ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 630 - (hide annotations) (download)
Fri Jul 22 10:00:10 2011 UTC (3 years, 9 months ago) by ph10
File size: 114594 byte(s)
Make (*MARK) work in positive assertions.

1 nigel 79 .TH PCREPATTERN 3
2 nigel 63 .SH NAME
3     PCRE - Perl-compatible regular expressions
5 nigel 63 .rs
6     .sp
7 ph10 208 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 ph10 333 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 ph10 208 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 nigel 75 .P
23     The original operation of PCRE was on strings of one-byte characters. However,
24 ph10 461 there is now also support for UTF-8 character strings. To use this,
25 ph10 456 PCRE must be built to include UTF-8 support, and you must call
26     \fBpcre_compile()\fP or \fBpcre_compile2()\fP with the PCRE_UTF8 option. There
27     is also a special sequence that can be given at the start of a pattern:
28 ph10 412 .sp
29     (*UTF8)
30 ph10 416 .sp
31 ph10 412 Starting a pattern with this sequence is equivalent to setting the PCRE_UTF8
32     option. This feature is not Perl-compatible. How setting UTF-8 mode affects
33     pattern matching is mentioned in several places below. There is also a summary
34     of UTF-8 features in the
35 nigel 63 .\" HTML <a href="pcre.html#utf8support">
36     .\" </a>
37     section on UTF-8 support
38     .\"
39     in the main
40     .\" HREF
41 nigel 75 \fBpcre\fP
42 nigel 63 .\"
43     page.
44 nigel 75 .P
45 ph10 535 Another special sequence that may appear at the start of a pattern or in
46 ph10 518 combination with (*UTF8) is:
47     .sp
48     (*UCP)
49     .sp
50 ph10 535 This has the same effect as setting the PCRE_UCP option: it causes sequences
51     such as \ed and \ew to use Unicode properties to determine character types,
52     instead of recognizing only characters with codes less than 128 via a lookup
53 ph10 518 table.
54     .P
55 ph10 576 If a pattern starts with (*NO_START_OPT), it has the same effect as setting the
56 ph10 579 PCRE_NO_START_OPTIMIZE option either at compile or matching time. There are
57     also some more of these special sequences that are concerned with the handling
58 ph10 576 of newlines; they are described below.
59     .P
60 nigel 77 The remainder of this document discusses the patterns that are supported by
61     PCRE when its main matching function, \fBpcre_exec()\fP, is used.
62     From release 6.0, PCRE offers a second matching function,
63     \fBpcre_dfa_exec()\fP, which matches using a different algorithm that is not
64 ph10 172 Perl-compatible. Some of the features discussed below are not available when
65 ph10 168 \fBpcre_dfa_exec()\fP is used. The advantages and disadvantages of the
66     alternative function, and how it differs from the normal function, are
67     discussed in the
68 nigel 77 .\" HREF
69     \fBpcrematching\fP
70     .\"
71     page.
72 nigel 93 .
73     .
74 ph10 556 .\" HTML <a name="newlines"></a>
76     .rs
77     .sp
78     PCRE supports five different conventions for indicating line breaks in
79     strings: a single CR (carriage return) character, a single LF (linefeed)
80     character, the two-character sequence CRLF, any of the three preceding, or any
81     Unicode newline sequence. The
82     .\" HREF
83     \fBpcreapi\fP
84     .\"
85     page has
86     .\" HTML <a href="pcreapi.html#newlines">
87     .\" </a>
88     further discussion
89     .\"
90     about newlines, and shows how to set the newline convention in the
91     \fIoptions\fP arguments for the compiling and matching functions.
92     .P
93     It is also possible to specify a newline convention by starting a pattern
94     string with one of the following five sequences:
95     .sp
96     (*CR) carriage return
97     (*LF) linefeed
98     (*CRLF) carriage return, followed by linefeed
99     (*ANYCRLF) any of the three above
100     (*ANY) all Unicode newline sequences
101     .sp
102 ph10 461 These override the default and the options given to \fBpcre_compile()\fP or
103 ph10 456 \fBpcre_compile2()\fP. For example, on a Unix system where LF is the default
104     newline sequence, the pattern
105 ph10 227 .sp
106     (*CR)a.b
107     .sp
108     changes the convention to CR. That pattern matches "a\enb" because LF is no
109     longer a newline. Note that these special settings, which are not
110     Perl-compatible, are recognized only at the very start of a pattern, and that
111 ph10 231 they must be in upper case. If more than one of them is present, the last one
112     is used.
113     .P
114 ph10 514 The newline convention affects the interpretation of the dot metacharacter when
115     PCRE_DOTALL is not set, and also the behaviour of \eN. However, it does not
116     affect what the \eR escape sequence matches. By default, this is any Unicode
117     newline sequence, for Perl compatibility. However, this can be changed; see the
118     description of \eR in the section entitled
119 ph10 231 .\" HTML <a href="#newlineseq">
120     .\" </a>
121     "Newline sequences"
122     .\"
123 ph10 247 below. A change of \eR setting can be combined with a change of newline
124 ph10 246 convention.
125 ph10 227 .
126     .
128     .rs
129     .sp
130 nigel 63 A regular expression is a pattern that is matched against a subject string from
131     left to right. Most characters stand for themselves in a pattern, and match the
132     corresponding characters in the subject. As a trivial example, the pattern
133 nigel 75 .sp
134 nigel 63 The quick brown fox
135 nigel 75 .sp
136 nigel 77 matches a portion of a subject string that is identical to itself. When
137     caseless matching is specified (the PCRE_CASELESS option), letters are matched
138     independently of case. In UTF-8 mode, PCRE always understands the concept of
139     case for characters whose values are less than 128, so caseless matching is
140     always possible. For characters with higher values, the concept of case is
141     supported if PCRE is compiled with Unicode property support, but not otherwise.
