .\" HREF
111 \fBpcreapi\fP
112 .\"
113 documentation.
114 .
115 .
116 .\" HTML <a name="newlines"></a>
117 .SS "Newline conventions"
118 .rs
119 .sp
120 PCRE supports five different conventions for indicating line breaks in
121 strings: a single CR (carriage return) character, a single LF (linefeed)
122 character, the two-character sequence CRLF, any of the three preceding, or any
123 Unicode newline sequence. The
124 .\" HREF
125 \fBpcreapi\fP
126 .\"
127 page has
128 .\" HTML <a href="pcreapi.html#newlines">
129 .\" </a>
130 further discussion
131 .\"
132 about newlines, and shows how to set the newline convention in the
133 \fIoptions\fP arguments for the compiling and matching functions.
134 .P
135 It is also possible to specify a newline convention by starting a pattern
136 string with one of the following five sequences:
137 .sp
138 (*CR) carriage return
139 (*LF) linefeed
140 (*CRLF) carriage return, followed by linefeed
141 (*ANYCRLF) any of the three above
142 (*ANY) all Unicode newline sequences
143 .sp
144 These override the default and the options given to the compiling function. For
145 example, on a Unix system where LF is the default newline sequence, the pattern
146 .sp
147 (*CR)a.b
148 .sp
149 changes the convention to CR. That pattern matches "a\enb" because LF is no
150 longer a newline. If more than one of these settings is present, the last one
151 is used.
152 .P
153 The newline convention affects where the circumflex and dollar assertions are
154 true. It also affects the interpretation of the dot metacharacter when
155 PCRE_DOTALL is not set, and the behaviour of \eN. However, it does not affect
156 what the \eR escape sequence matches. By default, this is any Unicode newline
157 sequence, for Perl compatibility. However, this can be changed; see the
158 description of \eR in the section entitled
159 .\" HTML <a href="#newlineseq">
160 .\" </a>
161 "Newline sequences"
162 .\"
163 below. A change of \eR setting can be combined with a change of newline
164 convention.
165 .
166 .
167 .SS "Setting match and recursion limits"
168 .rs
169 .sp
170 The caller of \fBpcre_exec()\fP can set a limit on the number of times the
171 internal \fBmatch()\fP function is called and on the maximum depth of
172 recursive calls. These facilities are provided to catch runaway matches that
173 are provoked by patterns with huge matching trees (a typical example is a
174 pattern with nested unlimited repeats) and to avoid running out of system stack
175 by too much recursion. When one of these limits is reached, \fBpcre_exec()\fP
176 gives an error return. The limits can also be set by items at the start of the
177 pattern of the form
178 .sp
179 (*LIMIT_MATCH=d)
181 .sp
182 where d is any number of decimal digits. However, the value of the setting must
183 be less than the value set (or defaulted) by the caller of \fBpcre_exec()\fP
184 for it to have any effect. In other words, the pattern writer can lower the
185 limits set by the programmer, but not raise them. If there is more than one
186 setting of one of these limits, the lower value is used.
187 .
188 .
190 .rs
191 .sp
192 PCRE can be compiled to run in an environment that uses EBCDIC as its character
193 code rather than ASCII or Unicode (typically a mainframe system). In the
194 sections below, character code values are ASCII or Unicode; in an EBCDIC
195 environment these characters may have different code values, and there are no
196 code points greater than 255.
197 .
198 .
200 .rs
201 .sp
202 A regular expression is a pattern that is matched against a subject string from
203 left to right. Most characters stand for themselves in a pattern, and match the
204 corresponding characters in the subject. As a trivial example, the pattern
205 .sp
206 The quick brown fox
207 .sp
208 matches a portion of a subject string that is identical to itself. When
209 caseless matching is specified (the PCRE_CASELESS option), letters are matched
210 independently of case. In a UTF mode, PCRE always understands the concept of
211 case for characters whose values are less than 128, so caseless matching is
212 always possible. For characters with higher values, the concept of case is
213 supported if PCRE is compiled with Unicode property support, but not otherwise.
214 If you want to use caseless matching for characters 128 and above, you must
215 ensure that PCRE is compiled with Unicode property support as well as with
216 UTF support.
217 .P
218 The power of regular expressions comes from the ability to include alternatives
219 and repetitions in the pattern. These are encoded in the pattern by the use of
220 \fImetacharacters\fP, which do not stand for themselves but instead are
221 interpreted in some special way.
222 .P
223 There are two different sets of metacharacters: those that are recognized
224 anywhere in the pattern except within square brackets, and those that are
225 recognized within square brackets. Outside square brackets, the metacharacters
226 are as follows:
227 .sp
228 \e general escape character with several uses
229 ^ assert start of string (or line, in multiline mode)
230 $ assert end of string (or line, in multiline mode)
231 . match any character except newline (by default)
232 [ start character class definition
233 | start of alternative branch
234 ( start subpattern
235 ) end subpattern
236 ? extends the meaning of (
237 also 0 or 1 quantifier
238 also quantifier minimizer
239 * 0 or more quantifier
240 + 1 or more quantifier
241 also "possessive quantifier"
242 { start min/max quantifier
243 .sp
244 Part of a pattern that is in square brackets is called a "character class". In
245 a character class the only metacharacters are:
246 .sp
247 \e general escape character
248 ^ negate the class, but only if the first character
249 - indicates character range
250 .\" JOIN
251 [ POSIX character class (only if followed by POSIX
252 syntax)
253 ] terminates the character class
254 .sp
255 The following sections describe the use of each of the metacharacters.
256 .
257 .
259 .rs
260 .sp
261 The backslash character has several uses. Firstly, if it is followed by a
262 character that is not a number or a letter, it takes away any special meaning
263 that character may have. This use of backslash as an escape character applies
264 both inside and outside character classes.
265 .P
266 For example, if you want to match a * character, you write \e* in the pattern.
267 This escaping action applies whether or not the following character would
268 otherwise be interpreted as a metacharacter, so it is always safe to precede a
269 non-alphanumeric with backslash to specify that it stands for itself. In
270 particular, if you want to match a backslash, you write \e\e.
271 .P
272 In a UTF mode, only ASCII numbers and letters have any special meaning after a
273 backslash. All other characters (in particular, those whose codepoints are
274 greater than 127) are treated as literals.
275 .P
276 If a pattern is compiled with the PCRE_EXTENDED option, most white space in the
277 pattern (other than in a character class), and characters between a # outside a
278 character class and the next newline, inclusive, are ignored. An escaping
279 backslash can be used to include a white space or # character as part of the
280 pattern.
281 .P
282 If you want to remove the special meaning from a sequence of characters, you
283 can do so by putting them between \eQ and \eE. This is different from Perl in
284 that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
285 Perl, $ and @ cause variable interpolation. Note the following examples:
286 .sp
287 Pattern PCRE matches Perl matches
288 .sp
289 .\" JOIN
290 \eQabc$xyz\eE abc$xyz abc followed by the
291 contents of $xyz
292 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
293 \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
294 .sp
295 The \eQ...\eE sequence is recognized both inside and outside character classes.
296 An isolated \eE that is not preceded by \eQ is ignored. If \eQ is not followed
297 by \eE later in the pattern, the literal interpretation continues to the end of
298 the pattern (that is, \eE is assumed at the end). If the isolated \eQ is inside
299 a character class, this causes an error, because the character class is not
300 terminated.
301 .
302 .
303 .\" HTML <a name="digitsafterbackslash"></a>
304 .SS "Non-printing characters"
305 .rs
306 .sp
307 A second use of backslash provides a way of encoding non-printing characters
308 in patterns in a visible manner. There is no restriction on the appearance of
309 non-printing characters, apart from the binary zero that terminates a pattern,
310 but when a pattern is being prepared by text editing, it is often easier to use
311 one of the following escape sequences than the binary character it represents.
312 In an ASCII or Unicode environment, these escapes are as follows:
313 .sp
314 \ea alarm, that is, the BEL character (hex 07)
315 \ecx "control-x", where x is any ASCII character
316 \ee escape (hex 1B)
317 \ef form feed (hex 0C)
318 \en linefeed (hex 0A)
319 \er carriage return (hex 0D)
320 \et tab (hex 09)
321 \e0dd character with octal code 0dd
322 \eddd character with octal code ddd, or back reference
323 \eo{ddd..} character with octal code ddd..
324 \exhh character with hex code hh
325 \ex{hhh..} character with hex code hhh.. (non-JavaScript mode)
326 \euhhhh character with hex code hhhh (JavaScript mode only)
327 .sp
328 The precise effect of \ecx on ASCII characters is as follows: if x is a lower
329 case letter, it is converted to upper case. Then bit 6 of the character (hex
330 40) is inverted. Thus \ecA to \ecZ become hex 01 to hex 1A (A is 41, Z is 5A),
331 but \ec{ becomes hex 3B ({ is 7B), and \ec; becomes hex 7B (; is 3B). If the
332 data item (byte or 16-bit value) following \ec has a value greater than 127, a
333 compile-time error occurs. This locks out non-ASCII characters in all modes.
334 .P
335 When PCRE is compiled in EBCDIC mode, \ea, \ee, \ef, \en, \er, and \et
336 generate the appropriate EBCDIC code values. The \ec escape is processed
337 as specified for Perl in the \fBperlebcdic\fP document. The only characters
338 that are allowed after \ec are A-Z, a-z, or one of @, [, \e, ], ^, _, or ?. Any
339 other character provokes a compile-time error. The sequence \ec@ encodes
340 character code 0; after \ec the letters (in either case) encode characters 1-26
341 (hex 01 to hex 1A); [, \e, ], ^, and _ encode characters 27-31 (hex 1B to hex
342 1F), and \ec? becomes either 255 (hex FF) or 95 (hex 5F).
343 .P
344 Thus, apart from \ec?, these escapes generate the same character code values as
345 they do in an ASCII environment, though the meanings of the values mostly
346 differ. For example, \ecG always generates code value 7, which is BEL in ASCII
347 but DEL in EBCDIC.
348 .P
349 The sequence \ec? generates DEL (127, hex 7F) in an ASCII environment, but
350 because 127 is not a control character in EBCDIC, Perl makes it generate the
351 APC character. Unfortunately, there are several variants of EBCDIC. In most of
352 them the APC character has the value 255 (hex FF), but in the one Perl calls
353 POSIX-BC its value is 95 (hex 5F). If certain other characters have POSIX-BC
354 values, PCRE makes \ec? generate 95; otherwise it generates 255.
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\e015
358 specifies two binary zeros followed by a CR character (code value 13). 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 escape \eo must be followed by a sequence of octal digits, enclosed in
363 braces. An error occurs if this is not the case. This escape is a recent
364 addition to Perl; it provides way of specifying character code points as octal
365 numbers greater than 0777, and it also allows octal numbers and back references
366 to be unambiguously specified.
367 .P
368 For greater clarity and unambiguity, it is best to avoid following \e by a
369 digit greater than zero. Instead, use \eo{} or \ex{} to specify character
370 numbers, and \eg{} to specify back references. The following paragraphs
371 describe the old, ambiguous syntax.
372 .P
373 The handling of a backslash followed by a digit other than 0 is complicated,
374 and Perl has changed in recent releases, causing PCRE also to change. Outside a
375 character class, PCRE reads the digit and any following digits as a decimal
376 number. If the number is less than 8, or if there have been at least that many
377 previous capturing left parentheses in the expression, the entire sequence is
378 taken as a \fIback reference\fP. A description of how this works is given
379 .\" HTML <a href="#backreferences">
380 .\" </a>
381 later,
382 .\"
383 following the discussion of
384 .\" HTML <a href="#subpattern">
385 .\" </a>
386 parenthesized subpatterns.
387 .\"
388 .P
389 Inside a character class, or if the decimal number following \e is greater than
390 7 and there have not been that many capturing subpatterns, PCRE handles \e8 and
391 \e9 as the literal characters "8" and "9", and otherwise re-reads up to three
392 octal digits following the backslash, using them to generate a data character.
393 Any subsequent digits stand for themselves. For example:
394 .sp
395 \e040 is another way of writing an ASCII space
396 .\" JOIN
397 \e40 is the same, provided there are fewer than 40
398 previous capturing subpatterns
399 \e7 is always a back reference
400 .\" JOIN
401 \e11 might be a back reference, or another way of
402 writing a tab
403 \e011 is always a tab
404 \e0113 is a tab followed by the character "3"
405 .\" JOIN
406 \e113 might be a back reference, otherwise the
407 character with octal code 113
408 .\" JOIN
409 \e377 might be a back reference, otherwise
410 the value 255 (decimal)
411 .\" JOIN
412 \e81 is either a back reference, or the two
413 characters "8" and "1"
414 .sp
415 Note that octal values of 100 or greater that are specified using this syntax
416 must not be introduced by a leading zero, because no more than three octal
417 digits are ever read.
418 .P
419 By default, after \ex that is not followed by {, from zero to two hexadecimal
420 digits are read (letters can be in upper or lower case). Any number of
421 hexadecimal digits may appear between \ex{ and }. If a character other than
422 a hexadecimal digit appears between \ex{ and }, or if there is no terminating
423 }, an error occurs.
424 .P
425 If the PCRE_JAVASCRIPT_COMPAT option is set, the interpretation of \ex is
426 as just described only when it is followed by two hexadecimal digits.
427 Otherwise, it matches a literal "x" character. In JavaScript mode, support for
428 code points greater than 256 is provided by \eu, which must be followed by
429 four hexadecimal digits; otherwise it matches a literal "u" character.
430 .P
431 Characters whose value is less than 256 can be defined by either of the two
432 syntaxes for \ex (or by \eu in JavaScript mode). There is no difference in the
433 way they are handled. For example, \exdc is exactly the same as \ex{dc} (or
434 \eu00dc in JavaScript mode).
435 .
436 .
437 .SS "Constraints on character values"
438 .rs
439 .sp
440 Characters that are specified using octal or hexadecimal numbers are
441 limited to certain values, as follows:
442 .sp
443 8-bit non-UTF mode less than 0x100
444 8-bit UTF-8 mode less than 0x10ffff and a valid codepoint
445 16-bit non-UTF mode less than 0x10000
446 16-bit UTF-16 mode less than 0x10ffff and a valid codepoint
447 32-bit non-UTF mode less than 0x100000000
448 32-bit UTF-32 mode less than 0x10ffff and a valid codepoint
449 .sp
450 Invalid Unicode codepoints are the range 0xd800 to 0xdfff (the so-called
451 "surrogate" codepoints), and 0xffef.
452 .
453 .
454 .SS "Escape sequences in character classes"
455 .rs
456 .sp
457 All the sequences that define a single character value can be used both inside
458 and outside character classes. In addition, inside a character class, \eb is
459 interpreted as the backspace character (hex 08).
460 .P
461 \eN is not allowed in a character class. \eB, \eR, and \eX are not special
462 inside a character class. Like other unrecognized escape sequences, they are
463 treated as the literal characters "B", "R", and "X" by default, but cause an
464 error if the PCRE_EXTRA option is set. Outside a character class, these
465 sequences have different meanings.
466 .
467 .
468 .SS "Unsupported escape sequences"
469 .rs
470 .sp
471 In Perl, the sequences \el, \eL, \eu, and \eU are recognized by its string
472 handler and used to modify the case of following characters. By default, PCRE
473 does not support these escape sequences. However, if the PCRE_JAVASCRIPT_COMPAT
474 option is set, \eU matches a "U" character, and \eu can be used to define a
475 character by code point, as described in the previous section.
476 .
477 .
478 .SS "Absolute and relative back references"
479 .rs
480 .sp
481 The sequence \eg followed by an unsigned or a negative number, optionally
482 enclosed in braces, is an absolute or relative back reference. A named back
483 reference can be coded as \eg{name}. Back references are discussed
484 .\" HTML <a href="#backreferences">
485 .\" </a>
486 later,
487 .\"
488 following the discussion of
489 .\" HTML <a href="#subpattern">
490 .\" </a>
491 parenthesized subpatterns.
492 .\"
493 .
494 .
495 .SS "Absolute and relative subroutine calls"
496 .rs
497 .sp
498 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
499 a number enclosed either in angle brackets or single quotes, is an alternative
500 syntax for referencing a subpattern as a "subroutine". Details are discussed
501 .\" HTML <a href="#onigurumasubroutines">
502 .\" </a>
503 later.