142     If you want to use caseless matching for characters 128 and above, you must
143     ensure that PCRE is compiled with Unicode property support as well as with
144     UTF-8 support.
145     .P
146     The power of regular expressions comes from the ability to include alternatives
147     and repetitions in the pattern. These are encoded in the pattern by the use of
148 nigel 75 \fImetacharacters\fP, which do not stand for themselves but instead are
149 nigel 63 interpreted in some special way.
150 nigel 75 .P
151     There are two different sets of metacharacters: those that are recognized
152 nigel 63 anywhere in the pattern except within square brackets, and those that are
153 nigel 93 recognized within square brackets. Outside square brackets, the metacharacters
154     are as follows:
155 nigel 75 .sp
156     \e general escape character with several uses
157 nigel 63 ^ assert start of string (or line, in multiline mode)
158     $ assert end of string (or line, in multiline mode)
159     . match any character except newline (by default)
160     [ start character class definition
161     | start of alternative branch
162     ( start subpattern
163     ) end subpattern
164     ? extends the meaning of (
165     also 0 or 1 quantifier
166     also quantifier minimizer
167     * 0 or more quantifier
168     + 1 or more quantifier
169     also "possessive quantifier"
170     { start min/max quantifier
171 nigel 75 .sp
172 nigel 63 Part of a pattern that is in square brackets is called a "character class". In
173 nigel 75 a character class the only metacharacters are:
174     .sp
175     \e general escape character
176 nigel 63 ^ negate the class, but only if the first character
177     - indicates character range
178 nigel 75 .\" JOIN
179 nigel 63 [ POSIX character class (only if followed by POSIX
180     syntax)
181     ] terminates the character class
182 nigel 75 .sp
183     The following sections describe the use of each of the metacharacters.
184     .
185 nigel 93 .
186 nigel 63 .SH BACKSLASH
187     .rs
188     .sp
189     The backslash character has several uses. Firstly, if it is followed by a
190 ph10 574 character that is not a number or a letter, it takes away any special meaning
191     that character may have. This use of backslash as an escape character applies
192 ph10 579 both inside and outside character classes.
193 nigel 75 .P
194     For example, if you want to match a * character, you write \e* in the pattern.
195 nigel 63 This escaping action applies whether or not the following character would
196 nigel 75 otherwise be interpreted as a metacharacter, so it is always safe to precede a
197     non-alphanumeric with backslash to specify that it stands for itself. In
198     particular, if you want to match a backslash, you write \e\e.
199     .P
200 ph10 574 In UTF-8 mode, only ASCII numbers and letters have any special meaning after a
201 ph10 579 backslash. All other characters (in particular, those whose codepoints are
202 ph10 574 greater than 127) are treated as literals.
203     .P
204 nigel 63 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
205     pattern (other than in a character class) and characters between a # outside
206 nigel 91 a character class and the next newline are ignored. An escaping backslash can
207     be used to include a whitespace or # character as part of the pattern.
208 nigel 75 .P
209 nigel 63 If you want to remove the special meaning from a sequence of characters, you
210 nigel 75 can do so by putting them between \eQ and \eE. This is different from Perl in
211     that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
212 nigel 63 Perl, $ and @ cause variable interpolation. Note the following examples:
213 nigel 75 .sp
214 nigel 63 Pattern PCRE matches Perl matches
215 nigel 75 .sp
216     .\" JOIN
217     \eQabc$xyz\eE abc$xyz abc followed by the
218 nigel 63 contents of $xyz
219 nigel 75 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
220     \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
221     .sp
222     The \eQ...\eE sequence is recognized both inside and outside character classes.
223 ph10 607 An isolated \eE that is not preceded by \eQ is ignored. If \eQ is not followed
224     by \eE later in the pattern, the literal interpretation continues to the end of
225     the pattern (that is, \eE is assumed at the end). If the isolated \eQ is inside
226     a character class, this causes an error, because the character class is not
227     terminated.
228 nigel 75 .
229     .
230     .\" HTML <a name="digitsafterbackslash"></a>
231     .SS "Non-printing characters"
232     .rs
233     .sp
234 nigel 63 A second use of backslash provides a way of encoding non-printing characters
235     in patterns in a visible manner. There is no restriction on the appearance of
236     non-printing characters, apart from the binary zero that terminates a pattern,
237 ph10 456 but when a pattern is being prepared by text editing, it is often easier to use
238     one of the following escape sequences than the binary character it represents:
239 nigel 75 .sp
240     \ea alarm, that is, the BEL character (hex 07)
241 ph10 574 \ecx "control-x", where x is any ASCII character
242 nigel 75 \ee escape (hex 1B)
243     \ef formfeed (hex 0C)
244 ph10 227 \en linefeed (hex 0A)
245 nigel 75 \er carriage return (hex 0D)
246     \et tab (hex 09)
247 ph10 488 \eddd character with octal code ddd, or back reference
248 nigel 75 \exhh character with hex code hh
249 nigel 87 \ex{hhh..} character with hex code hhh..
250 nigel 75 .sp
251     The precise effect of \ecx is as follows: if x is a lower case letter, it
252 nigel 63 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
253 ph10 574 Thus \ecz becomes hex 1A (z is 7A), but \ec{ becomes hex 3B ({ is 7B), while
254 ph10 579 \ec; becomes hex 7B (; is 3B). If the byte following \ec has a value greater
255     than 127, a compile-time error occurs. This locks out non-ASCII characters in
256     both byte mode and UTF-8 mode. (When PCRE is compiled in EBCDIC mode, all byte
257     values are valid. A lower case letter is converted to upper case, and then the
258 ph10 574 0xc0 bits are flipped.)