504 .\"
505 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
506 synonymous. The former is a back reference; the latter is a
507 .\" HTML <a href="#subpatternsassubroutines">
508 .\" </a>
509 subroutine
510 .\"
511 call.
512 .
513 .
514 .\" HTML <a name="genericchartypes"></a>
515 .SS "Generic character types"
516 .rs
517 .sp
518 Another use of backslash is for specifying generic character types:
519 .sp
520 \ed any decimal digit
521 \eD any character that is not a decimal digit
522 \eh any horizontal white space character
523 \eH any character that is not a horizontal white space character
524 \es any white space character
525 \eS any character that is not a white space character
526 \ev any vertical white space character
527 \eV any character that is not a vertical white space character
528 \ew any "word" character
529 \eW any "non-word" character
530 .sp
531 There is also the single sequence \eN, which matches a non-newline character.
532 This is the same as
533 .\" HTML <a href="#fullstopdot">
534 .\" </a>
535 the "." metacharacter
536 .\"
537 when PCRE_DOTALL is not set. Perl also uses \eN to match characters by name;
538 PCRE does not support this.
539 .P
540 Each pair of lower and upper case escape sequences partitions the complete set
541 of characters into two disjoint sets. Any given character matches one, and only
542 one, of each pair. The sequences can appear both inside and outside character
543 classes. They each match one character of the appropriate type. If the current
544 matching point is at the end of the subject string, all of them fail, because
545 there is no character to match.
546 .P
547 For compatibility with Perl, \es did not used to match the VT character (code
548 11), which made it different from the the POSIX "space" class. However, Perl
549 added VT at release 5.18, and PCRE followed suit at release 8.34. The default
550 \es characters are now HT (9), LF (10), VT (11), FF (12), CR (13), and space
551 (32), which are defined as white space in the "C" locale. This list may vary if
552 locale-specific matching is taking place. For example, in some locales the
553 "non-breaking space" character (\exA0) is recognized as white space, and in
554 others the VT character is not.
555 .P
556 A "word" character is an underscore or any character that is a letter or digit.
557 By default, the definition of letters and digits is controlled by PCRE's
558 low-valued character tables, and may vary if locale-specific matching is taking
559 place (see
560 .\" HTML <a href="pcreapi.html#localesupport">
561 .\" </a>
562 "Locale support"
563 .\"
564 in the
565 .\" HREF
566 \fBpcreapi\fP
567 .\"
568 page). For example, in a French locale such as "fr_FR" in Unix-like systems,
569 or "french" in Windows, some character codes greater than 127 are used for
570 accented letters, and these are then matched by \ew. The use of locales with
571 Unicode is discouraged.
572 .P
573 By default, characters whose code points are greater than 127 never match \ed,
574 \es, or \ew, and always match \eD, \eS, and \eW, although this may vary for
575 characters in the range 128-255 when locale-specific matching is happening.
576 These escape sequences retain their original meanings from before Unicode
577 support was available, mainly for efficiency reasons. If PCRE is compiled with
578 Unicode property support, and the PCRE_UCP option is set, the behaviour is
579 changed so that Unicode properties are used to determine character types, as
580 follows:
581 .sp
582 \ed any character that matches \ep{Nd} (decimal digit)
583 \es any character that matches \ep{Z} or \eh or \ev
584 \ew any character that matches \ep{L} or \ep{N}, plus underscore
585 .sp
586 The upper case escapes match the inverse sets of characters. Note that \ed
587 matches only decimal digits, whereas \ew matches any Unicode digit, as well as
588 any Unicode letter, and underscore. Note also that PCRE_UCP affects \eb, and
589 \eB because they are defined in terms of \ew and \eW. Matching these sequences
590 is noticeably slower when PCRE_UCP is set.
591 .P
592 The sequences \eh, \eH, \ev, and \eV are features that were added to Perl at
593 release 5.10. In contrast to the other sequences, which match only ASCII
594 characters by default, these always match certain high-valued code points,
595 whether or not PCRE_UCP is set. The horizontal space characters are:
596 .sp
597 U+0009 Horizontal tab (HT)
598 U+0020 Space
599 U+00A0 Non-break space
600 U+1680 Ogham space mark
601 U+180E Mongolian vowel separator
602 U+2000 En quad
603 U+2001 Em quad
604 U+2002 En space
605 U+2003 Em space
606 U+2004 Three-per-em space
607 U+2005 Four-per-em space
608 U+2006 Six-per-em space
609 U+2007 Figure space
610 U+2008 Punctuation space
611 U+2009 Thin space
612 U+200A Hair space
613 U+202F Narrow no-break space
614 U+205F Medium mathematical space
615 U+3000 Ideographic space
616 .sp
617 The vertical space characters are:
618 .sp
619 U+000A Linefeed (LF)
620 U+000B Vertical tab (VT)
621 U+000C Form feed (FF)
622 U+000D Carriage return (CR)
623 U+0085 Next line (NEL)
624 U+2028 Line separator
625 U+2029 Paragraph separator
626 .sp
627 In 8-bit, non-UTF-8 mode, only the characters with codepoints less than 256 are
628 relevant.
629 .
630 .
631 .\" HTML <a name="newlineseq"></a>
632 .SS "Newline sequences"
633 .rs
634 .sp
635 Outside a character class, by default, the escape sequence \eR matches any
636 Unicode newline sequence. In 8-bit non-UTF-8 mode \eR is equivalent to the
637 following:
638 .sp
639 (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
640 .sp
641 This is an example of an "atomic group", details of which are given
642 .\" HTML <a href="#atomicgroup">
643 .\" </a>
644 below.
645 .\"
646 This particular group matches either the two-character sequence CR followed by
647 LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
648 U+000B), FF (form feed, U+000C), CR (carriage return, U+000D), or NEL (next
649 line, U+0085). The two-character sequence is treated as a single unit that
650 cannot be split.
651 .P
652 In other modes, two additional characters whose codepoints are greater than 255
653 are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
654 Unicode character property support is not needed for these characters to be
655 recognized.
656 .P
657 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
658 complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
659 either at compile time or when the pattern is matched. (BSR is an abbrevation
660 for "backslash R".) This can be made the default when PCRE is built; if this is
661 the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
662 It is also possible to specify these settings by starting a pattern string with
663 one of the following sequences:
664 .sp
665 (*BSR_ANYCRLF) CR, LF, or CRLF only
666 (*BSR_UNICODE) any Unicode newline sequence
667 .sp
668 These override the default and the options given to the compiling function, but
669 they can themselves be overridden by options given to a matching function. Note
670 that these special settings, which are not Perl-compatible, are recognized only
671 at the very start of a pattern, and that they must be in upper case. If more
672 than one of them is present, the last one is used. They can be combined with a
673 change of newline convention; for example, a pattern can start with:
674 .sp
676 .sp
677 They can also be combined with the (*UTF8), (*UTF16), (*UTF32), (*UTF) or
678 (*UCP) special sequences. Inside a character class, \eR is treated as an
679 unrecognized escape sequence, and so matches the letter "R" by default, but
680 causes an error if PCRE_EXTRA is set.
681 .
682 .
683 .\" HTML <a name="uniextseq"></a>
684 .SS Unicode character properties
685 .rs
686 .sp
687 When PCRE is built with Unicode character property support, three additional
688 escape sequences that match characters with specific properties are available.
689 When in 8-bit non-UTF-8 mode, these sequences are of course limited to testing
690 characters whose codepoints are less than 256, but they do work in this mode.
691 The extra escape sequences are:
692 .sp
693 \ep{\fIxx\fP} a character with the \fIxx\fP property
694 \eP{\fIxx\fP} a character without the \fIxx\fP property
695 \eX a Unicode extended grapheme cluster
696 .sp
697 The property names represented by \fIxx\fP above are limited to the Unicode
698 script names, the general category properties, "Any", which matches any
699 character (including newline), and some special PCRE properties (described
700 in the
701 .\" HTML <a href="#extraprops">
702 .\" </a>
703 next section).
704 .\"
705 Other Perl properties such as "InMusicalSymbols" are not currently supported by
706 PCRE. Note that \eP{Any} does not match any characters, so always causes a
707 match failure.
708 .P
709 Sets of Unicode characters are defined as belonging to certain scripts. A
710 character from one of these sets can be matched using a script name. For
711 example:
712 .sp
713 \ep{Greek}
714 \eP{Han}
715 .sp
716 Those that are not part of an identified script are lumped together as
717 "Common". The current list of scripts is:
718 .P
719 Arabic,
720 Armenian,
721 Avestan,
722 Balinese,
723 Bamum,
724 Bassa_Vah,
725 Batak,
726 Bengali,
727 Bopomofo,
728 Brahmi,
729 Braille,
730 Buginese,
731 Buhid,
732 Canadian_Aboriginal,
733 Carian,
734 Caucasian_Albanian,
735 Chakma,
736 Cham,
737 Cherokee,
738 Common,
739 Coptic,
740 Cuneiform,
741 Cypriot,
742 Cyrillic,
743 Deseret,
744 Devanagari,
745 Duployan,
746 Egyptian_Hieroglyphs,
747 Elbasan,
748 Ethiopic,
749 Georgian,
750 Glagolitic,
751 Gothic,
752 Grantha,
753 Greek,
754 Gujarati,
755 Gurmukhi,
756 Han,
757 Hangul,
758 Hanunoo,
759 Hebrew,
760 Hiragana,
761 Imperial_Aramaic,
762 Inherited,
763 Inscriptional_Pahlavi,
764 Inscriptional_Parthian,
765 Javanese,
766 Kaithi,
767 Kannada,
768 Katakana,
769 Kayah_Li,
770 Kharoshthi,
771 Khmer,
772 Khojki,
773 Khudawadi,
774 Lao,
775 Latin,
776 Lepcha,
777 Limbu,
778 Linear_A,
779 Linear_B,
780 Lisu,
781 Lycian,
782 Lydian,
783 Mahajani,
784 Malayalam,
785 Mandaic,
786 Manichaean,
787 Meetei_Mayek,
788 Mende_Kikakui,
789 Meroitic_Cursive,
790 Meroitic_Hieroglyphs,
791 Miao,
792 Modi,
793 Mongolian,
794 Mro,
795 Myanmar,
796 Nabataean,
797 New_Tai_Lue,
798 Nko,
799 Ogham,
800 Ol_Chiki,
801 Old_Italic,
802 Old_North_Arabian,
803 Old_Permic,
804 Old_Persian,
805 Old_South_Arabian,
806 Old_Turkic,
807 Oriya,
808 Osmanya,
809 Pahawh_Hmong,
810 Palmyrene,
811 Pau_Cin_Hau,
812 Phags_Pa,
813 Phoenician,
814 Psalter_Pahlavi,
815 Rejang,
816 Runic,
817 Samaritan,
818 Saurashtra,
819 Sharada,
820 Shavian,
821 Siddham,
822 Sinhala,
823 Sora_Sompeng,
824 Sundanese,
825 Syloti_Nagri,
826 Syriac,
827 Tagalog,
828 Tagbanwa,
829 Tai_Le,
830 Tai_Tham,
831 Tai_Viet,
832 Takri,
833 Tamil,
834 Telugu,
835 Thaana,
836 Thai,
837 Tibetan,
838 Tifinagh,
839 Tirhuta,
840 Ugaritic,
841 Vai,
842 Warang_Citi,
843 Yi.
844 .P
845 Each character has exactly one Unicode general category property, specified by
846 a two-letter abbreviation. For compatibility with Perl, negation can be
847 specified by including a circumflex between the opening brace and the property
848 name. For example, \ep{^Lu} is the same as \eP{Lu}.
849 .P
850 If only one letter is specified with \ep or \eP, it includes all the general
851 category properties that start with that letter. In this case, in the absence
852 of negation, the curly brackets in the escape sequence are optional; these two
853 examples have the same effect:
854 .sp
855 \ep{L}
856 \epL
857 .sp
858 The following general category property codes are supported:
859 .sp
860 C Other
861 Cc Control
862 Cf Format
863 Cn Unassigned
864 Co Private use
865 Cs Surrogate
866 .sp
867 L Letter
868 Ll Lower case letter
869 Lm Modifier letter
870 Lo Other letter
871 Lt Title case letter
872 Lu Upper case letter
873 .sp
874 M Mark
875 Mc Spacing mark
876 Me Enclosing mark
877 Mn Non-spacing mark
878 .sp
879 N Number
880 Nd Decimal number
881 Nl Letter number
882 No Other number
883 .sp
884 P Punctuation
885 Pc Connector punctuation
886 Pd Dash punctuation
887 Pe Close punctuation
888 Pf Final punctuation
889 Pi Initial punctuation
890 Po Other punctuation
891 Ps Open punctuation
892 .sp
893 S Symbol
894 Sc Currency symbol
895 Sk Modifier symbol
896 Sm Mathematical symbol
897 So Other symbol
898 .sp
899 Z Separator
900 Zl Line separator
901 Zp Paragraph separator
902 Zs Space separator
903 .sp
904 The special property L& is also supported: it matches a character that has
905 the Lu, Ll, or Lt property, in other words, a letter that is not classified as
906 a modifier or "other".
907 .P
908 The Cs (Surrogate) property applies only to characters in the range U+D800 to
909 U+DFFF. Such characters are not valid in Unicode strings and so
910 cannot be tested by PCRE, unless UTF validity checking has been turned off
911 (see the discussion of PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK and
912 PCRE_NO_UTF32_CHECK in the
913 .\" HREF
914 \fBpcreapi\fP
915 .\"
916 page). Perl does not support the Cs property.
917 .P
918 The long synonyms for property names that Perl supports (such as \ep{Letter})
919 are not supported by PCRE, nor is it permitted to prefix any of these
920 properties with "Is".
921 .P
922 No character that is in the Unicode table has the Cn (unassigned) property.
923 Instead, this property is assumed for any code point that is not in the
924 Unicode table.
925 .P
926 Specifying caseless matching does not affect these escape sequences. For
927 example, \ep{Lu} always matches only upper case letters. This is different from
928 the behaviour of current versions of Perl.
929 .P
930 Matching characters by Unicode property is not fast, because PCRE has to do a
931 multistage table lookup in order to find a character's property. That is why
932 the traditional escape sequences such as \ed and \ew do not use Unicode
933 properties in PCRE by default, though you can make them do so by setting the
934 PCRE_UCP option or by starting the pattern with (*UCP).
935 .
936 .
937 .SS Extended grapheme clusters
938 .rs
939 .sp
940 The \eX escape matches any number of Unicode characters that form an "extended
941 grapheme cluster", and treats the sequence as an atomic group
942 .\" HTML <a href="#atomicgroup">
943 .\" </a>
944 (see below).
945 .\"
946 Up to and including release 8.31, PCRE matched an earlier, simpler definition
947 that was equivalent to
948 .sp
949 (?>\ePM\epM*)
950 .sp
951 That is, it matched a character without the "mark" property, followed by zero
952 or more characters with the "mark" property. Characters with the "mark"
953 property are typically non-spacing accents that affect the preceding character.
954 .P
955 This simple definition was extended in Unicode to include more complicated
956 kinds of composite character by giving each character a grapheme breaking
957 property, and creating rules that use these properties to define the boundaries
958 of extended grapheme clusters. In releases of PCRE later than 8.31, \eX matches
959 one of these clusters.
960 .P
961 \eX always matches at least one character. Then it decides whether to add
962 additional characters according to the following rules for ending a cluster:
963 .P
964 1. End at the end of the subject string.
965 .P
966 2. Do not end between CR and LF; otherwise end after any control character.
967 .P
968 3. Do not break Hangul (a Korean script) syllable sequences. Hangul characters
969 are of five types: L, V, T, LV, and LVT. An L character may be followed by an
970 L, V, LV, or LVT character; an LV or V character may be followed by a V or T
971 character; an LVT or T character may be follwed only by a T character.
972 .P
973 4. Do not end before extending characters or spacing marks. Characters with
974 the "mark" property always have the "extend" grapheme breaking property.
975 .P
976 5. Do not end after prepend characters.
977 .P
978 6. Otherwise, end the cluster.
979 .
980 .