259 nigel 75 .P
260     After \ex, from zero to two hexadecimal digits are read (letters can be in
261 nigel 87 upper or lower case). Any number of hexadecimal digits may appear between \ex{
262     and }, but the value of the character code must be less than 256 in non-UTF-8
263 ph10 211 mode, and less than 2**31 in UTF-8 mode. That is, the maximum value in
264     hexadecimal is 7FFFFFFF. Note that this is bigger than the largest Unicode code
265     point, which is 10FFFF.
266 nigel 75 .P
267 ph10 211 If characters other than hexadecimal digits appear between \ex{ and }, or if
268     there is no terminating }, this form of escape is not recognized. Instead, the
269     initial \ex will be interpreted as a basic hexadecimal escape, with no
270     following digits, giving a character whose value is zero.
271     .P
272 nigel 63 Characters whose value is less than 256 can be defined by either of the two
273 nigel 87 syntaxes for \ex. There is no difference in the way they are handled. For
274     example, \exdc is exactly the same as \ex{dc}.
275 nigel 75 .P
276 nigel 91 After \e0 up to two further octal digits are read. If there are fewer than two
277     digits, just those that are present are used. Thus the sequence \e0\ex\e07
278     specifies two binary zeros followed by a BEL character (code value 7). Make
279     sure you supply two digits after the initial zero if the pattern character that
280     follows is itself an octal digit.
281 nigel 75 .P
282 nigel 63 The handling of a backslash followed by a digit other than 0 is complicated.
283     Outside a character class, PCRE reads it and any following digits as a decimal
284     number. If the number is less than 10, or if there have been at least that many
285     previous capturing left parentheses in the expression, the entire sequence is
286 nigel 75 taken as a \fIback reference\fP. A description of how this works is given
287     .\" HTML <a href="#backreferences">
288     .\" </a>
289     later,
290     .\"
291     following the discussion of
292     .\" HTML <a href="#subpattern">
293     .\" </a>
294     parenthesized subpatterns.
295     .\"
296     .P
297 nigel 63 Inside a character class, or if the decimal number is greater than 9 and there
298     have not been that many capturing subpatterns, PCRE re-reads up to three octal
299 nigel 93 digits following the backslash, and uses them to generate a data character. Any
300 nigel 91 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
301     character specified in octal must be less than \e400. In UTF-8 mode, values up
302     to \e777 are permitted. For example:
303 nigel 75 .sp
304     \e040 is another way of writing a space
305     .\" JOIN
306     \e40 is the same, provided there are fewer than 40
307 nigel 63 previous capturing subpatterns
308 nigel 75 \e7 is always a back reference
309     .\" JOIN
310     \e11 might be a back reference, or another way of
311 nigel 63 writing a tab
312 nigel 75 \e011 is always a tab
313     \e0113 is a tab followed by the character "3"
314     .\" JOIN
315     \e113 might be a back reference, otherwise the
316 nigel 63 character with octal code 113
317 nigel 75 .\" JOIN
318     \e377 might be a back reference, otherwise
319 nigel 63 the byte consisting entirely of 1 bits
320 nigel 75 .\" JOIN
321     \e81 is either a back reference, or a binary zero
322 nigel 63 followed by the two characters "8" and "1"
323 nigel 75 .sp
324 nigel 63 Note that octal values of 100 or greater must not be introduced by a leading
325     zero, because no more than three octal digits are ever read.
326 nigel 75 .P
327 nigel 91 All the sequences that define a single character value can be used both inside
328     and outside character classes. In addition, inside a character class, the
329 ph10 513 sequence \eb is interpreted as the backspace character (hex 08). The sequences
330 ph10 514 \eB, \eN, \eR, and \eX are not special inside a character class. Like any other
331 ph10 513 unrecognized escape sequences, they are treated as the literal characters "B",
332 ph10 514 "N", "R", and "X" by default, but cause an error if the PCRE_EXTRA option is
333     set. Outside a character class, these sequences have different meanings.
334 nigel 75 .
335     .
336 nigel 93 .SS "Absolute and relative back references"
337     .rs
338     .sp
339 ph10 208 The sequence \eg followed by an unsigned or a negative number, optionally
340     enclosed in braces, is an absolute or relative back reference. A named back
341     reference can be coded as \eg{name}. Back references are discussed
342 nigel 93 .\" HTML <a href="#backreferences">
343     .\" </a>
344     later,
345     .\"
346     following the discussion of
347     .\" HTML <a href="#subpattern">
348     .\" </a>
349     parenthesized subpatterns.
350     .\"
351     .
352     .
353 ph10 333 .SS "Absolute and relative subroutine calls"
354     .rs
355     .sp
356 ph10 345 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
357     a number enclosed either in angle brackets or single quotes, is an alternative
358     syntax for referencing a subpattern as a "subroutine". Details are discussed
359 ph10 333 .\" HTML <a href="#onigurumasubroutines">
360     .\" </a>
361     later.
362     .\"
363 ph10 345 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
364 ph10 461 synonymous. The former is a back reference; the latter is a
365 ph10 454 .\" HTML <a href="#subpatternsassubroutines">
366     .\" </a>
367     subroutine
368     .\"
369     call.
370 ph10 333 .
371     .