981 .\" HTML <a name="extraprops"></a>
982 .SS PCRE's additional properties
983 .rs
984 .sp
985 As well as the standard Unicode properties described above, PCRE supports four
986 more that make it possible to convert traditional escape sequences such as \ew
987 and \es to use Unicode properties. PCRE uses these non-standard, non-Perl
988 properties internally when PCRE_UCP is set. However, they may also be used
989 explicitly. These properties are:
990 .sp
991 Xan Any alphanumeric character
992 Xps Any POSIX space character
993 Xsp Any Perl space character
994 Xwd Any Perl "word" character
995 .sp
996 Xan matches characters that have either the L (letter) or the N (number)
997 property. Xps matches the characters tab, linefeed, vertical tab, form feed, or
998 carriage return, and any other character that has the Z (separator) property.
999 Xsp is the same as Xps; it used to exclude vertical tab, for Perl
1000 compatibility, but Perl changed, and so PCRE followed at release 8.34. Xwd
1001 matches the same characters as Xan, plus underscore.
1002 .P
1003 There is another non-standard property, Xuc, which matches any character that
1004 can be represented by a Universal Character Name in C++ and other programming
1005 languages. These are the characters $, @, ` (grave accent), and all characters
1006 with Unicode code points greater than or equal to U+00A0, except for the
1007 surrogates U+D800 to U+DFFF. Note that most base (ASCII) characters are
1008 excluded. (Universal Character Names are of the form \euHHHH or \eUHHHHHHHH
1009 where H is a hexadecimal digit. Note that the Xuc property does not match these
1010 sequences but the characters that they represent.)
1011 .
1012 .
1013 .\" HTML <a name="resetmatchstart"></a>
1014 .SS "Resetting the match start"
1015 .rs
1016 .sp
1017 The escape sequence \eK causes any previously matched characters not to be
1018 included in the final matched sequence. For example, the pattern:
1019 .sp
1020 foo\eKbar
1021 .sp
1022 matches "foobar", but reports that it has matched "bar". This feature is
1023 similar to a lookbehind assertion
1024 .\" HTML <a href="#lookbehind">
1025 .\" </a>
1026 (described below).
1027 .\"
1028 However, in this case, the part of the subject before the real match does not
1029 have to be of fixed length, as lookbehind assertions do. The use of \eK does
1030 not interfere with the setting of
1031 .\" HTML <a href="#subpattern">
1032 .\" </a>
1033 captured substrings.
1034 .\"
1035 For example, when the pattern
1036 .sp
1037 (foo)\eKbar
1038 .sp
1039 matches "foobar", the first substring is still set to "foo".
1040 .P
1041 Perl documents that the use of \eK within assertions is "not well defined". In
1042 PCRE, \eK is acted upon when it occurs inside positive assertions, but is
1043 ignored in negative assertions. Note that when a pattern such as (?=ab\eK)
1044 matches, the reported start of the match can be greater than the end of the
1045 match.
1046 .
1047 .
1048 .\" HTML <a name="smallassertions"></a>
1049 .SS "Simple assertions"
1050 .rs
1051 .sp
1052 The final use of backslash is for certain simple assertions. An assertion
1053 specifies a condition that has to be met at a particular point in a match,
1054 without consuming any characters from the subject string. The use of
1055 subpatterns for more complicated assertions is described
1056 .\" HTML <a href="#bigassertions">
1057 .\" </a>
1058 below.
1059 .\"
1060 The backslashed assertions are:
1061 .sp
1062 \eb matches at a word boundary
1063 \eB matches when not at a word boundary
1064 \eA matches at the start of the subject
1065 \eZ matches at the end of the subject
1066 also matches before a newline at the end of the subject
1067 \ez matches only at the end of the subject
1068 \eG matches at the first matching position in the subject
1069 .sp
1070 Inside a character class, \eb has a different meaning; it matches the backspace
1071 character. If any other of these assertions appears in a character class, by
1072 default it matches the corresponding literal character (for example, \eB
1073 matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
1074 escape sequence" error is generated instead.
1075 .P
1076 A word boundary is a position in the subject string where the current character
1077 and the previous character do not both match \ew or \eW (i.e. one matches
1078 \ew and the other matches \eW), or the start or end of the string if the
1079 first or last character matches \ew, respectively. In a UTF mode, the meanings
1080 of \ew and \eW can be changed by setting the PCRE_UCP option. When this is
1081 done, it also affects \eb and \eB. Neither PCRE nor Perl has a separate "start
1082 of word" or "end of word" metasequence. However, whatever follows \eb normally
1083 determines which it is. For example, the fragment \eba matches "a" at the start
1084 of a word.
1085 .P
1086 The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
1087 dollar (described in the next section) in that they only ever match at the very
1088 start and end of the subject string, whatever options are set. Thus, they are
1089 independent of multiline mode. These three assertions are not affected by the
1090 PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
1091 circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
1092 argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
1093 at a point other than the beginning of the subject, \eA can never match. The
1094 difference between \eZ and \ez is that \eZ matches before a newline at the end
1095 of the string as well as at the very end, whereas \ez matches only at the end.
1096 .P
1097 The \eG assertion is true only when the current matching position is at the
1098 start point of the match, as specified by the \fIstartoffset\fP argument of
1099 \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
1100 non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
1101 arguments, you can mimic Perl's /g option, and it is in this kind of
1102 implementation where \eG can be useful.
1103 .P
1104 Note, however, that PCRE's interpretation of \eG, as the start of the current
1105 match, is subtly different from Perl's, which defines it as the end of the
1106 previous match. In Perl, these can be different when the previously matched
1107 string was empty. Because PCRE does just one match at a time, it cannot
1108 reproduce this behaviour.
1109 .P
1110 If all the alternatives of a pattern begin with \eG, the expression is anchored
1111 to the starting match position, and the "anchored" flag is set in the compiled
1112 regular expression.
1113 .
1114 .
1116 .rs
1117 .sp
1118 The circumflex and dollar metacharacters are zero-width assertions. That is,
1119 they test for a particular condition being true without consuming any
1120 characters from the subject string.
1121 .P
1122 Outside a character class, in the default matching mode, the circumflex
1123 character is an assertion that is true only if the current matching point is at
1124 the start of the subject string. If the \fIstartoffset\fP argument of
1125 \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
1126 option is unset. Inside a character class, circumflex has an entirely different
1127 meaning
1128 .\" HTML <a href="#characterclass">
1129 .\" </a>
1130 (see below).
1131 .\"
1132 .P
1133 Circumflex need not be the first character of the pattern if a number of
1134 alternatives are involved, but it should be the first thing in each alternative
1135 in which it appears if the pattern is ever to match that branch. If all
1136 possible alternatives start with a circumflex, that is, if the pattern is
1137 constrained to match only at the start of the subject, it is said to be an
1138 "anchored" pattern. (There are also other constructs that can cause a pattern
1139 to be anchored.)
1140 .P
1141 The dollar character is an assertion that is true only if the current matching
1142 point is at the end of the subject string, or immediately before a newline at
1143 the end of the string (by default). Note, however, that it does not actually
1144 match the newline. Dollar need not be the last character of the pattern if a
1145 number of alternatives are involved, but it should be the last item in any
1146 branch in which it appears. Dollar has no special meaning in a character class.
1147 .P
1148 The meaning of dollar can be changed so that it matches only at the very end of
1149 the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
1150 does not affect the \eZ assertion.
1151 .P
1152 The meanings of the circumflex and dollar characters are changed if the
1153 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
1154 immediately after internal newlines as well as at the start of the subject
1155 string. It does not match after a newline that ends the string. A dollar
1156 matches before any newlines in the string, as well as at the very end, when
1157 PCRE_MULTILINE is set. When newline is specified as the two-character
1158 sequence CRLF, isolated CR and LF characters do not indicate newlines.
1159 .P
1160 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
1161 \en represents a newline) in multiline mode, but not otherwise. Consequently,
1162 patterns that are anchored in single line mode because all branches start with
1163 ^ are not anchored in multiline mode, and a match for circumflex is possible
1164 when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
1165 PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
1166 .P
1167 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
1168 end of the subject in both modes, and if all branches of a pattern start with
1169 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
1170 .
1171 .
1172 .\" HTML <a name="fullstopdot"></a>
1174 .rs
1175 .sp
1176 Outside a character class, a dot in the pattern matches any one character in
1177 the subject string except (by default) a character that signifies the end of a
1178 line.
1179 .P
1180 When a line ending is defined as a single character, dot never matches that
1181 character; when the two-character sequence CRLF is used, dot does not match CR
1182 if it is immediately followed by LF, but otherwise it matches all characters
1183 (including isolated CRs and LFs). When any Unicode line endings are being
1184 recognized, dot does not match CR or LF or any of the other line ending
1185 characters.
1186 .P
1187 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
1188 option is set, a dot matches any one character, without exception. If the
1189 two-character sequence CRLF is present in the subject string, it takes two dots
1190 to match it.
1191 .P
1192 The handling of dot is entirely independent of the handling of circumflex and
1193 dollar, the only relationship being that they both involve newlines. Dot has no
1194 special meaning in a character class.
1195 .P
1196 The escape sequence \eN behaves like a dot, except that it is not affected by
1197 the PCRE_DOTALL option. In other words, it matches any character except one
1198 that signifies the end of a line. Perl also uses \eN to match characters by
1199 name; PCRE does not support this.
1200 .
1201 .
1203 .rs
1204 .sp
1205 Outside a character class, the escape sequence \eC matches any one data unit,
1206 whether or not a UTF mode is set. In the 8-bit library, one data unit is one
1207 byte; in the 16-bit library it is a 16-bit unit; in the 32-bit library it is
1208 a 32-bit unit. Unlike a dot, \eC always
1209 matches line-ending characters. The feature is provided in Perl in order to
1210 match individual bytes in UTF-8 mode, but it is unclear how it can usefully be
1211 used. Because \eC breaks up characters into individual data units, matching one
1212 unit with \eC in a UTF mode means that the rest of the string may start with a
1213 malformed UTF character. This has undefined results, because PCRE assumes that
1214 it is dealing with valid UTF strings (and by default it checks this at the
1215 start of processing unless the PCRE_NO_UTF8_CHECK, PCRE_NO_UTF16_CHECK or
1216 PCRE_NO_UTF32_CHECK option is used).
1217 .P
1218 PCRE does not allow \eC to appear in lookbehind assertions
1219 .\" HTML <a href="#lookbehind">
1220 .\" </a>
1221 (described below)
1222 .\"
1223 in a UTF mode, because this would make it impossible to calculate the length of
1224 the lookbehind.
1225 .P
1226 In general, the \eC escape sequence is best avoided. However, one
1227 way of using it that avoids the problem of malformed UTF characters is to use a
1228 lookahead to check the length of the next character, as in this pattern, which
1229 could be used with a UTF-8 string (ignore white space and line breaks):
1230 .sp
1231 (?| (?=[\ex00-\ex7f])(\eC) |
1232 (?=[\ex80-\ex{7ff}])(\eC)(\eC) |
1233 (?=[\ex{800}-\ex{ffff}])(\eC)(\eC)(\eC) |
1234 (?=[\ex{10000}-\ex{1fffff}])(\eC)(\eC)(\eC)(\eC))
1235 .sp
1236 A group that starts with (?| resets the capturing parentheses numbers in each
1237 alternative (see
1238 .\" HTML <a href="#dupsubpatternnumber">
1239 .\" </a>
1240 "Duplicate Subpattern Numbers"
1241 .\"
1242 below). The assertions at the start of each branch check the next UTF-8
1243 character for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The
1244 character's individual bytes are then captured by the appropriate number of
1245 groups.
1246 .
1247 .
1248 .\" HTML <a name="characterclass"></a>
1250 .rs
1251 .sp
1252 An opening square bracket introduces a character class, terminated by a closing
1253 square bracket. A closing square bracket on its own is not special by default.
1254 However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
1255 bracket causes a compile-time error. If a closing square bracket is required as
1256 a member of the class, it should be the first data character in the class
1257 (after an initial circumflex, if present) or escaped with a backslash.
1258 .P
1259 A character class matches a single character in the subject. In a UTF mode, the
1260 character may be more than one data unit long. A matched character must be in
1261 the set of characters defined by the class, unless the first character in the
1262 class definition is a circumflex, in which case the subject character must not
1263 be in the set defined by the class. If a circumflex is actually required as a
1264 member of the class, ensure it is not the first character, or escape it with a
1265 backslash.
1266 .P
1267 For example, the character class [aeiou] matches any lower case vowel, while
1268 [^aeiou] matches any character that is not a lower case vowel. Note that a
1269 circumflex is just a convenient notation for specifying the characters that
1270 are in the class by enumerating those that are not. A class that starts with a
1271 circumflex is not an assertion; it still consumes a character from the subject
1272 string, and therefore it fails if the current pointer is at the end of the
1273 string.
1274 .P
1275 In UTF-8 (UTF-16, UTF-32) mode, characters with values greater than 255 (0xffff)
1276 can be included in a class as a literal string of data units, or by using the
1277 \ex{ escaping mechanism.
1278 .P
1279 When caseless matching is set, any letters in a class represent both their
1280 upper case and lower case versions, so for example, a caseless [aeiou] matches
1281 "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1282 caseful version would. In a UTF mode, PCRE always understands the concept of
1283 case for characters whose values are less than 128, so caseless matching is
1284 always possible. For characters with higher values, the concept of case is
1285 supported if PCRE is compiled with Unicode property support, but not otherwise.
1286 If you want to use caseless matching in a UTF mode for characters 128 and
1287 above, you must ensure that PCRE is compiled with Unicode property support as
1288 well as with UTF support.
1289 .P
1290 Characters that might indicate line breaks are never treated in any special way
1291 when matching character classes, whatever line-ending sequence is in use, and
1292 whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
1293 such as [^a] always matches one of these characters.
1294 .P
1295 The minus (hyphen) character can be used to specify a range of characters in a
1296 character class. For example, [d-m] matches any letter between d and m,
1297 inclusive. If a minus character is required in a class, it must be escaped with
1298 a backslash or appear in a position where it cannot be interpreted as
1299 indicating a range, typically as the first or last character in the class, or
1300 immediately after a range. For example, [b-d-z] matches letters in the range b
1301 to d, a hyphen character, or z.
1302 .P
1303 It is not possible to have the literal character "]" as the end character of a
1304 range. A pattern such as [W-]46] is interpreted as a class of two characters
1305 ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1306 "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1307 the end of range, so [W-\e]46] is interpreted as a class containing a range
1308 followed by two other characters. The octal or hexadecimal representation of
1309 "]" can also be used to end a range.
1310 .P
1311 An error is generated if a POSIX character class (see below) or an escape
1312 sequence other than one that defines a single character appears at a point
1313 where a range ending character is expected. For example, [z-\exff] is valid,
1314 but [A-\ed] and [A-[:digit:]] are not.
1315 .P
1316 Ranges operate in the collating sequence of character values. They can also be
1317 used for characters specified numerically, for example [\e000-\e037]. Ranges
1318 can include any characters that are valid for the current mode.
1319 .P
1320 If a range that includes letters is used when caseless matching is set, it
1321 matches the letters in either case. For example, [W-c] is equivalent to
1322 [][\e\e^_`wxyzabc], matched caselessly, and in a non-UTF mode, if character
1323 tables for a French locale are in use, [\exc8-\excb] matches accented E
1324 characters in both cases. In UTF modes, PCRE supports the concept of case for
1325 characters with values greater than 128 only when it is compiled with Unicode
1326 property support.
1327 .P
1328 The character escape sequences \ed, \eD, \eh, \eH, \ep, \eP, \es, \eS, \ev,
1329 \eV, \ew, and \eW may appear in a character class, and add the characters that
1330 they match to the class. For example, [\edABCDEF] matches any hexadecimal
1331 digit. In UTF modes, the PCRE_UCP option affects the meanings of \ed, \es, \ew
1332 and their upper case partners, just as it does when they appear outside a
1333 character class, as described in the section entitled
1334 .\" HTML <a href="#genericchartypes">
1335 .\" </a>
1336 "Generic character types"
1337 .\"
1338 above. The escape sequence \eb has a different meaning inside a character
1339 class; it matches the backspace character. The sequences \eB, \eN, \eR, and \eX
1340 are not special inside a character class. Like any other unrecognized escape
1341 sequences, they are treated as the literal characters "B", "N", "R", and "X" by
1342 default, but cause an error if the PCRE_EXTRA option is set.
1343 .P
1344 A circumflex can conveniently be used with the upper case character types to
1345 specify a more restricted set of characters than the matching lower case type.
1346 For example, the class [^\eW_] matches any letter or digit, but not underscore,
1347 whereas [\ew] includes underscore. A positive character class should be read as
1348 "something OR something OR ..." and a negative class as "NOT something AND NOT
1349 something AND NOT ...".
1350 .P
1351 The only metacharacters that are recognized in character classes are backslash,
1352 hyphen (only where it can be interpreted as specifying a range), circumflex
1353 (only at the start), opening square bracket (only when it can be interpreted as
1354 introducing a POSIX class name, or for a special compatibility feature - see
1355 the next two sections), and the terminating closing square bracket. However,
1356 escaping other non-alphanumeric characters does no harm.
1357 .
1358 .
1360 .rs
1361 .sp
1362 Perl supports the POSIX notation for character classes. This uses names
1363 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1364 this notation. For example,
1365 .sp
1366 [01[:alpha:]%]
1367 .sp
1368 matches "0", "1", any alphabetic character, or "%". The supported class names
1369 are:
1370 .sp
1371 alnum letters and digits
1372 alpha letters
1373 ascii character codes 0 - 127
1374 blank space or tab only
1375 cntrl control characters
1376 digit decimal digits (same as \ed)
1377 graph printing characters, excluding space
1378 lower lower case letters
1379 print printing characters, including space
1380 punct printing characters, excluding letters and digits and space
1381 space white space (the same as \es from PCRE 8.34)
1382 upper upper case letters
1383 word "word" characters (same as \ew)
1384 xdigit hexadecimal digits
1385 .sp
1386 The default "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13),
1387 and space (32). If locale-specific matching is taking place, the list of space
1388 characters may be different; there may be fewer or more of them. "Space" used
1389 to be different to \es, which did not include VT, for Perl compatibility.
1390 However, Perl changed at release 5.18, and PCRE followed at release 8.34.
1391 "Space" and \es now match the same set of characters.
1392 .P
1393 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1394 5.8. Another Perl extension is negation, which is indicated by a ^ character
1395 after the colon. For example,
1396 .sp
1397 [12[:^digit:]]
1398 .sp
1399 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1400 syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1401 supported, and an error is given if they are encountered.
1402 .P
1403 By default, characters with values greater than 128 do not match any of the
1404 POSIX character classes. However, if the PCRE_UCP option is passed to
1405 \fBpcre_compile()\fP, some of the classes are changed so that Unicode character
1406 properties are used. This is achieved by replacing certain POSIX classes by
1407 other sequences, as follows:
1408 .sp
1409 [:alnum:] becomes \ep{Xan}
1410 [:alpha:] becomes \ep{L}
1411 [:blank:] becomes \eh
1412 [:digit:] becomes \ep{Nd}
1413 [:lower:] becomes \ep{Ll}
1414 [:space:] becomes \ep{Xps}
1415 [:upper:] becomes \ep{Lu}
1416 [:word:] becomes \ep{Xwd}
1417 .sp
1418 Negated versions, such as [:^alpha:] use \eP instead of \ep. Three other POSIX
1419 classes are handled specially in UCP mode:
1420 .TP 10
1421 [:graph:]
1422 This matches characters that have glyphs that mark the page when printed. In
1423 Unicode property terms, it matches all characters with the L, M, N, P, S, or Cf
1424 properties, except for:
1425 .sp
1426 U+061C Arabic Letter Mark
1427 U+180E Mongolian Vowel Separator
1428 U+2066 - U+2069 Various "isolate"s
1429 .sp
1430 .TP 10
1431 [:print:]
1432 This matches the same characters as [:graph:] plus space characters that are
1433 not controls, that is, characters with the Zs property.
1434 .TP 10
1435 [:punct:]
1436 This matches all characters that have the Unicode P (punctuation) property,
1437 plus those characters whose code points are less than 128 that have the S
1438 (Symbol) property.
1439 .P
1440 The other POSIX classes are unchanged, and match only characters with code
1441 points less than 128.
1442 .
1443 .
1445 .rs
1446 .sp
1447 In the POSIX.2 compliant library that was included in 4.4BSD Unix, the ugly
1448 syntax [[:<:]] and [[:>:]] is used for matching "start of word" and "end of
1449 word". PCRE treats these items as follows:
1450 .sp
1451 [[:<:]] is converted to \eb(?=\ew)
1452 [[:>:]] is converted to \eb(?<=\ew)
1453 .sp
1454 Only these exact character sequences are recognized. A sequence such as
1455 [a[:<:]b] provokes error for an unrecognized POSIX class name. This support is
1456 not compatible with Perl. It is provided to help migrations from other
1457 environments, and is best not used in any new patterns. Note that \eb matches
1458 at the start and the end of a word (see
1459 .\" HTML <a href="#smallassertions">
1460 .\" </a>
1461 "Simple assertions"
1462 .\"
1463 above), and in a Perl-style pattern the preceding or following character
1464 normally shows which is wanted, without the need for the assertions that are
1465 used above in order to give exactly the POSIX behaviour.
1466 .
1467 .
1469 .rs
1470 .sp
1471 Vertical bar characters are used to separate alternative patterns. For example,
1472 the pattern
1473 .sp
1474 gilbert|sullivan
1475 .sp
1476 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1477 and an empty alternative is permitted (matching the empty string). The matching
1478 process tries each alternative in turn, from left to right, and the first one
1479 that succeeds is used. If the alternatives are within a subpattern
1480 .\" HTML <a href="#subpattern">
1481 .\" </a>
1482 (defined below),
1483 .\"
1484 "succeeds" means matching the rest of the main pattern as well as the
1485 alternative in the subpattern.
1486 .
1487 .
1489 .rs
1490 .sp
1491 The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1492 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1493 the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1494 The option letters are
1495 .sp
1496 i for PCRE_CASELESS
1498 s for PCRE_DOTALL
1499 x for PCRE_EXTENDED
1500 .sp
1501 For example, (?im) sets caseless, multiline matching. It is also possible to
1502 unset these options by preceding the letter with a hyphen, and a combined
1503 setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1504 PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1505 permitted. If a letter appears both before and after the hyphen, the option is
1506 unset.
1507 .P
1508 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1509 changed in the same way as the Perl-compatible options by using the characters
1510 J, U and X respectively.
1511 .P
1512 When one of these option changes occurs at top level (that is, not inside
1513 subpattern parentheses), the change applies to the remainder of the pattern
1514 that follows. An option change within a subpattern (see below for a description
1515 of subpatterns) affects only that part of the subpattern that follows it, so
1516 .sp
1517 (a(?i)b)c
1518 .sp
1519 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1520 By this means, options can be made to have different settings in different
1521 parts of the pattern. Any changes made in one alternative do carry on
1522 into subsequent branches within the same subpattern. For example,
1523 .sp
1524 (a(?i)b|c)
1525 .sp
1526 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1527 branch is abandoned before the option setting. This is because the effects of
1528 option settings happen at compile time. There would be some very weird
1529 behaviour otherwise.
1530 .P
1531 \fBNote:\fP There are other PCRE-specific options that can be set by the
1532 application when the compiling or matching functions are called. In some cases
1533 the pattern can contain special leading sequences such as (*CRLF) to override
1534 what the application has set or what has been defaulted. Details are given in
1535 the section entitled
1536 .\" HTML <a href="#newlineseq">
1537 .\" </a>
1538 "Newline sequences"
1539 .\"
1540 above. There are also the (*UTF8), (*UTF16),(*UTF32), and (*UCP) leading
1541 sequences that can be used to set UTF and Unicode property modes; they are
1542 equivalent to setting the PCRE_UTF8, PCRE_UTF16, PCRE_UTF32 and the PCRE_UCP
1543 options, respectively. The (*UTF) sequence is a generic version that can be
1544 used with any of the libraries. However, the application can set the
1545 PCRE_NEVER_UTF option, which locks out the use of the (*UTF) sequences.
1546 .
1547 .
1548 .\" HTML <a name="subpattern"></a>
1550 .rs
1551 .sp
1552 Subpatterns are delimited by parentheses (round brackets), which can be nested.
1553 Turning part of a pattern into a subpattern does two things:
1554 .sp
1555 1. It localizes a set of alternatives. For example, the pattern
1556 .sp
1557 cat(aract|erpillar|)
1558 .sp
1559 matches "cataract", "caterpillar", or "cat". Without the parentheses, it would
1560 match "cataract", "erpillar" or an empty string.
1561 .sp
1562 2. It sets up the subpattern as a capturing subpattern. This means that, when
1563 the whole pattern matches, that portion of the subject string that matched the
1564 subpattern is passed back to the caller via the \fIovector\fP argument of the
1565 matching function. (This applies only to the traditional matching functions;
1566 the DFA matching functions do not support capturing.)
1567 .P
1568 Opening parentheses are counted from left to right (starting from 1) to obtain
1569 numbers for the capturing subpatterns. For example, if the string "the red
1570 king" is matched against the pattern
1571 .sp
1572 the ((red|white) (king|queen))
1573 .sp
1574 the captured substrings are "red king", "red", and "king", and are numbered 1,
1575 2, and 3, respectively.
1576 .P
1577 The fact that plain parentheses fulfil two functions is not always helpful.
1578 There are often times when a grouping subpattern is required without a
1579 capturing requirement. If an opening parenthesis is followed by a question mark
1580 and a colon, the subpattern does not do any capturing, and is not counted when
1581 computing the number of any subsequent capturing subpatterns. For example, if
1582 the string "the white queen" is matched against the pattern
1583 .sp
1584 the ((?:red|white) (king|queen))
1585 .sp
1586 the captured substrings are "white queen" and "queen", and are numbered 1 and
1587 2. The maximum number of capturing subpatterns is 65535.
1588 .P
1589 As a convenient shorthand, if any option settings are required at the start of
1590 a non-capturing subpattern, the option letters may appear between the "?" and
1591 the ":". Thus the two patterns
1592 .sp
1593 (?i:saturday|sunday)
1594 (?:(?i)saturday|sunday)
1595 .sp
1596 match exactly the same set of strings. Because alternative branches are tried
1597 from left to right, and options are not reset until the end of the subpattern
1598 is reached, an option setting in one branch does affect subsequent branches, so
1599 the above patterns match "SUNDAY" as well as "Saturday".
1600 .
1601 .
1602 .\" HTML <a name="dupsubpatternnumber"></a>
1604 .rs
1605 .sp
1606 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1607 the same numbers for its capturing parentheses. Such a subpattern starts with
1608 (?| and is itself a non-capturing subpattern. For example, consider this
1609 pattern:
1610 .sp
1611 (?|(Sat)ur|(Sun))day
1612 .sp
1613 Because the two alternatives are inside a (?| group, both sets of capturing
1614 parentheses are numbered one. Thus, when the pattern matches, you can look
1615 at captured substring number one, whichever alternative matched. This construct
1616 is useful when you want to capture part, but not all, of one of a number of
1617 alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1618 number is reset at the start of each branch. The numbers of any capturing
1619 parentheses that follow the subpattern start after the highest number used in
1620 any branch. The following example is taken from the Perl documentation. The
1621 numbers underneath show in which buffer the captured content will be stored.
1622 .sp
1623 # before ---------------branch-reset----------- after
1624 / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1625 # 1 2 2 3 2 3 4
1626 .sp
1627 A back reference to a numbered subpattern uses the most recent value that is
1628 set for that number by any subpattern. The following pattern matches "abcabc"
1629 or "defdef":
1630 .sp
1631 /(?|(abc)|(def))\e1/
1632 .sp
1633 In contrast, a subroutine call to a numbered subpattern always refers to the
1634 first one in the pattern with the given number. The following pattern matches
1635 "abcabc" or "defabc":
1636 .sp
1637 /(?|(abc)|(def))(?1)/
1638 .sp
1639 If a
1640 .\" HTML <a href="#conditions">
1641 .\" </a>
1642 condition test
1643 .\"
1644 for a subpattern's having matched refers to a non-unique number, the test is
1645 true if any of the subpatterns of that number have matched.
1646 .P
1647 An alternative approach to using this "branch reset" feature is to use
1648 duplicate named subpatterns, as described in the next section.
1649 .
1650 .
1652 .rs
1653 .sp
1654 Identifying capturing parentheses by number is simple, but it can be very hard
1655 to keep track of the numbers in complicated regular expressions. Furthermore,
1656 if an expression is modified, the numbers may change. To help with this
1657 difficulty, PCRE supports the naming of subpatterns. This feature was not
1658 added to Perl until release 5.10. Python had the feature earlier, and PCRE
1659 introduced it at release 4.0, using the Python syntax. PCRE now supports both
1660 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1661 have different names, but PCRE does not.
1662 .P
1663 In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1664 (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1665 parentheses from other parts of the pattern, such as
1666 .\" HTML <a href="#backreferences">
1667 .\" </a>
1668 back references,
1669 .\"
1670 .\" HTML <a href="#recursion">
1671 .\" </a>
1672 recursion,
1673 .\"
1674 and
1675 .\" HTML <a href="#conditions">
1676 .\" </a>
1677 conditions,
1678 .\"
1679 can be made by name as well as by number.
1680 .P
1681 Names consist of up to 32 alphanumeric characters and underscores, but must
1682 start with a non-digit. Named capturing parentheses are still allocated numbers
1683 as well as names, exactly as if the names were not present. The PCRE API
1684 provides function calls for extracting the name-to-number translation table
1685 from a compiled pattern. There is also a convenience function for extracting a
1686 captured substring by name.
1687 .P
1688 By default, a name must be unique within a pattern, but it is possible to relax
1689 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1690 names are also always permitted for subpatterns with the same number, set up as
1691 described in the previous section.) Duplicate names can be useful for patterns
1692 where only one instance of the named parentheses can match. Suppose you want to
1693 match the name of a weekday, either as a 3-letter abbreviation or as the full
1694 name, and in both cases you want to extract the abbreviation. This pattern
1695 (ignoring the line breaks) does the job:
1696 .sp
1697 (?<DN>Mon|Fri|Sun)(?:day)?|
1698 (?<DN>Tue)(?:sday)?|
1699 (?<DN>Wed)(?:nesday)?|
1700 (?<DN>Thu)(?:rsday)?|
1701 (?<DN>Sat)(?:urday)?
1702 .sp
1703 There are five capturing substrings, but only one is ever set after a match.
1704 (An alternative way of solving this problem is to use a "branch reset"
1705 subpattern, as described in the previous section.)
1706 .P
1707 The convenience function for extracting the data by name returns the substring
1708 for the first (and in this example, the only) subpattern of that name that
1709 matched. This saves searching to find which numbered subpattern it was.
1710 .P
1711 If you make a back reference to a non-unique named subpattern from elsewhere in
1712 the pattern, the subpatterns to which the name refers are checked in the order
1713 in which they appear in the overall pattern. The first one that is set is used
1714 for the reference. For example, this pattern matches both "foofoo" and
1715 "barbar" but not "foobar" or "barfoo":
1716 .sp
1717 (?:(?<n>foo)|(?<n>bar))\ek<n>
1718 .sp
1719 .P
1720 If you make a subroutine call to a non-unique named subpattern, the one that
1721 corresponds to the first occurrence of the name is used. In the absence of
1722 duplicate numbers (see the previous section) this is the one with the lowest
1723 number.
1724 .P
1725 If you use a named reference in a condition
1726 test (see the
1727 .\"
1728 .\" HTML <a href="#conditions">
1729 .\" </a>
1730 section about conditions
1731 .\"
1732 below), either to check whether a subpattern has matched, or to check for
1733 recursion, all subpatterns with the same name are tested. If the condition is
1734 true for any one of them, the overall condition is true. This is the same
1735 behaviour as testing by number. For further details of the interfaces for
1736 handling named subpatterns, see the
1737 .\" HREF
1738 \fBpcreapi\fP
1739 .\"
1740 documentation.
1741 .P
1742 \fBWarning:\fP You cannot use different names to distinguish between two
1743 subpatterns with the same number because PCRE uses only the numbers when
1744 matching. For this reason, an error is given at compile time if different names
1745 are given to subpatterns with the same number. However, you can always give the
1746 same name to subpatterns with the same number, even when PCRE_DUPNAMES is not
1747 set.