372 ph10 518 .\" HTML <a name="genericchartypes"></a>
373 nigel 75 .SS "Generic character types"
374     .rs
375     .sp
376 ph10 514 Another use of backslash is for specifying generic character types:
377 nigel 75 .sp
378 ph10 182 \ed any decimal digit
379 nigel 75 \eD any character that is not a decimal digit
380 ph10 178 \eh any horizontal whitespace character
381 ph10 182 \eH any character that is not a horizontal whitespace character
382 nigel 75 \es any whitespace character
383     \eS any character that is not a whitespace character
384 ph10 178 \ev any vertical whitespace character
385 ph10 182 \eV any character that is not a vertical whitespace character
386 nigel 75 \ew any "word" character
387     \eW any "non-word" character
388     .sp
389 ph10 535 There is also the single sequence \eN, which matches a non-newline character.
390     This is the same as
391 ph10 514 .\" HTML <a href="#fullstopdot">
392     .\" </a>
393 ph10 535 the "." metacharacter
394 ph10 514 .\"
395     when PCRE_DOTALL is not set.
396 nigel 75 .P
397 ph10 514 Each pair of lower and upper case escape sequences partitions the complete set
398     of characters into two disjoint sets. Any given character matches one, and only
399 ph10 518 one, of each pair. The sequences can appear both inside and outside character
400 nigel 75 classes. They each match one character of the appropriate type. If the current
401 ph10 518 matching point is at the end of the subject string, all of them fail, because
402 nigel 75 there is no character to match.
403     .P
404     For compatibility with Perl, \es does not match the VT character (code 11).
405     This makes it different from the the POSIX "space" class. The \es characters
406 ph10 178 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
407 nigel 91 included in a Perl script, \es may match the VT character. In PCRE, it never
408 ph10 178 does.
409 nigel 75 .P
410 ph10 518 A "word" character is an underscore or any character that is a letter or digit.
411     By default, the definition of letters and digits is controlled by PCRE's
412     low-valued character tables, and may vary if locale-specific matching is taking
413     place (see
414     .\" HTML <a href="pcreapi.html#localesupport">
415     .\" </a>
416     "Locale support"
417     .\"
418     in the
419     .\" HREF
420     \fBpcreapi\fP
421     .\"
422     page). For example, in a French locale such as "fr_FR" in Unix-like systems,
423     or "french" in Windows, some character codes greater than 128 are used for
424     accented letters, and these are then matched by \ew. The use of locales with
425     Unicode is discouraged.
426 ph10 178 .P
427 ph10 518 By default, in UTF-8 mode, characters with values greater than 128 never match
428     \ed, \es, or \ew, and always match \eD, \eS, and \eW. These sequences retain
429     their original meanings from before UTF-8 support was available, mainly for
430 ph10 535 efficiency reasons. However, if PCRE is compiled with Unicode property support,
431 ph10 518 and the PCRE_UCP option is set, the behaviour is changed so that Unicode
432     properties are used to determine character types, as follows:
433     .sp
434     \ed any character that \ep{Nd} matches (decimal digit)
435     \es any character that \ep{Z} matches, plus HT, LF, FF, CR
436     \ew any character that \ep{L} or \ep{N} matches, plus underscore
437     .sp
438     The upper case escapes match the inverse sets of characters. Note that \ed
439 ph10 535 matches only decimal digits, whereas \ew matches any Unicode digit, as well as
440 ph10 518 any Unicode letter, and underscore. Note also that PCRE_UCP affects \eb, and
441     \eB because they are defined in terms of \ew and \eW. Matching these sequences
442     is noticeably slower when PCRE_UCP is set.
443     .P
444 ph10 579 The sequences \eh, \eH, \ev, and \eV are features that were added to Perl at
445 ph10 572 release 5.10. In contrast to the other sequences, which match only ASCII
446     characters by default, these always match certain high-valued codepoints in
447     UTF-8 mode, whether or not PCRE_UCP is set. The horizontal space characters
448     are:
449 ph10 178 .sp
450     U+0009 Horizontal tab
451     U+0020 Space
452     U+00A0 Non-break space
453     U+1680 Ogham space mark
454     U+180E Mongolian vowel separator
455     U+2000 En quad
456     U+2001 Em quad
457     U+2002 En space
458     U+2003 Em space
459     U+2004 Three-per-em space
460     U+2005 Four-per-em space
461     U+2006 Six-per-em space
462     U+2007 Figure space
463     U+2008 Punctuation space
464     U+2009 Thin space
465     U+200A Hair space
466     U+202F Narrow no-break space
467     U+205F Medium mathematical space
468     U+3000 Ideographic space
469     .sp
470     The vertical space characters are:
471     .sp
472     U+000A Linefeed
473     U+000B Vertical tab
474     U+000C Formfeed
475     U+000D Carriage return
476     U+0085 Next line
477     U+2028 Line separator
478     U+2029 Paragraph separator
479 nigel 75 .
480     .
481 ph10 231 .\" HTML <a name="newlineseq"></a>
482 nigel 93 .SS "Newline sequences"
483     .rs
484     .sp
485 ph10 231 Outside a character class, by default, the escape sequence \eR matches any
486 ph10 572 Unicode newline sequence. In non-UTF-8 mode \eR is equivalent to the following:
487 nigel 93 .sp
488     (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
489     .sp
490     This is an example of an "atomic group", details of which are given
491     .\" HTML <a href="#atomicgroup">
492     .\" </a>
493     below.
494     .\"
495     This particular group matches either the two-character sequence CR followed by
496     LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
497     U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
498     line, U+0085). The two-character sequence is treated as a single unit that
499     cannot be split.
500     .P
501     In UTF-8 mode, two additional characters whose codepoints are greater than 255
502     are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
503     Unicode character property support is not needed for these characters to be
504     recognized.