1748 .
1749 .
1751 .rs
1752 .sp
1753 Repetition is specified by quantifiers, which can follow any of the following
1754 items:
1755 .sp
1756 a literal data character
1757 the dot metacharacter
1758 the \eC escape sequence
1759 the \eX escape sequence
1760 the \eR escape sequence
1761 an escape such as \ed or \epL that matches a single character
1762 a character class
1763 a back reference (see next section)
1764 a parenthesized subpattern (including assertions)
1765 a subroutine call to a subpattern (recursive or otherwise)
1766 .sp
1767 The general repetition quantifier specifies a minimum and maximum number of
1768 permitted matches, by giving the two numbers in curly brackets (braces),
1769 separated by a comma. The numbers must be less than 65536, and the first must
1770 be less than or equal to the second. For example:
1771 .sp
1772 z{2,4}
1773 .sp
1774 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1775 character. If the second number is omitted, but the comma is present, there is
1776 no upper limit; if the second number and the comma are both omitted, the
1777 quantifier specifies an exact number of required matches. Thus
1778 .sp
1779 [aeiou]{3,}
1780 .sp
1781 matches at least 3 successive vowels, but may match many more, while
1782 .sp
1783 \ed{8}
1784 .sp
1785 matches exactly 8 digits. An opening curly bracket that appears in a position
1786 where a quantifier is not allowed, or one that does not match the syntax of a
1787 quantifier, is taken as a literal character. For example, {,6} is not a
1788 quantifier, but a literal string of four characters.
1789 .P
1790 In UTF modes, quantifiers apply to characters rather than to individual data
1791 units. Thus, for example, \ex{100}{2} matches two characters, each of
1792 which is represented by a two-byte sequence in a UTF-8 string. Similarly,
1793 \eX{3} matches three Unicode extended grapheme clusters, each of which may be
1794 several data units long (and they may be of different lengths).
1795 .P
1796 The quantifier {0} is permitted, causing the expression to behave as if the
1797 previous item and the quantifier were not present. This may be useful for
1798 subpatterns that are referenced as
1799 .\" HTML <a href="#subpatternsassubroutines">
1800 .\" </a>
1801 subroutines
1802 .\"
1803 from elsewhere in the pattern (but see also the section entitled
1804 .\" HTML <a href="#subdefine">
1805 .\" </a>
1806 "Defining subpatterns for use by reference only"
1807 .\"
1808 below). Items other than subpatterns that have a {0} quantifier are omitted
1809 from the compiled pattern.
1810 .P
1811 For convenience, the three most common quantifiers have single-character
1812 abbreviations:
1813 .sp
1814 * is equivalent to {0,}
1815 + is equivalent to {1,}
1816 ? is equivalent to {0,1}
1817 .sp
1818 It is possible to construct infinite loops by following a subpattern that can
1819 match no characters with a quantifier that has no upper limit, for example:
1820 .sp
1821 (a?)*
1822 .sp
1823 Earlier versions of Perl and PCRE used to give an error at compile time for
1824 such patterns. However, because there are cases where this can be useful, such
1825 patterns are now accepted, but if any repetition of the subpattern does in fact
1826 match no characters, the loop is forcibly broken.
1827 .P
1828 By default, the quantifiers are "greedy", that is, they match as much as
1829 possible (up to the maximum number of permitted times), without causing the
1830 rest of the pattern to fail. The classic example of where this gives problems
1831 is in trying to match comments in C programs. These appear between /* and */
1832 and within the comment, individual * and / characters may appear. An attempt to
1833 match C comments by applying the pattern
1834 .sp
1835 /\e*.*\e*/
1836 .sp
1837 to the string
1838 .sp
1839 /* first comment */ not comment /* second comment */
1840 .sp
1841 fails, because it matches the entire string owing to the greediness of the .*
1842 item.
1843 .P
1844 However, if a quantifier is followed by a question mark, it ceases to be
1845 greedy, and instead matches the minimum number of times possible, so the
1846 pattern
1847 .sp
1848 /\e*.*?\e*/
1849 .sp
1850 does the right thing with the C comments. The meaning of the various
1851 quantifiers is not otherwise changed, just the preferred number of matches.
1852 Do not confuse this use of question mark with its use as a quantifier in its
1853 own right. Because it has two uses, it can sometimes appear doubled, as in
1854 .sp
1855 \ed??\ed
1856 .sp
1857 which matches one digit by preference, but can match two if that is the only
1858 way the rest of the pattern matches.
1859 .P
1860 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1861 the quantifiers are not greedy by default, but individual ones can be made
1862 greedy by following them with a question mark. In other words, it inverts the
1863 default behaviour.
1864 .P
1865 When a parenthesized subpattern is quantified with a minimum repeat count that
1866 is greater than 1 or with a limited maximum, more memory is required for the
1867 compiled pattern, in proportion to the size of the minimum or maximum.
1868 .P
1869 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1870 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1871 implicitly anchored, because whatever follows will be tried against every
1872 character position in the subject string, so there is no point in retrying the
1873 overall match at any position after the first. PCRE normally treats such a
1874 pattern as though it were preceded by \eA.
1875 .P
1876 In cases where it is known that the subject string contains no newlines, it is
1877 worth setting PCRE_DOTALL in order to obtain this optimization, or
1878 alternatively using ^ to indicate anchoring explicitly.
1879 .P
1880 However, there are some cases where the optimization cannot be used. When .*
1881 is inside capturing parentheses that are the subject of a back reference
1882 elsewhere in the pattern, a match at the start may fail where a later one
1883 succeeds. Consider, for example:
1884 .sp
1885 (.*)abc\e1
1886 .sp
1887 If the subject is "xyz123abc123" the match point is the fourth character. For
1888 this reason, such a pattern is not implicitly anchored.
1889 .P
1890 Another case where implicit anchoring is not applied is when the leading .* is
1891 inside an atomic group. Once again, a match at the start may fail where a later
1892 one succeeds. Consider this pattern:
1893 .sp
1894 (?>.*?a)b
1895 .sp
1896 It matches "ab" in the subject "aab". The use of the backtracking control verbs
1897 (*PRUNE) and (*SKIP) also disable this optimization.
1898 .P
1899 When a capturing subpattern is repeated, the value captured is the substring
1900 that matched the final iteration. For example, after
1901 .sp
1902 (tweedle[dume]{3}\es*)+
1903 .sp
1904 has matched "tweedledum tweedledee" the value of the captured substring is
1905 "tweedledee". However, if there are nested capturing subpatterns, the
1906 corresponding captured values may have been set in previous iterations. For
1907 example, after
1908 .sp
1909 /(a|(b))+/
1910 .sp
1911 matches "aba" the value of the second captured substring is "b".
1912 .
1913 .
1914 .\" HTML <a name="atomicgroup"></a>
1916 .rs
1917 .sp
1918 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1919 repetition, failure of what follows normally causes the repeated item to be
1920 re-evaluated to see if a different number of repeats allows the rest of the
1921 pattern to match. Sometimes it is useful to prevent this, either to change the
1922 nature of the match, or to cause it fail earlier than it otherwise might, when
1923 the author of the pattern knows there is no point in carrying on.
1924 .P
1925 Consider, for example, the pattern \ed+foo when applied to the subject line
1926 .sp
1927 123456bar
1928 .sp
1929 After matching all 6 digits and then failing to match "foo", the normal
1930 action of the matcher is to try again with only 5 digits matching the \ed+
1931 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1932 (a term taken from Jeffrey Friedl's book) provides the means for specifying
1933 that once a subpattern has matched, it is not to be re-evaluated in this way.
1934 .P
1935 If we use atomic grouping for the previous example, the matcher gives up
1936 immediately on failing to match "foo" the first time. The notation is a kind of
1937 special parenthesis, starting with (?> as in this example:
1938 .sp
1939 (?>\ed+)foo
1940 .sp
1941 This kind of parenthesis "locks up" the part of the pattern it contains once
1942 it has matched, and a failure further into the pattern is prevented from
1943 backtracking into it. Backtracking past it to previous items, however, works as
1944 normal.
1945 .P
1946 An alternative description is that a subpattern of this type matches the string
1947 of characters that an identical standalone pattern would match, if anchored at
1948 the current point in the subject string.
1949 .P
1950 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1951 the above example can be thought of as a maximizing repeat that must swallow
1952 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1953 number of digits they match in order to make the rest of the pattern match,
1954 (?>\ed+) can only match an entire sequence of digits.
1955 .P
1956 Atomic groups in general can of course contain arbitrarily complicated
1957 subpatterns, and can be nested. However, when the subpattern for an atomic
1958 group is just a single repeated item, as in the example above, a simpler
1959 notation, called a "possessive quantifier" can be used. This consists of an
1960 additional + character following a quantifier. Using this notation, the
1961 previous example can be rewritten as
1962 .sp
1963 \ed++foo
1964 .sp
1965 Note that a possessive quantifier can be used with an entire group, for
1966 example:
1967 .sp
1968 (abc|xyz){2,3}+
1969 .sp
1970 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1971 option is ignored. They are a convenient notation for the simpler forms of
1972 atomic group. However, there is no difference in the meaning of a possessive
1973 quantifier and the equivalent atomic group, though there may be a performance
1974 difference; possessive quantifiers should be slightly faster.
1975 .P
1976 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1977 Jeffrey Friedl originated the idea (and the name) in the first edition of his
1978 book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1979 package, and PCRE copied it from there. It ultimately found its way into Perl
1980 at release 5.10.
1981 .P
1982 PCRE has an optimization that automatically "possessifies" certain simple
1983 pattern constructs. For example, the sequence A+B is treated as A++B because
1984 there is no point in backtracking into a sequence of A's when B must follow.
1985 .P
1986 When a pattern contains an unlimited repeat inside a subpattern that can itself
1987 be repeated an unlimited number of times, the use of an atomic group is the
1988 only way to avoid some failing matches taking a very long time indeed. The
1989 pattern
1990 .sp
1991 (\eD+|<\ed+>)*[!?]
1992 .sp
1993 matches an unlimited number of substrings that either consist of non-digits, or
1994 digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1995 quickly. However, if it is applied to
1996 .sp
1997 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1998 .sp
1999 it takes a long time before reporting failure. This is because the string can
2000 be divided between the internal \eD+ repeat and the external * repeat in a
2001 large number of ways, and all have to be tried. (The example uses [!?] rather
2002 than a single character at the end, because both PCRE and Perl have an
2003 optimization that allows for fast failure when a single character is used. They
2004 remember the last single character that is required for a match, and fail early
2005 if it is not present in the string.) If the pattern is changed so that it uses
2006 an atomic group, like this:
2007 .sp
2008 ((?>\eD+)|<\ed+>)*[!?]
2009 .sp
2010 sequences of non-digits cannot be broken, and failure happens quickly.
2011 .
2012 .
2013 .\" HTML <a name="backreferences"></a>
2015 .rs
2016 .sp
2017 Outside a character class, a backslash followed by a digit greater than 0 (and
2018 possibly further digits) is a back reference to a capturing subpattern earlier
2019 (that is, to its left) in the pattern, provided there have been that many
2020 previous capturing left parentheses.
2021 .P
2022 However, if the decimal number following the backslash is less than 10, it is
2023 always taken as a back reference, and causes an error only if there are not
2024 that many capturing left parentheses in the entire pattern. In other words, the
2025 parentheses that are referenced need not be to the left of the reference for
2026 numbers less than 10. A "forward back reference" of this type can make sense
2027 when a repetition is involved and the subpattern to the right has participated
2028 in an earlier iteration.
2029 .P
2030 It is not possible to have a numerical "forward back reference" to a subpattern
2031 whose number is 10 or more using this syntax because a sequence such as \e50 is
2032 interpreted as a character defined in octal. See the subsection entitled
2033 "Non-printing characters"
2034 .\" HTML <a href="#digitsafterbackslash">
2035 .\" </a>
2036 above
2037 .\"
2038 for further details of the handling of digits following a backslash. There is
2039 no such problem when named parentheses are used. A back reference to any
2040 subpattern is possible using named parentheses (see below).
2041 .P
2042 Another way of avoiding the ambiguity inherent in the use of digits following a
2043 backslash is to use the \eg escape sequence. This escape must be followed by an
2044 unsigned number or a negative number, optionally enclosed in braces. These
2045 examples are all identical:
2046 .sp
2047 (ring), \e1
2048 (ring), \eg1
2049 (ring), \eg{1}
2050 .sp
2051 An unsigned number specifies an absolute reference without the ambiguity that
2052 is present in the older syntax. It is also useful when literal digits follow
2053 the reference. A negative number is a relative reference. Consider this
2054 example:
2055 .sp
2056 (abc(def)ghi)\eg{-1}
2057 .sp
2058 The sequence \eg{-1} is a reference to the most recently started capturing
2059 subpattern before \eg, that is, is it equivalent to \e2 in this example.
2060 Similarly, \eg{-2} would be equivalent to \e1. The use of relative references
2061 can be helpful in long patterns, and also in patterns that are created by
2062 joining together fragments that contain references within themselves.
2063 .P
2064 A back reference matches whatever actually matched the capturing subpattern in
2065 the current subject string, rather than anything matching the subpattern
2066 itself (see
2067 .\" HTML <a href="#subpatternsassubroutines">
2068 .\" </a>
2069 "Subpatterns as subroutines"
2070 .\"
2071 below for a way of doing that). So the pattern
2072 .sp
2073 (sens|respons)e and \e1ibility
2074 .sp
2075 matches "sense and sensibility" and "response and responsibility", but not
2076 "sense and responsibility". If caseful matching is in force at the time of the
2077 back reference, the case of letters is relevant. For example,
2078 .sp
2079 ((?i)rah)\es+\e1
2080 .sp
2081 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
2082 capturing subpattern is matched caselessly.
2083 .P
2084 There are several different ways of writing back references to named
2085 subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
2086 \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
2087 back reference syntax, in which \eg can be used for both numeric and named
2088 references, is also supported. We could rewrite the above example in any of
2089 the following ways:
2090 .sp
2091 (?<p1>(?i)rah)\es+\ek<p1>
2092 (?'p1'(?i)rah)\es+\ek{p1}
2093 (?P<p1>(?i)rah)\es+(?P=p1)
2094 (?<p1>(?i)rah)\es+\eg{p1}
2095 .sp
2096 A subpattern that is referenced by name may appear in the pattern before or
2097 after the reference.
2098 .P
2099 There may be more than one back reference to the same subpattern. If a
2100 subpattern has not actually been used in a particular match, any back
2101 references to it always fail by default. For example, the pattern
2102 .sp
2103 (a|(bc))\e2
2104 .sp
2105 always fails if it starts to match "a" rather than "bc". However, if the
2106 PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
2107 unset value matches an empty string.
2108 .P
2109 Because there may be many capturing parentheses in a pattern, all digits
2110 following a backslash are taken as part of a potential back reference number.
2111 If the pattern continues with a digit character, some delimiter must be used to
2112 terminate the back reference. If the PCRE_EXTENDED option is set, this can be
2113 white space. Otherwise, the \eg{ syntax or an empty comment (see
2114 .\" HTML <a href="#comments">
2115 .\" </a>
2116 "Comments"
2117 .\"
2118 below) can be used.
2119 .
2120 .SS "Recursive back references"
2121 .rs
2122 .sp
2123 A back reference that occurs inside the parentheses to which it refers fails
2124 when the subpattern is first used, so, for example, (a\e1) never matches.
2125 However, such references can be useful inside repeated subpatterns. For
2126 example, the pattern
2127 .sp
2128 (a|b\e1)+
2129 .sp
2130 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
2131 the subpattern, the back reference matches the character string corresponding
2132 to the previous iteration. In order for this to work, the pattern must be such
2133 that the first iteration does not need to match the back reference. This can be
2134 done using alternation, as in the example above, or by a quantifier with a
2135 minimum of zero.
2136 .P
2137 Back references of this type cause the group that they reference to be treated
2138 as an
2139 .\" HTML <a href="#atomicgroup">
2140 .\" </a>
2141 atomic group.
2142 .\"
2143 Once the whole group has been matched, a subsequent matching failure cannot
2144 cause backtracking into the middle of the group.