505     .P
506 ph10 231 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
507     complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
508 ph10 247 either at compile time or when the pattern is matched. (BSR is an abbrevation
509 ph10 246 for "backslash R".) This can be made the default when PCRE is built; if this is
510     the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
511     It is also possible to specify these settings by starting a pattern string with
512     one of the following sequences:
513 ph10 231 .sp
514     (*BSR_ANYCRLF) CR, LF, or CRLF only
515     (*BSR_UNICODE) any Unicode newline sequence
516     .sp
517 ph10 461 These override the default and the options given to \fBpcre_compile()\fP or
518 ph10 456 \fBpcre_compile2()\fP, but they can be overridden by options given to
519     \fBpcre_exec()\fP or \fBpcre_dfa_exec()\fP. Note that these special settings,
520     which are not Perl-compatible, are recognized only at the very start of a
521     pattern, and that they must be in upper case. If more than one of them is
522     present, the last one is used. They can be combined with a change of newline
523 ph10 518 convention; for example, a pattern can start with:
524 ph10 246 .sp
525     (*ANY)(*BSR_ANYCRLF)
526     .sp
527 ph10 518 They can also be combined with the (*UTF8) or (*UCP) special sequences. Inside
528     a character class, \eR is treated as an unrecognized escape sequence, and so
529     matches the letter "R" by default, but causes an error if PCRE_EXTRA is set.
530 nigel 93 .
531     .
532 nigel 75 .\" HTML <a name="uniextseq"></a>
533     .SS Unicode character properties
534     .rs
535     .sp
536     When PCRE is built with Unicode character property support, three additional
537 ph10 184 escape sequences that match characters with specific properties are available.
538     When not in UTF-8 mode, these sequences are of course limited to testing
539     characters whose codepoints are less than 256, but they do work in this mode.
540     The extra escape sequences are:
541 nigel 75 .sp
542 nigel 87 \ep{\fIxx\fP} a character with the \fIxx\fP property
543     \eP{\fIxx\fP} a character without the \fIxx\fP property
544     \eX an extended Unicode sequence
545 nigel 75 .sp
546 nigel 87 The property names represented by \fIxx\fP above are limited to the Unicode
547 ph10 517 script names, the general category properties, "Any", which matches any
548     character (including newline), and some special PCRE properties (described
549 ph10 535 in the
550 ph10 517 .\" HTML <a href="#extraprops">
551     .\" </a>
552 ph10 535 next section).
553 ph10 517 .\"
554     Other Perl properties such as "InMusicalSymbols" are not currently supported by
555     PCRE. Note that \eP{Any} does not match any characters, so always causes a
556     match failure.
557 nigel 75 .P
558 nigel 87 Sets of Unicode characters are defined as belonging to certain scripts. A
559     character from one of these sets can be matched using a script name. For
560     example:
561 nigel 75 .sp
562 nigel 87 \ep{Greek}
563     \eP{Han}
564     .sp
565     Those that are not part of an identified script are lumped together as
566     "Common". The current list of scripts is:
567     .P
568     Arabic,
569     Armenian,
570 ph10 491 Avestan,
571 nigel 93 Balinese,
572 ph10 491 Bamum,
573 nigel 87 Bengali,
574     Bopomofo,
575     Braille,
576     Buginese,
577     Buhid,
578     Canadian_Aboriginal,
579 ph10 491 Carian,
580     Cham,
581 nigel 87 Cherokee,
582     Common,
583     Coptic,
584 nigel 93 Cuneiform,
585 nigel 87 Cypriot,
586     Cyrillic,
587     Deseret,
588     Devanagari,
589 ph10 491 Egyptian_Hieroglyphs,
590 nigel 87 Ethiopic,
591     Georgian,
592     Glagolitic,
593     Gothic,
594     Greek,
595     Gujarati,
596     Gurmukhi,
597     Han,
598     Hangul,
599     Hanunoo,
600     Hebrew,
601     Hiragana,
602 ph10 491 Imperial_Aramaic,
603 nigel 87 Inherited,
604 ph10 491 Inscriptional_Pahlavi,
605     Inscriptional_Parthian,
606     Javanese,
607     Kaithi,
608 nigel 87 Kannada,
609     Katakana,
610 ph10 491 Kayah_Li,
611 nigel 87 Kharoshthi,
612     Khmer,
613     Lao,
614     Latin,
615 ph10 491 Lepcha,
616 nigel 87 Limbu,
617     Linear_B,
618 ph10 491 Lisu,
619     Lycian,
620     Lydian,
621 nigel 87 Malayalam,
622 ph10 491 Meetei_Mayek,
623 nigel 87 Mongolian,
624     Myanmar,
625     New_Tai_Lue,
626 nigel 93 Nko,
627 nigel 87 Ogham,
628     Old_Italic,
629     Old_Persian,
630 ph10 491 Old_South_Arabian,
631     Old_Turkic,
632     Ol_Chiki,
633 nigel 87 Oriya,
634     Osmanya,
635 nigel 93 Phags_Pa,
636     Phoenician,
637 ph10 491 Rejang,
638 nigel 87 Runic,
639 ph10 491 Samaritan,
640     Saurashtra,
641 nigel 87 Shavian,
642     Sinhala,
643 ph10 491 Sundanese,
644 nigel 87 Syloti_Nagri,
645     Syriac,
646     Tagalog,
647     Tagbanwa,
648     Tai_Le,
649 ph10 491 Tai_Tham,
650     Tai_Viet,
651 nigel 87 Tamil,
652     Telugu,
653     Thaana,
654     Thai,
655     Tibetan,
656     Tifinagh,
657     Ugaritic,
658 ph10 491 Vai,
659 nigel 87 Yi.
660     .P
661 ph10 517 Each character has exactly one Unicode general category property, specified by
662     a two-letter abbreviation. For compatibility with Perl, negation can be
663     specified by including a circumflex between the opening brace and the property
664     name. For example, \ep{^Lu} is the same as \eP{Lu}.
665 nigel 87 .P
666     If only one letter is specified with \ep or \eP, it includes all the general
667     category properties that start with that letter. In this case, in the absence
668     of negation, the curly brackets in the escape sequence are optional; these two
669     examples have the same effect:
670     .sp
671 nigel 75 \ep{L}
672     \epL
673     .sp
674 nigel 87 The following general category property codes are supported:
675 nigel 75 .sp
676     C Other
677     Cc Control
678     Cf Format
679     Cn Unassigned
680     Co Private use
681     Cs Surrogate
682     .sp
683     L Letter
684     Ll Lower case letter
685     Lm Modifier letter
686     Lo Other letter
687     Lt Title case letter
688     Lu Upper case letter
689     .sp
690     M Mark
691     Mc Spacing mark
692     Me Enclosing mark
693     Mn Non-spacing mark
694     .sp
695     N Number
696     Nd Decimal number
697     Nl Letter number
698     No Other number
699     .sp
700     P Punctuation
701     Pc Connector punctuation
702     Pd Dash punctuation
703     Pe Close punctuation
704     Pf Final punctuation
705     Pi Initial punctuation
706     Po Other punctuation
707     Ps Open punctuation
708     .sp
709     S Symbol
710     Sc Currency symbol
711     Sk Modifier symbol
712     Sm Mathematical symbol
713     So Other symbol
714     .sp
715     Z Separator
716     Zl Line separator
717     Zp Paragraph separator
718     Zs Space separator
719     .sp
720 nigel 87 The special property L& is also supported: it matches a character that has
721     the Lu, Ll, or Lt property, in other words, a letter that is not classified as
722     a modifier or "other".
723 nigel 75 .P
724 ph10 211 The Cs (Surrogate) property applies only to characters in the range U+D800 to
725     U+DFFF. Such characters are not valid in UTF-8 strings (see RFC 3629) and so
726     cannot be tested by PCRE, unless UTF-8 validity checking has been turned off
727     (see the discussion of PCRE_NO_UTF8_CHECK in the
728     .\" HREF
729     \fBpcreapi\fP
730     .\"
731 ph10 451 page). Perl does not support the Cs property.
732 ph10 211 .P
733 ph10 451 The long synonyms for property names that Perl supports (such as \ep{Letter})
734 nigel 91 are not supported by PCRE, nor is it permitted to prefix any of these
735 nigel 87 properties with "Is".
736     .P
737     No character that is in the Unicode table has the Cn (unassigned) property.
738     Instead, this property is assumed for any code point that is not in the
739     Unicode table.
740     .P
741 nigel 75 Specifying caseless matching does not affect these escape sequences. For
742     example, \ep{Lu} always matches only upper case letters.
743     .P
744     The \eX escape matches any number of Unicode characters that form an extended
745     Unicode sequence. \eX is equivalent to
746     .sp
747     (?>\ePM\epM*)
748     .sp
749     That is, it matches a character without the "mark" property, followed by zero
750     or more characters with the "mark" property, and treats the sequence as an
751     atomic group
752     .\" HTML <a href="#atomicgroup">
753     .\" </a>
754     (see below).
755     .\"
756     Characters with the "mark" property are typically accents that affect the
757 ph10 185 preceding character. None of them have codepoints less than 256, so in
758 ph10 184 non-UTF-8 mode \eX matches any one character.
759 nigel 75 .P
760 ph10 628 Note that recent versions of Perl have changed \eX to match what Unicode calls
761     an "extended grapheme cluster", which has a more complicated definition.
762     .P
763 nigel 75 Matching characters by Unicode property is not fast, because PCRE has to search
764     a structure that contains data for over fifteen thousand characters. That is
765     why the traditional escape sequences such as \ed and \ew do not use Unicode
766 ph10 535 properties in PCRE by default, though you can make them do so by setting the
767 ph10 518 PCRE_UCP option for \fBpcre_compile()\fP or by starting the pattern with
768     (*UCP).
769 nigel 75 .
770     .
771 ph10 517 .\" HTML <a name="extraprops"></a>
772     .SS PCRE's additional properties
773     .rs
774     .sp
775 ph10 535 As well as the standard Unicode properties described in the previous
776     section, PCRE supports four more that make it possible to convert traditional
777 ph10 517 escape sequences such as \ew and \es and POSIX character classes to use Unicode
778 ph10 518 properties. PCRE uses these non-standard, non-Perl properties internally when
779     PCRE_UCP is set. They are:
780 ph10 517 .sp
781     Xan Any alphanumeric character
782     Xps Any POSIX space character
783     Xsp Any Perl space character
784     Xwd Any Perl "word" character
785     .sp
786 ph10 535 Xan matches characters that have either the L (letter) or the N (number)
787     property. Xps matches the characters tab, linefeed, vertical tab, formfeed, or
788 ph10 517 carriage return, and any other character that has the Z (separator) property.
789 ph10 535 Xsp is the same as Xps, except that vertical tab is excluded. Xwd matches the
790 ph10 517 same characters as Xan, plus underscore.
791     .
792     .