2145 .
2146 .
2147 .\" HTML <a name="bigassertions"></a>
2149 .rs
2150 .sp
2151 An assertion is a test on the characters following or preceding the current
2152 matching point that does not actually consume any characters. The simple
2153 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
2154 .\" HTML <a href="#smallassertions">
2155 .\" </a>
2156 above.
2157 .\"
2158 .P
2159 More complicated assertions are coded as subpatterns. There are two kinds:
2160 those that look ahead of the current position in the subject string, and those
2161 that look behind it. An assertion subpattern is matched in the normal way,
2162 except that it does not cause the current matching position to be changed.
2163 .P
2164 Assertion subpatterns are not capturing subpatterns. If such an assertion
2165 contains capturing subpatterns within it, these are counted for the purposes of
2166 numbering the capturing subpatterns in the whole pattern. However, substring
2167 capturing is carried out only for positive assertions. (Perl sometimes, but not
2168 always, does do capturing in negative assertions.)
2169 .P
2170 For compatibility with Perl, assertion subpatterns may be repeated; though
2171 it makes no sense to assert the same thing several times, the side effect of
2172 capturing parentheses may occasionally be useful. In practice, there only three
2173 cases:
2174 .sp
2175 (1) If the quantifier is {0}, the assertion is never obeyed during matching.
2176 However, it may contain internal capturing parenthesized groups that are called
2177 from elsewhere via the
2178 .\" HTML <a href="#subpatternsassubroutines">
2179 .\" </a>
2180 subroutine mechanism.
2181 .\"
2182 .sp
2183 (2) If quantifier is {0,n} where n is greater than zero, it is treated as if it
2184 were {0,1}. At run time, the rest of the pattern match is tried with and
2185 without the assertion, the order depending on the greediness of the quantifier.
2186 .sp
2187 (3) If the minimum repetition is greater than zero, the quantifier is ignored.
2188 The assertion is obeyed just once when encountered during matching.
2189 .
2190 .
2191 .SS "Lookahead assertions"
2192 .rs
2193 .sp
2194 Lookahead assertions start with (?= for positive assertions and (?! for
2195 negative assertions. For example,
2196 .sp
2197 \ew+(?=;)
2198 .sp
2199 matches a word followed by a semicolon, but does not include the semicolon in
2200 the match, and
2201 .sp
2202 foo(?!bar)
2203 .sp
2204 matches any occurrence of "foo" that is not followed by "bar". Note that the
2205 apparently similar pattern
2206 .sp
2207 (?!foo)bar
2208 .sp
2209 does not find an occurrence of "bar" that is preceded by something other than
2210 "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
2211 (?!foo) is always true when the next three characters are "bar". A
2212 lookbehind assertion is needed to achieve the other effect.
2213 .P
2214 If you want to force a matching failure at some point in a pattern, the most
2215 convenient way to do it is with (?!) because an empty string always matches, so
2216 an assertion that requires there not to be an empty string must always fail.
2217 The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).
2218 .
2219 .
2220 .\" HTML <a name="lookbehind"></a>
2221 .SS "Lookbehind assertions"
2222 .rs
2223 .sp
2224 Lookbehind assertions start with (?<= for positive assertions and (?<! for
2225 negative assertions. For example,
2226 .sp
2227 (?<!foo)bar
2228 .sp
2229 does find an occurrence of "bar" that is not preceded by "foo". The contents of
2230 a lookbehind assertion are restricted such that all the strings it matches must
2231 have a fixed length. However, if there are several top-level alternatives, they
2232 do not all have to have the same fixed length. Thus
2233 .sp
2234 (?<=bullock|donkey)
2235 .sp
2236 is permitted, but
2237 .sp
2238 (?<!dogs?|cats?)
2239 .sp
2240 causes an error at compile time. Branches that match different length strings
2241 are permitted only at the top level of a lookbehind assertion. This is an
2242 extension compared with Perl, which requires all branches to match the same
2243 length of string. An assertion such as
2244 .sp
2245 (?<=ab(c|de))
2246 .sp
2247 is not permitted, because its single top-level branch can match two different
2248 lengths, but it is acceptable to PCRE if rewritten to use two top-level
2249 branches:
2250 .sp
2251 (?<=abc|abde)
2252 .sp
2253 In some cases, the escape sequence \eK
2254 .\" HTML <a href="#resetmatchstart">
2255 .\" </a>
2256 (see above)
2257 .\"
2258 can be used instead of a lookbehind assertion to get round the fixed-length
2259 restriction.
2260 .P
2261 The implementation of lookbehind assertions is, for each alternative, to
2262 temporarily move the current position back by the fixed length and then try to
2263 match. If there are insufficient characters before the current position, the
2264 assertion fails.
2265 .P
2266 In a UTF mode, PCRE does not allow the \eC escape (which matches a single data
2267 unit even in a UTF mode) to appear in lookbehind assertions, because it makes
2268 it impossible to calculate the length of the lookbehind. The \eX and \eR
2269 escapes, which can match different numbers of data units, are also not
2270 permitted.
2271 .P
2272 .\" HTML <a href="#subpatternsassubroutines">
2273 .\" </a>
2274 "Subroutine"
2275 .\"
2276 calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
2277 as the subpattern matches a fixed-length string.
2278 .\" HTML <a href="#recursion">
2279 .\" </a>
2280 Recursion,
2281 .\"
2282 however, is not supported.
2283 .P
2284 Possessive quantifiers can be used in conjunction with lookbehind assertions to
2285 specify efficient matching of fixed-length strings at the end of subject
2286 strings. Consider a simple pattern such as
2287 .sp
2288 abcd$
2289 .sp
2290 when applied to a long string that does not match. Because matching proceeds
2291 from left to right, PCRE will look for each "a" in the subject and then see if
2292 what follows matches the rest of the pattern. If the pattern is specified as
2293 .sp
2294 ^.*abcd$
2295 .sp
2296 the initial .* matches the entire string at first, but when this fails (because
2297 there is no following "a"), it backtracks to match all but the last character,
2298 then all but the last two characters, and so on. Once again the search for "a"
2299 covers the entire string, from right to left, so we are no better off. However,
2300 if the pattern is written as
2301 .sp
2302 ^.*+(?<=abcd)
2303 .sp
2304 there can be no backtracking for the .*+ item; it can match only the entire
2305 string. The subsequent lookbehind assertion does a single test on the last four
2306 characters. If it fails, the match fails immediately. For long strings, this
2307 approach makes a significant difference to the processing time.
2308 .
2309 .
2310 .SS "Using multiple assertions"
2311 .rs
2312 .sp
2313 Several assertions (of any sort) may occur in succession. For example,
2314 .sp
2315 (?<=\ed{3})(?<!999)foo
2316 .sp
2317 matches "foo" preceded by three digits that are not "999". Notice that each of
2318 the assertions is applied independently at the same point in the subject
2319 string. First there is a check that the previous three characters are all
2320 digits, and then there is a check that the same three characters are not "999".
2321 This pattern does \fInot\fP match "foo" preceded by six characters, the first
2322 of which are digits and the last three of which are not "999". For example, it
2323 doesn't match "123abcfoo". A pattern to do that is
2324 .sp
2325 (?<=\ed{3}...)(?<!999)foo
2326 .sp
2327 This time the first assertion looks at the preceding six characters, checking
2328 that the first three are digits, and then the second assertion checks that the
2329 preceding three characters are not "999".
2330 .P
2331 Assertions can be nested in any combination. For example,
2332 .sp
2333 (?<=(?<!foo)bar)baz
2334 .sp
2335 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
2336 preceded by "foo", while
2337 .sp
2338 (?<=\ed{3}(?!999)...)foo
2339 .sp
2340 is another pattern that matches "foo" preceded by three digits and any three
2341 characters that are not "999".
2342 .
2343 .
2344 .\" HTML <a name="conditions"></a>
2346 .rs
2347 .sp
2348 It is possible to cause the matching process to obey a subpattern
2349 conditionally or to choose between two alternative subpatterns, depending on
2350 the result of an assertion, or whether a specific capturing subpattern has
2351 already been matched. The two possible forms of conditional subpattern are:
2352 .sp
2353 (?(condition)yes-pattern)
2354 (?(condition)yes-pattern|no-pattern)
2355 .sp
2356 If the condition is satisfied, the yes-pattern is used; otherwise the
2357 no-pattern (if present) is used. If there are more than two alternatives in the
2358 subpattern, a compile-time error occurs. Each of the two alternatives may
2359 itself contain nested subpatterns of any form, including conditional
2360 subpatterns; the restriction to two alternatives applies only at the level of
2361 the condition. This pattern fragment is an example where the alternatives are
2362 complex:
2363 .sp
2364 (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
2365 .sp
2366 .P
2367 There are four kinds of condition: references to subpatterns, references to
2368 recursion, a pseudo-condition called DEFINE, and assertions.
2369 .
2370 .SS "Checking for a used subpattern by number"
2371 .rs
2372 .sp
2373 If the text between the parentheses consists of a sequence of digits, the
2374 condition is true if a capturing subpattern of that number has previously
2375 matched. If there is more than one capturing subpattern with the same number
2376 (see the earlier
2377 .\"
2378 .\" HTML <a href="#recursion">
2379 .\" </a>
2380 section about duplicate subpattern numbers),
2381 .\"
2382 the condition is true if any of them have matched. An alternative notation is
2383 to precede the digits with a plus or minus sign. In this case, the subpattern
2384 number is relative rather than absolute. The most recently opened parentheses
2385 can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside
2386 loops it can also make sense to refer to subsequent groups. The next
2387 parentheses to be opened can be referenced as (?(+1), and so on. (The value
2388 zero in any of these forms is not used; it provokes a compile-time error.)
2389 .P
2390 Consider the following pattern, which contains non-significant white space to
2391 make it more readable (assume the PCRE_EXTENDED option) and to divide it into
2392 three parts for ease of discussion:
2393 .sp
2394 ( \e( )? [^()]+ (?(1) \e) )
2395 .sp
2396 The first part matches an optional opening parenthesis, and if that
2397 character is present, sets it as the first captured substring. The second part
2398 matches one or more characters that are not parentheses. The third part is a
2399 conditional subpattern that tests whether or not the first set of parentheses
2400 matched. If they did, that is, if subject started with an opening parenthesis,
2401 the condition is true, and so the yes-pattern is executed and a closing
2402 parenthesis is required. Otherwise, since no-pattern is not present, the
2403 subpattern matches nothing. In other words, this pattern matches a sequence of
2404 non-parentheses, optionally enclosed in parentheses.
2405 .P
2406 If you were embedding this pattern in a larger one, you could use a relative
2407 reference:
2408 .sp
2409 ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
2410 .sp
2411 This makes the fragment independent of the parentheses in the larger pattern.
2412 .
2413 .SS "Checking for a used subpattern by name"
2414 .rs
2415 .sp
2416 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
2417 subpattern by name. For compatibility with earlier versions of PCRE, which had
2418 this facility before Perl, the syntax (?(name)...) is also recognized.
2419 .P
2420 Rewriting the above example to use a named subpattern gives this:
2421 .sp
2422 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
2423 .sp
2424 If the name used in a condition of this kind is a duplicate, the test is
2425 applied to all subpatterns of the same name, and is true if any one of them has
2426 matched.
2427 .
2428 .SS "Checking for pattern recursion"
2429 .rs
2430 .sp
2431 If the condition is the string (R), and there is no subpattern with the name R,
2432 the condition is true if a recursive call to the whole pattern or any
2433 subpattern has been made. If digits or a name preceded by ampersand follow the
2434 letter R, for example:
2435 .sp
2436 (?(R3)...) or (?(R&name)...)
2437 .sp
2438 the condition is true if the most recent recursion is into a subpattern whose
2439 number or name is given. This condition does not check the entire recursion
2440 stack. If the name used in a condition of this kind is a duplicate, the test is
2441 applied to all subpatterns of the same name, and is true if any one of them is
2442 the most recent recursion.
2443 .P
2444 At "top level", all these recursion test conditions are false.
2445 .\" HTML <a href="#recursion">
2446 .\" </a>
2447 The syntax for recursive patterns
2448 .\"
2449 is described below.
2450 .
2451 .\" HTML <a name="subdefine"></a>
2452 .SS "Defining subpatterns for use by reference only"
2453 .rs
2454 .sp
2455 If the condition is the string (DEFINE), and there is no subpattern with the
2456 name DEFINE, the condition is always false. In this case, there may be only one
2457 alternative in the subpattern. It is always skipped if control reaches this
2458 point in the pattern; the idea of DEFINE is that it can be used to define
2459 subroutines that can be referenced from elsewhere. (The use of
2460 .\" HTML <a href="#subpatternsassubroutines">
2461 .\" </a>
2462 subroutines
2463 .\"
2464 is described below.) For example, a pattern to match an IPv4 address such as
2465 "" could be written like this (ignore white space and line
2466 breaks):
2467 .sp
2468 (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
2469 \eb (?&byte) (\e.(?&byte)){3} \eb
2470 .sp
2471 The first part of the pattern is a DEFINE group inside which a another group
2472 named "byte" is defined. This matches an individual component of an IPv4
2473 address (a number less than 256). When matching takes place, this part of the
2474 pattern is skipped because DEFINE acts like a false condition. The rest of the
2475 pattern uses references to the named group to match the four dot-separated
2476 components of an IPv4 address, insisting on a word boundary at each end.
2477 .
2478 .SS "Assertion conditions"
2479 .rs
2480 .sp
2481 If the condition is not in any of the above formats, it must be an assertion.
2482 This may be a positive or negative lookahead or lookbehind assertion. Consider
2483 this pattern, again containing non-significant white space, and with the two
2484 alternatives on the second line:
2485 .sp
2486 (?(?=[^a-z]*[a-z])
2487 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
2488 .sp
2489 The condition is a positive lookahead assertion that matches an optional
2490 sequence of non-letters followed by a letter. In other words, it tests for the
2491 presence of at least one letter in the subject. If a letter is found, the
2492 subject is matched against the first alternative; otherwise it is matched
2493 against the second. This pattern matches strings in one of the two forms
2494 dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2495 .
2496 .
2497 .\" HTML <a name="comments"></a>
2499 .rs
2500 .sp
2501 There are two ways of including comments in patterns that are processed by
2502 PCRE. In both cases, the start of the comment must not be in a character class,
2503 nor in the middle of any other sequence of related characters such as (?: or a
2504 subpattern name or number. The characters that make up a comment play no part
2505 in the pattern matching.
2506 .P
2507 The sequence (?# marks the start of a comment that continues up to the next
2508 closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED
2509 option is set, an unescaped # character also introduces a comment, which in
2510 this case continues to immediately after the next newline character or
2511 character sequence in the pattern. Which characters are interpreted as newlines
2512 is controlled by the options passed to a compiling function or by a special
2513 sequence at the start of the pattern, as described in the section entitled
2514 .\" HTML <a href="#newlines">
2515 .\" </a>
2516 "Newline conventions"
2517 .\"
2518 above. Note that the end of this type of comment is a literal newline sequence
2519 in the pattern; escape sequences that happen to represent a newline do not
2520 count. For example, consider this pattern when PCRE_EXTENDED is set, and the
2521 default newline convention is in force:
2522 .sp
2523 abc #comment \en still comment
2524 .sp
2525 On encountering the # character, \fBpcre_compile()\fP skips along, looking for
2526 a newline in the pattern. The sequence \en is still literal at this stage, so
2527 it does not terminate the comment. Only an actual character with the code value
2528 0x0a (the default newline) does so.
2529 .
2530 .
2531 .\" HTML <a name="recursion"></a>
2533 .rs
2534 .sp
2535 Consider the problem of matching a string in parentheses, allowing for
2536 unlimited nested parentheses. Without the use of recursion, the best that can
2537 be done is to use a pattern that matches up to some fixed depth of nesting. It
2538 is not possible to handle an arbitrary nesting depth.
2539 .P
2540 For some time, Perl has provided a facility that allows regular expressions to
2541 recurse (amongst other things). It does this by interpolating Perl code in the
2542 expression at run time, and the code can refer to the expression itself. A Perl
2543 pattern using code interpolation to solve the parentheses problem can be
2544 created like this:
2545 .sp
2546 $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
2547 .sp
2548 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2549 recursively to the pattern in which it appears.