793 ph10 168 .\" HTML <a name="resetmatchstart"></a>
794     .SS "Resetting the match start"
795     .rs
796     .sp
797 ph10 572 The escape sequence \eK causes any previously matched characters not to be
798     included in the final matched sequence. For example, the pattern:
799 ph10 168 .sp
800     foo\eKbar
801     .sp
802 ph10 172 matches "foobar", but reports that it has matched "bar". This feature is
803 ph10 168 similar to a lookbehind assertion
804     .\" HTML <a href="#lookbehind">
805     .\" </a>
806     (described below).
807     .\"
808 ph10 172 However, in this case, the part of the subject before the real match does not
809     have to be of fixed length, as lookbehind assertions do. The use of \eK does
810 ph10 168 not interfere with the setting of
811     .\" HTML <a href="#subpattern">
812     .\" </a>
813     captured substrings.
814 ph10 172 .\"
815 ph10 168 For example, when the pattern
816     .sp
817     (foo)\eKbar
818     .sp
819 ph10 172 matches "foobar", the first substring is still set to "foo".
820 ph10 500 .P
821 ph10 507 Perl documents that the use of \eK within assertions is "not well defined". In
822     PCRE, \eK is acted upon when it occurs inside positive assertions, but is
823 ph10 500 ignored in negative assertions.
824 ph10 168 .
825     .
826 nigel 75 .\" HTML <a name="smallassertions"></a>
827     .SS "Simple assertions"
828     .rs
829     .sp
830 nigel 93 The final use of backslash is for certain simple assertions. An assertion
831 nigel 63 specifies a condition that has to be met at a particular point in a match,
832     without consuming any characters from the subject string. The use of
833 nigel 75 subpatterns for more complicated assertions is described
834     .\" HTML <a href="#bigassertions">
835     .\" </a>
836     below.
837     .\"
838 nigel 91 The backslashed assertions are:
839 nigel 75 .sp
840     \eb matches at a word boundary
841     \eB matches when not at a word boundary
842 nigel 93 \eA matches at the start of the subject
843     \eZ matches at the end of the subject
844     also matches before a newline at the end of the subject
845     \ez matches only at the end of the subject
846     \eG matches at the first matching position in the subject
847 nigel 75 .sp
848 ph10 513 Inside a character class, \eb has a different meaning; it matches the backspace
849 ph10 535 character. If any other of these assertions appears in a character class, by
850 ph10 513 default it matches the corresponding literal character (for example, \eB
851     matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
852     escape sequence" error is generated instead.
853 nigel 75 .P
854 nigel 63 A word boundary is a position in the subject string where the current character
855 nigel 75 and the previous character do not both match \ew or \eW (i.e. one matches
856     \ew and the other matches \eW), or the start or end of the string if the
857 ph10 518 first or last character matches \ew, respectively. In UTF-8 mode, the meanings
858     of \ew and \eW can be changed by setting the PCRE_UCP option. When this is
859     done, it also affects \eb and \eB. Neither PCRE nor Perl has a separate "start
860     of word" or "end of word" metasequence. However, whatever follows \eb normally
861     determines which it is. For example, the fragment \eba matches "a" at the start
862     of a word.
863 nigel 75 .P
864     The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
865     dollar (described in the next section) in that they only ever match at the very
866     start and end of the subject string, whatever options are set. Thus, they are
867     independent of multiline mode. These three assertions are not affected by the
868     PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
869     circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
870     argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
871     at a point other than the beginning of the subject, \eA can never match. The
872 nigel 91 difference between \eZ and \ez is that \eZ matches before a newline at the end
873     of the string as well as at the very end, whereas \ez matches only at the end.
874 nigel 75 .P
875     The \eG assertion is true only when the current matching position is at the
876     start point of the match, as specified by the \fIstartoffset\fP argument of
877     \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
878     non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
879 nigel 63 arguments, you can mimic Perl's /g option, and it is in this kind of
880 nigel 75 implementation where \eG can be useful.
881     .P
882     Note, however, that PCRE's interpretation of \eG, as the start of the current
883 nigel 63 match, is subtly different from Perl's, which defines it as the end of the
884     previous match. In Perl, these can be different when the previously matched
885     string was empty. Because PCRE does just one match at a time, it cannot
886     reproduce this behaviour.
887 nigel 75 .P
888     If all the alternatives of a pattern begin with \eG, the expression is anchored
889 nigel 63 to the starting match position, and the "anchored" flag is set in the compiled
890     regular expression.
891 nigel 75 .
892     .
894 nigel 63 .rs
895     .sp
896     Outside a character class, in the default matching mode, the circumflex
897 nigel 75 character is an assertion that is true only if the current matching point is
898     at the start of the subject string. If the \fIstartoffset\fP argument of
899     \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
900 nigel 63 option is unset. Inside a character class, circumflex has an entirely different
901 nigel 75 meaning
902     .\" HTML <a href="#characterclass">
903     .\" </a>
904     (see below).
905     .\"
906     .P
907 nigel 63 Circumflex need not be the first character of the pattern if a number of
908     alternatives are involved, but it should be the first thing in each alternative
909     in which it appears if the pattern is ever to match that branch. If all
910     possible alternatives start with a circumflex, that is, if the pattern is
911     constrained to match only at the start of the subject, it is said to be an
912     "anchored" pattern. (There are also other constructs that can cause a pattern
913     to be anchored.)
914 nigel 75 .P
915     A dollar character is an assertion that is true only if the current matching
916 nigel 63 point is at the end of the subject string, or immediately before a newline
917 nigel 91 at the end of the string (by default). Dollar need not be the last character of
918     the pattern if a number of alternatives are involved, but it should be the last
919     item in any branch in which it appears. Dollar has no special meaning in a
920     character class.