2550 .P
2551 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2552 supports special syntax for recursion of the entire pattern, and also for
2553 individual subpattern recursion. After its introduction in PCRE and Python,
2554 this kind of recursion was subsequently introduced into Perl at release 5.10.
2555 .P
2556 A special item that consists of (? followed by a number greater than zero and a
2557 closing parenthesis is a recursive subroutine call of the subpattern of the
2558 given number, provided that it occurs inside that subpattern. (If not, it is a
2559 .\" HTML <a href="#subpatternsassubroutines">
2560 .\" </a>
2561 non-recursive subroutine
2562 .\"
2563 call, which is described in the next section.) The special item (?R) or (?0) is
2564 a recursive call of the entire regular expression.
2565 .P
2566 This PCRE pattern solves the nested parentheses problem (assume the
2567 PCRE_EXTENDED option is set so that white space is ignored):
2568 .sp
2569 \e( ( [^()]++ | (?R) )* \e)
2570 .sp
2571 First it matches an opening parenthesis. Then it matches any number of
2572 substrings which can either be a sequence of non-parentheses, or a recursive
2573 match of the pattern itself (that is, a correctly parenthesized substring).
2574 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2575 to avoid backtracking into sequences of non-parentheses.
2576 .P
2577 If this were part of a larger pattern, you would not want to recurse the entire
2578 pattern, so instead you could use this:
2579 .sp
2580 ( \e( ( [^()]++ | (?1) )* \e) )
2581 .sp
2582 We have put the pattern into parentheses, and caused the recursion to refer to
2583 them instead of the whole pattern.
2584 .P
2585 In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2586 is made easier by the use of relative references. Instead of (?1) in the
2587 pattern above you can write (?-2) to refer to the second most recently opened
2588 parentheses preceding the recursion. In other words, a negative number counts
2589 capturing parentheses leftwards from the point at which it is encountered.
2590 .P
2591 It is also possible to refer to subsequently opened parentheses, by writing
2592 references such as (?+2). However, these cannot be recursive because the
2593 reference is not inside the parentheses that are referenced. They are always
2594 .\" HTML <a href="#subpatternsassubroutines">
2595 .\" </a>
2596 non-recursive subroutine
2597 .\"
2598 calls, as described in the next section.
2599 .P
2600 An alternative approach is to use named parentheses instead. The Perl syntax
2601 for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
2602 could rewrite the above example as follows:
2603 .sp
2604 (?<pn> \e( ( [^()]++ | (?&pn) )* \e) )
2605 .sp
2606 If there is more than one subpattern with the same name, the earliest one is
2607 used.
2608 .P
2609 This particular example pattern that we have been looking at contains nested
2610 unlimited repeats, and so the use of a possessive quantifier for matching
2611 strings of non-parentheses is important when applying the pattern to strings
2612 that do not match. For example, when this pattern is applied to
2613 .sp
2614 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2615 .sp
2616 it yields "no match" quickly. However, if a possessive quantifier is not used,
2617 the match runs for a very long time indeed because there are so many different
2618 ways the + and * repeats can carve up the subject, and all have to be tested
2619 before failure can be reported.
2620 .P
2621 At the end of a match, the values of capturing parentheses are those from
2622 the outermost level. If you want to obtain intermediate values, a callout
2623 function can be used (see below and the
2624 .\" HREF
2625 \fBpcrecallout\fP
2626 .\"
2627 documentation). If the pattern above is matched against
2628 .sp
2629 (ab(cd)ef)
2630 .sp
2631 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2632 the last value taken on at the top level. If a capturing subpattern is not
2633 matched at the top level, its final captured value is unset, even if it was
2634 (temporarily) set at a deeper level during the matching process.
2635 .P
2636 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2637 obtain extra memory to store data during a recursion, which it does by using
2638 \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no memory can
2639 be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2640 .P
2641 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2642 Consider this pattern, which matches text in angle brackets, allowing for
2643 arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2644 recursing), whereas any characters are permitted at the outer level.
2645 .sp
2646 < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
2647 .sp
2648 In this pattern, (?(R) is the start of a conditional subpattern, with two
2649 different alternatives for the recursive and non-recursive cases. The (?R) item
2650 is the actual recursive call.
2651 .
2652 .
2653 .\" HTML <a name="recursiondifference"></a>
2654 .SS "Differences in recursion processing between PCRE and Perl"
2655 .rs
2656 .sp
2657 Recursion processing in PCRE differs from Perl in two important ways. In PCRE
2658 (like Python, but unlike Perl), a recursive subpattern call is always treated
2659 as an atomic group. That is, once it has matched some of the subject string, it
2660 is never re-entered, even if it contains untried alternatives and there is a
2661 subsequent matching failure. This can be illustrated by the following pattern,
2662 which purports to match a palindromic string that contains an odd number of
2663 characters (for example, "a", "aba", "abcba", "abcdcba"):
2664 .sp
2665 ^(.|(.)(?1)\e2)$
2666 .sp
2667 The idea is that it either matches a single character, or two identical
2668 characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2669 it does not if the pattern is longer than three characters. Consider the
2670 subject string "abcba":
2671 .P
2672 At the top level, the first character is matched, but as it is not at the end
2673 of the string, the first alternative fails; the second alternative is taken
2674 and the recursion kicks in. The recursive call to subpattern 1 successfully
2675 matches the next character ("b"). (Note that the beginning and end of line
2676 tests are not part of the recursion).
2677 .P
2678 Back at the top level, the next character ("c") is compared with what
2679 subpattern 2 matched, which was "a". This fails. Because the recursion is
2680 treated as an atomic group, there are now no backtracking points, and so the
2681 entire match fails. (Perl is able, at this point, to re-enter the recursion and
2682 try the second alternative.) However, if the pattern is written with the
2683 alternatives in the other order, things are different:
2684 .sp
2685 ^((.)(?1)\e2|.)$
2686 .sp
2687 This time, the recursing alternative is tried first, and continues to recurse
2688 until it runs out of characters, at which point the recursion fails. But this
2689 time we do have another alternative to try at the higher level. That is the big
2690 difference: in the previous case the remaining alternative is at a deeper
2691 recursion level, which PCRE cannot use.
2692 .P
2693 To change the pattern so that it matches all palindromic strings, not just
2694 those with an odd number of characters, it is tempting to change the pattern to
2695 this:
2696 .sp
2697 ^((.)(?1)\e2|.?)$
2698 .sp
2699 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2700 deeper recursion has matched a single character, it cannot be entered again in
2701 order to match an empty string. The solution is to separate the two cases, and
2702 write out the odd and even cases as alternatives at the higher level:
2703 .sp
2704 ^(?:((.)(?1)\e2|)|((.)(?3)\e4|.))
2705 .sp
2706 If you want to match typical palindromic phrases, the pattern has to ignore all
2707 non-word characters, which can be done like this:
2708 .sp
2709 ^\eW*+(?:((.)\eW*+(?1)\eW*+\e2|)|((.)\eW*+(?3)\eW*+\e4|\eW*+.\eW*+))\eW*+$
2710 .sp
2711 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2712 man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2713 the use of the possessive quantifier *+ to avoid backtracking into sequences of
2714 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2715 more) to match typical phrases, and Perl takes so long that you think it has
2716 gone into a loop.
2717 .P
2718 \fBWARNING\fP: The palindrome-matching patterns above work only if the subject
2719 string does not start with a palindrome that is shorter than the entire string.
2720 For example, although "abcba" is correctly matched, if the subject is "ababa",
2721 PCRE finds the palindrome "aba" at the start, then fails at top level because
2722 the end of the string does not follow. Once again, it cannot jump back into the
2723 recursion to try other alternatives, so the entire match fails.
2724 .P
2725 The second way in which PCRE and Perl differ in their recursion processing is
2726 in the handling of captured values. In Perl, when a subpattern is called
2727 recursively or as a subpattern (see the next section), it has no access to any
2728 values that were captured outside the recursion, whereas in PCRE these values
2729 can be referenced. Consider this pattern:
2730 .sp
2731 ^(.)(\e1|a(?2))
2732 .sp
2733 In PCRE, this pattern matches "bab". The first capturing parentheses match "b",
2734 then in the second group, when the back reference \e1 fails to match "b", the
2735 second alternative matches "a" and then recurses. In the recursion, \e1 does
2736 now match "b" and so the whole match succeeds. In Perl, the pattern fails to
2737 match because inside the recursive call \e1 cannot access the externally set
2738 value.
2739 .
2740 .
2741 .\" HTML <a name="subpatternsassubroutines"></a>
2743 .rs
2744 .sp
2745 If the syntax for a recursive subpattern call (either by number or by
2746 name) is used outside the parentheses to which it refers, it operates like a
2747 subroutine in a programming language. The called subpattern may be defined
2748 before or after the reference. A numbered reference can be absolute or
2749 relative, as in these examples:
2750 .sp
2751 (...(absolute)...)...(?2)...
2752 (...(relative)...)...(?-1)...
2753 (...(?+1)...(relative)...
2754 .sp
2755 An earlier example pointed out that the pattern
2756 .sp
2757 (sens|respons)e and \e1ibility
2758 .sp
2759 matches "sense and sensibility" and "response and responsibility", but not
2760 "sense and responsibility". If instead the pattern
2761 .sp
2762 (sens|respons)e and (?1)ibility
2763 .sp
2764 is used, it does match "sense and responsibility" as well as the other two
2765 strings. Another example is given in the discussion of DEFINE above.
2766 .P
2767 All subroutine calls, whether recursive or not, are always treated as atomic
2768 groups. That is, once a subroutine has matched some of the subject string, it
2769 is never re-entered, even if it contains untried alternatives and there is a
2770 subsequent matching failure. Any capturing parentheses that are set during the
2771 subroutine call revert to their previous values afterwards.
2772 .P
2773 Processing options such as case-independence are fixed when a subpattern is
2774 defined, so if it is used as a subroutine, such options cannot be changed for
2775 different calls. For example, consider this pattern:
2776 .sp
2777 (abc)(?i:(?-1))
2778 .sp
2779 It matches "abcabc". It does not match "abcABC" because the change of
2780 processing option does not affect the called subpattern.
2781 .
2782 .
2783 .\" HTML <a name="onigurumasubroutines"></a>
2785 .rs
2786 .sp
2787 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
2788 a number enclosed either in angle brackets or single quotes, is an alternative
2789 syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2790 are two of the examples used above, rewritten using this syntax:
2791 .sp
2792 (?<pn> \e( ( (?>[^()]+) | \eg<pn> )* \e) )
2793 (sens|respons)e and \eg'1'ibility
2794 .sp
2795 PCRE supports an extension to Oniguruma: if a number is preceded by a
2796 plus or a minus sign it is taken as a relative reference. For example:
2797 .sp
2798 (abc)(?i:\eg<-1>)
2799 .sp
2800 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
2801 synonymous. The former is a back reference; the latter is a subroutine call.
2802 .
2803 .
2805 .rs
2806 .sp
2807 Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2808 code to be obeyed in the middle of matching a regular expression. This makes it
2809 possible, amongst other things, to extract different substrings that match the
2810 same pair of parentheses when there is a repetition.
2811 .P
2812 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2813 code. The feature is called "callout". The caller of PCRE provides an external
2814 function by putting its entry point in the global variable \fIpcre_callout\fP
2815 (8-bit library) or \fIpcre[16|32]_callout\fP (16-bit or 32-bit library).
2816 By default, this variable contains NULL, which disables all calling out.
2817 .P
2818 Within a regular expression, (?C) indicates the points at which the external
2819 function is to be called. If you want to identify different callout points, you
2820 can put a number less than 256 after the letter C. The default value is zero.
2821 For example, this pattern has two callout points:
2822 .sp
2823 (?C1)abc(?C2)def
2824 .sp
2825 If the PCRE_AUTO_CALLOUT flag is passed to a compiling function, callouts are
2826 automatically installed before each item in the pattern. They are all numbered
2827 255. If there is a conditional group in the pattern whose condition is an
2828 assertion, an additional callout is inserted just before the condition. An
2829 explicit callout may also be set at this position, as in this example:
2830 .sp
2831 (?(?C9)(?=a)abc|def)
2832 .sp
2833 Note that this applies only to assertion conditions, not to other types of
2834 condition.
2835 .P
2836 During matching, when PCRE reaches a callout point, the external function is
2837 called. It is provided with the number of the callout, the position in the
2838 pattern, and, optionally, one item of data originally supplied by the caller of
2839 the matching function. The callout function may cause matching to proceed, to
2840 backtrack, or to fail altogether.
2841 .P
2842 By default, PCRE implements a number of optimizations at compile time and
2843 matching time, and one side-effect is that sometimes callouts are skipped. If
2844 you need all possible callouts to happen, you need to set options that disable
2845 the relevant optimizations. More details, and a complete description of the
2846 interface to the callout function, are given in the
2847 .\" HREF
2848 \fBpcrecallout\fP
2849 .\"
2850 documentation.
2851 .
2852 .
2853 .\" HTML <a name="backtrackcontrol"></a>
2855 .rs
2856 .sp
2857 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2858 are still described in the Perl documentation as "experimental and subject to
2859 change or removal in a future version of Perl". It goes on to say: "Their usage
2860 in production code should be noted to avoid problems during upgrades." The same
2861 remarks apply to the PCRE features described in this section.
2862 .P
2863 The new verbs make use of what was previously invalid syntax: an opening
2864 parenthesis followed by an asterisk. They are generally of the form
2865 (*VERB) or (*VERB:NAME). Some may take either form, possibly behaving
2866 differently depending on whether or not a name is present. A name is any
2867 sequence of characters that does not include a closing parenthesis. The maximum
2868 length of name is 255 in the 8-bit library and 65535 in the 16-bit and 32-bit
2869 libraries. If the name is empty, that is, if the closing parenthesis
2870 immediately follows the colon, the effect is as if the colon were not there.
2871 Any number of these verbs may occur in a pattern.
2872 .P
2873 Since these verbs are specifically related to backtracking, most of them can be
2874 used only when the pattern is to be matched using one of the traditional
2875 matching functions, because these use a backtracking algorithm. With the
2876 exception of (*FAIL), which behaves like a failing negative assertion, the
2877 backtracking control verbs cause an error if encountered by a DFA matching
2878 function.
2879 .P
2880 The behaviour of these verbs in
2881 .\" HTML <a href="#btrepeat">
2882 .\" </a>
2883 repeated groups,
2884 .\"
2885 .\" HTML <a href="#btassert">
2886 .\" </a>
2887 assertions,
2888 .\"
2889 and in
2890 .\" HTML <a href="#btsub">
2891 .\" </a>
2892 subpatterns called as subroutines
2893 .\"
2894 (whether or not recursively) is documented below.
2895 .
2896 .
2897 .\" HTML <a name="nooptimize"></a>
2898 .SS "Optimizations that affect backtracking verbs"
2899 .rs
2900 .sp
2901 PCRE contains some optimizations that are used to speed up matching by running
2902 some checks at the start of each match attempt. For example, it may know the
2903 minimum length of matching subject, or that a particular character must be
2904 present. When one of these optimizations bypasses the running of a match, any
2905 included backtracking verbs will not, of course, be processed. You can suppress
2906 the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2907 when calling \fBpcre_compile()\fP or \fBpcre_exec()\fP, or by starting the
2908 pattern with (*NO_START_OPT). There is more discussion of this option in the
2909 section entitled
2910 .\" HTML <a href="pcreapi.html#execoptions">
2911 .\" </a>
2912 "Option bits for \fBpcre_exec()\fP"
2913 .\"
2914 in the
2915 .\" HREF
2916 \fBpcreapi\fP
2917 .\"
2918 documentation.
2919 .P
2920 Experiments with Perl suggest that it too has similar optimizations, sometimes
2921 leading to anomalous results.
2922 .
2923 .
2924 .SS "Verbs that act immediately"
2925 .rs
2926 .sp
2927 The following verbs act as soon as they are encountered. They may not be
2928 followed by a name.
2929 .sp
2930 (*ACCEPT)
2931 .sp
2932 This verb causes the match to end successfully, skipping the remainder of the
2933 pattern. However, when it is inside a subpattern that is called as a
2934 subroutine, only that subpattern is ended successfully. Matching then continues
2935 at the outer level. If (*ACCEPT) in triggered in a positive assertion, the
2936 assertion succeeds; in a negative assertion, the assertion fails.
2937 .P
2938 If (*ACCEPT) is inside capturing parentheses, the data so far is captured. For
2939 example:
2940 .sp
2941 A((?:A|B(*ACCEPT)|C)D)
2942 .sp
2943 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2944 the outer parentheses.
2945 .sp
2946 (*FAIL) or (*F)
2947 .sp
2948 This verb causes a matching failure, forcing backtracking to occur. It is
2949 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2950 probably useful only when combined with (?{}) or (??{}). Those are, of course,
2951 Perl features that are not present in PCRE. The nearest equivalent is the
2952 callout feature, as for example in this pattern:
2953 .sp
2954 a+(?C)(*FAIL)
2955 .sp
2956 A match with the string "aaaa" always fails, but the callout is taken before
2957 each backtrack happens (in this example, 10 times).
2958 .
2959 .
2960 .SS "Recording which path was taken"
2961 .rs
2962 .sp
2963 There is one verb whose main purpose is to track how a match was arrived at,
2964 though it also has a secondary use in conjunction with advancing the match
2965 starting point (see (*SKIP) below).
2966 .sp
2967 (*MARK:NAME) or (*:NAME)
2968 .sp
2969 A name is always required with this verb. There may be as many instances of
2970 (*MARK) as you like in a pattern, and their names do not have to be unique.
2971 .P
2972 When a match succeeds, the name of the last-encountered (*MARK:NAME),
2973 (*PRUNE:NAME), or (*THEN:NAME) on the matching path is passed back to the
2974 caller as described in the section entitled
2975 .\" HTML <a href="pcreapi.html#extradata">
2976 .\" </a>
2977 "Extra data for \fBpcre_exec()\fP"
2978 .\"
2979 in the
2980 .\" HREF
2981 \fBpcreapi\fP
2982 .\"
2983 documentation. Here is an example of \fBpcretest\fP output, where the /K
2984 modifier requests the retrieval and outputting of (*MARK) data:
2985 .sp
2986 re> /X(*MARK:A)Y|X(*MARK:B)Z/K
2987 data> XY
2988 0: XY
2989 MK: A
2990 XZ
2991 0: XZ
2992 MK: B
2993 .sp
2994 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2995 indicates which of the two alternatives matched. This is a more efficient way
2996 of obtaining this information than putting each alternative in its own
2997 capturing parentheses.
2998 .P
2999 If a verb with a name is encountered in a positive assertion that is true, the
3000 name is recorded and passed back if it is the last-encountered. This does not
3001 happen for negative assertions or failing positive assertions.
3002 .P
3003 After a partial match or a failed match, the last encountered name in the
3004 entire match process is returned. For example:
3005 .sp
3006 re> /X(*MARK:A)Y|X(*MARK:B)Z/K
3007 data> XP
3008 No match, mark = B
3009 .sp
3010 Note that in this unanchored example the mark is retained from the match
3011 attempt that started at the letter "X" in the subject. Subsequent match
3012 attempts starting at "P" and then with an empty string do not get as far as the
3013 (*MARK) item, but nevertheless do not reset it.
3014 .P
3015 If you are interested in (*MARK) values after failed matches, you should
3016 probably set the PCRE_NO_START_OPTIMIZE option
3017 .\" HTML <a href="#nooptimize">
3018 .\" </a>
3019 (see above)
3020 .\"
3021 to ensure that the match is always attempted.
3022 .
3023 .
3024 .SS "Verbs that act after backtracking"
3025 .rs
3026 .sp
3027 The following verbs do nothing when they are encountered. Matching continues
3028 with what follows, but if there is no subsequent match, causing a backtrack to
3029 the verb, a failure is forced. That is, backtracking cannot pass to the left of
3030 the verb. However, when one of these verbs appears inside an atomic group or an
3031 assertion that is true, its effect is confined to that group, because once the
3032 group has been matched, there is never any backtracking into it. In this
3033 situation, backtracking can "jump back" to the left of the entire atomic group
3034 or assertion. (Remember also, as stated above, that this localization also
3035 applies in subroutine calls.)
3036 .P
3037 These verbs differ in exactly what kind of failure occurs when backtracking
3038 reaches them. The behaviour described below is what happens when the verb is
3039 not in a subroutine or an assertion. Subsequent sections cover these special
3040 cases.
3041 .sp
3042 (*COMMIT)
3043 .sp
3044 This verb, which may not be followed by a name, causes the whole match to fail
3045 outright if there is a later matching failure that causes backtracking to reach
3046 it. Even if the pattern is unanchored, no further attempts to find a match by
3047 advancing the starting point take place. If (*COMMIT) is the only backtracking
3048 verb that is encountered, once it has been passed \fBpcre_exec()\fP is
3049 committed to finding a match at the current starting point, or not at all. For
3050 example:
3051 .sp
3052 a+(*COMMIT)b
3053 .sp
3054 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
3055 dynamic anchor, or "I've started, so I must finish." The name of the most
3056 recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
3057 match failure.
3058 .P
3059 If there is more than one backtracking verb in a pattern, a different one that
3060 follows (*COMMIT) may be triggered first, so merely passing (*COMMIT) during a
3061 match does not always guarantee that a match must be at this starting point.
3062 .P
3063 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
3064 unless PCRE's start-of-match optimizations are turned off, as shown in this
3065 output from \fBpcretest\fP:
3066 .sp
3067 re> /(*COMMIT)abc/
3068 data> xyzabc
3069 0: abc
3070 data> xyzabc\eY
3071 No match
3072 .sp
3073 For this pattern, PCRE knows that any match must start with "a", so the
3074 optimization skips along the subject to "a" before applying the pattern to the
3075 first set of data. The match attempt then succeeds. In the second set of data,
3076 the escape sequence \eY is interpreted by the \fBpcretest\fP program. It causes
3077 the PCRE_NO_START_OPTIMIZE option to be set when \fBpcre_exec()\fP is called.
3078 This disables the optimization that skips along to the first character. The
3079 pattern is now applied starting at "x", and so the (*COMMIT) causes the match
3080 to fail without trying any other starting points.
3081 .sp
3082 (*PRUNE) or (*PRUNE:NAME)
3083 .sp
3084 This verb causes the match to fail at the current starting position in the
3085 subject if there is a later matching failure that causes backtracking to reach
3086 it. If the pattern is unanchored, the normal "bumpalong" advance to the next
3087 starting character then happens. Backtracking can occur as usual to the left of
3088 (*PRUNE), before it is reached, or when matching to the right of (*PRUNE), but
3089 if there is no match to the right, backtracking cannot cross (*PRUNE). In
3090 simple cases, the use of (*PRUNE) is just an alternative to an atomic group or
3091 possessive quantifier, but there are some uses of (*PRUNE) that cannot be
3092 expressed in any other way. In an anchored pattern (*PRUNE) has the same effect
3093 as (*COMMIT).
3094 .P
3095 The behaviour of (*PRUNE:NAME) is the not the same as (*MARK:NAME)(*PRUNE).
3096 It is like (*MARK:NAME) in that the name is remembered for passing back to the
3097 caller. However, (*SKIP:NAME) searches only for names set with (*MARK).
3098 .sp
3099 (*SKIP)
3100 .sp
3101 This verb, when given without a name, is like (*PRUNE), except that if the
3102 pattern is unanchored, the "bumpalong" advance is not to the next character,
3103 but to the position in the subject where (*SKIP) was encountered. (*SKIP)
3104 signifies that whatever text was matched leading up to it cannot be part of a
3105 successful match. Consider:
3106 .sp
3107 a+(*SKIP)b
3108 .sp
3109 If the subject is "aaaac...", after the first match attempt fails (starting at
3110 the first character in the string), the starting point skips on to start the
3111 next attempt at "c". Note that a possessive quantifer does not have the same
3112 effect as this example; although it would suppress backtracking during the
3113 first match attempt, the second attempt would start at the second character
3114 instead of skipping on to "c".
3115 .sp
3116 (*SKIP:NAME)
3117 .sp
3118 When (*SKIP) has an associated name, its behaviour is modified. When it is
3119 triggered, the previous path through the pattern is searched for the most
3120 recent (*MARK) that has the same name. If one is found, the "bumpalong" advance
3121 is to the subject position that corresponds to that (*MARK) instead of to where
3122 (*SKIP) was encountered. If no (*MARK) with a matching name is found, the
3123 (*SKIP) is ignored.
3124 .P
3125 Note that (*SKIP:NAME) searches only for names set by (*MARK:NAME). It ignores
3126 names that are set by (*PRUNE:NAME) or (*THEN:NAME).
3127 .sp
3128 (*THEN) or (*THEN:NAME)
3129 .sp
3130 This verb causes a skip to the next innermost alternative when backtracking
3131 reaches it. That is, it cancels any further backtracking within the current
3132 alternative. Its name comes from the observation that it can be used for a
3133 pattern-based if-then-else block:
3134 .sp
3135 ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
3136 .sp
3137 If the COND1 pattern matches, FOO is tried (and possibly further items after
3138 the end of the group if FOO succeeds); on failure, the matcher skips to the
3139 second alternative and tries COND2, without backtracking into COND1. If that
3140 succeeds and BAR fails, COND3 is tried. If subsequently BAZ fails, there are no
3141 more alternatives, so there is a backtrack to whatever came before the entire
3142 group. If (*THEN) is not inside an alternation, it acts like (*PRUNE).
3143 .P
3144 The behaviour of (*THEN:NAME) is the not the same as (*MARK:NAME)(*THEN).
3145 It is like (*MARK:NAME) in that the name is remembered for passing back to the
3146 caller. However, (*SKIP:NAME) searches only for names set with (*MARK).
3147 .P
3148 A subpattern that does not contain a | character is just a part of the
3149 enclosing alternative; it is not a nested alternation with only one
3150 alternative. The effect of (*THEN) extends beyond such a subpattern to the
3151 enclosing alternative. Consider this pattern, where A, B, etc. are complex
3152 pattern fragments that do not contain any | characters at this level:
3153 .sp
3154 A (B(*THEN)C) | D
3155 .sp
3156 If A and B are matched, but there is a failure in C, matching does not
3157 backtrack into A; instead it moves to the next alternative, that is, D.
3158 However, if the subpattern containing (*THEN) is given an alternative, it
3159 behaves differently:
3160 .sp
3161 A (B(*THEN)C | (*FAIL)) | D
3162 .sp
3163 The effect of (*THEN) is now confined to the inner subpattern. After a failure
3164 in C, matching moves to (*FAIL), which causes the whole subpattern to fail
3165 because there are no more alternatives to try. In this case, matching does now
3166 backtrack into A.
3167 .P
3168 Note that a conditional subpattern is not considered as having two
3169 alternatives, because only one is ever used. In other words, the | character in
3170 a conditional subpattern has a different meaning. Ignoring white space,
3171 consider:
3172 .sp
3173 ^.*? (?(?=a) a | b(*THEN)c )
3174 .sp
3175 If the subject is "ba", this pattern does not match. Because .*? is ungreedy,
3176 it initially matches zero characters. The condition (?=a) then fails, the
3177 character "b" is matched, but "c" is not. At this point, matching does not
3178 backtrack to .*? as might perhaps be expected from the presence of the |
3179 character. The conditional subpattern is part of the single alternative that
3180 comprises the whole pattern, and so the match fails. (If there was a backtrack
3181 into .*?, allowing it to match "b", the match would succeed.)
3182 .P
3183 The verbs just described provide four different "strengths" of control when
3184 subsequent matching fails. (*THEN) is the weakest, carrying on the match at the
3185 next alternative. (*PRUNE) comes next, failing the match at the current
3186 starting position, but allowing an advance to the next character (for an
3187 unanchored pattern). (*SKIP) is similar, except that the advance may be more
3188 than one character. (*COMMIT) is the strongest, causing the entire match to
3189 fail.
3190 .
3191 .
3192 .SS "More than one backtracking verb"
3193 .rs
3194 .sp
3195 If more than one backtracking verb is present in a pattern, the one that is
3196 backtracked onto first acts. For example, consider this pattern, where A, B,
3197 etc. are complex pattern fragments:
3198 .sp
3200 .sp
3201 If A matches but B fails, the backtrack to (*COMMIT) causes the entire match to
3202 fail. However, if A and B match, but C fails, the backtrack to (*THEN) causes
3203 the next alternative (ABD) to be tried. This behaviour is consistent, but is
3204 not always the same as Perl's. It means that if two or more backtracking verbs
3205 appear in succession, all the the last of them has no effect. Consider this
3206 example:
3207 .sp
3208 ...(*COMMIT)(*PRUNE)...
3209 .sp
3210 If there is a matching failure to the right, backtracking onto (*PRUNE) causes
3211 it to be triggered, and its action is taken. There can never be a backtrack
3212 onto (*COMMIT).
3213 .
3214 .
3215 .\" HTML <a name="btrepeat"></a>
3216 .SS "Backtracking verbs in repeated groups"
3217 .rs
3218 .sp
3219 PCRE differs from Perl in its handling of backtracking verbs in repeated
3220 groups. For example, consider:
3221 .sp
3222 /(a(*COMMIT)b)+ac/
3223 .sp
3224 If the subject is "abac", Perl matches, but PCRE fails because the (*COMMIT) in
3225 the second repeat of the group acts.
3226 .
3227 .
3228 .\" HTML <a name="btassert"></a>
3229 .SS "Backtracking verbs in assertions"
3230 .rs
3231 .sp
3232 (*FAIL) in an assertion has its normal effect: it forces an immediate backtrack.
3233 .P
3234 (*ACCEPT) in a positive assertion causes the assertion to succeed without any
3235 further processing. In a negative assertion, (*ACCEPT) causes the assertion to
3236 fail without any further processing.
3237 .P
3238 The other backtracking verbs are not treated specially if they appear in a
3239 positive assertion. In particular, (*THEN) skips to the next alternative in the
3240 innermost enclosing group that has alternations, whether or not this is within
3241 the assertion.
3242 .P
3243 Negative assertions are, however, different, in order to ensure that changing a
3244 positive assertion into a negative assertion changes its result. Backtracking
3245 into (*COMMIT), (*SKIP), or (*PRUNE) causes a negative assertion to be true,
3246 without considering any further alternative branches in the assertion.
3247 Backtracking into (*THEN) causes it to skip to the next enclosing alternative
3248 within the assertion (the normal behaviour), but if the assertion does not have
3249 such an alternative, (*THEN) behaves like (*PRUNE).
3250 .
3251 .
3252 .\" HTML <a name="btsub"></a>
3253 .SS "Backtracking verbs in subroutines"
3254 .rs
3255 .sp
3256 These behaviours occur whether or not the subpattern is called recursively.
3257 Perl's treatment of subroutines is different in some cases.
3258 .P
3259 (*FAIL) in a subpattern called as a subroutine has its normal effect: it forces
3260 an immediate backtrack.
3261 .P
3262 (*ACCEPT) in a subpattern called as a subroutine causes the subroutine match to
3263 succeed without any further processing. Matching then continues after the
3264 subroutine call.
3265 .P
3266 (*COMMIT), (*SKIP), and (*PRUNE) in a subpattern called as a subroutine cause
3267 the subroutine match to fail.
3268 .P
3269 (*THEN) skips to the next alternative in the innermost enclosing group within
3270 the subpattern that has alternatives. If there is no such group within the
3271 subpattern, (*THEN) causes the subroutine match to fail.
3272 .
3273 .
3274 .SH "SEE ALSO"
3275 .rs
3276 .sp
3277 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3),
3278 \fBpcresyntax\fP(3), \fBpcre\fP(3), \fBpcre16(3)\fP, \fBpcre32(3)\fP.
3279 .
3280 .
3282 .rs
3283 .sp
3284 .nf
3285 Philip Hazel
3286 University Computing Service
3287 Cambridge CB2 3QH, England.
3288 .fi
3289 .
3290 .
3292 .rs
3293 .sp
3294 .nf
3295 Last updated: 14 October 2016
3296 Copyright (c) 1997-2016 University of Cambridge.
3297 .fi


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

  ViewVC Help
Powered by ViewVC 1.1.5