921 nigel 75 .P
922 nigel 63 The meaning of dollar can be changed so that it matches only at the very end of
923     the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
924 nigel 75 does not affect the \eZ assertion.
925     .P
926 nigel 63 The meanings of the circumflex and dollar characters are changed if the
927 nigel 91 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
928     immediately after internal newlines as well as at the start of the subject
929     string. It does not match after a newline that ends the string. A dollar
930     matches before any newlines in the string, as well as at the very end, when
931     PCRE_MULTILINE is set. When newline is specified as the two-character
932     sequence CRLF, isolated CR and LF characters do not indicate newlines.
933 nigel 75 .P
934 nigel 91 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
935     \en represents a newline) in multiline mode, but not otherwise. Consequently,
936     patterns that are anchored in single line mode because all branches start with
937     ^ are not anchored in multiline mode, and a match for circumflex is possible
938     when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
939     PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
940     .P
941 nigel 75 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
942 nigel 63 end of the subject in both modes, and if all branches of a pattern start with
943 nigel 91 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
944 nigel 75 .
945     .
946 ph10 514 .\" HTML <a name="fullstopdot"></a>
948 nigel 63 .rs
949     .sp
950     Outside a character class, a dot in the pattern matches any one character in
951 nigel 91 the subject string except (by default) a character that signifies the end of a
952 nigel 93 line. In UTF-8 mode, the matched character may be more than one byte long.
953 nigel 91 .P
954 nigel 93 When a line ending is defined as a single character, dot never matches that
955     character; when the two-character sequence CRLF is used, dot does not match CR
956     if it is immediately followed by LF, but otherwise it matches all characters
957     (including isolated CRs and LFs). When any Unicode line endings are being
958     recognized, dot does not match CR or LF or any of the other line ending
959     characters.
960     .P
961 nigel 91 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
962 nigel 93 option is set, a dot matches any one character, without exception. If the
963     two-character sequence CRLF is present in the subject string, it takes two dots
964     to match it.
965 nigel 91 .P
966     The handling of dot is entirely independent of the handling of circumflex and
967     dollar, the only relationship being that they both involve newlines. Dot has no
968     special meaning in a character class.
969 ph10 514 .P
970 ph10 579 The escape sequence \eN behaves like a dot, except that it is not affected by
971 ph10 572 the PCRE_DOTALL option. In other words, it matches any character except one
972     that signifies the end of a line.
973 nigel 75 .
974     .
976 nigel 63 .rs
977     .sp
978 nigel 75 Outside a character class, the escape sequence \eC matches any one byte, both
979 nigel 93 in and out of UTF-8 mode. Unlike a dot, it always matches any line-ending
980     characters. The feature is provided in Perl in order to match individual bytes
981 ph10 572 in UTF-8 mode. Because it breaks up UTF-8 characters into individual bytes, the
982     rest of the string may start with a malformed UTF-8 character. For this reason,
983 nigel 93 the \eC escape sequence is best avoided.
984 nigel 75 .P
985     PCRE does not allow \eC to appear in lookbehind assertions
986     .\" HTML <a href="#lookbehind">
987     .\" </a>
988     (described below),
989     .\"
990     because in UTF-8 mode this would make it impossible to calculate the length of
991     the lookbehind.
992     .
993     .
994     .\" HTML <a name="characterclass"></a>
996 nigel 63 .rs
997     .sp
998     An opening square bracket introduces a character class, terminated by a closing
999 ph10 461 square bracket. A closing square bracket on its own is not special by default.
1000     However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
1001 ph10 456 bracket causes a compile-time error. If a closing square bracket is required as
1002     a member of the class, it should be the first data character in the class
1003     (after an initial circumflex, if present) or escaped with a backslash.
1004 nigel 75 .P
1005 nigel 63 A character class matches a single character in the subject. In UTF-8 mode, the
1006 ph10 456 character may be more than one byte long. A matched character must be in the
1007     set of characters defined by the class, unless the first character in the class
1008 nigel 63 definition is a circumflex, in which case the subject character must not be in
1009     the set defined by the class. If a circumflex is actually required as a member
1010     of the class, ensure it is not the first character, or escape it with a
1011     backslash.
1012 nigel 75 .P
1013 nigel 63 For example, the character class [aeiou] matches any lower case vowel, while
1014     [^aeiou] matches any character that is not a lower case vowel. Note that a
1015 nigel 75 circumflex is just a convenient notation for specifying the characters that
1016     are in the class by enumerating those that are not. A class that starts with a
1017 ph10 456 circumflex is not an assertion; it still consumes a character from the subject
1018 nigel 75 string, and therefore it fails if the current pointer is at the end of the
1019     string.
1020     .P
1021 nigel 63 In UTF-8 mode, characters with values greater than 255 can be included in a
1022 nigel 75 class as a literal string of bytes, or by using the \ex{ escaping mechanism.
1023     .P
1024 nigel 63 When caseless matching is set, any letters in a class represent both their
1025     upper case and lower case versions, so for example, a caseless [aeiou] matches
1026     "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1027 nigel 77 caseful version would. In UTF-8 mode, PCRE always understands the concept of
1028     case for characters whose values are less than 128, so caseless matching is
1029     always possible. For characters with higher values, the concept of case is
1030     supported if PCRE is compiled with Unicode property support, but not otherwise.
1031 ph10 456 If you want to use caseless matching in UTF8-mode for characters 128 and above,
1032     you must ensure that PCRE is compiled with Unicode property support as well as
1033     with UTF-8 support.
1034 nigel 75 .P
1035 nigel 93 Characters that might indicate line breaks are never treated in any special way
1036     when matching character classes, whatever line-ending sequence is in use, and
1037     whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
1038     such as [^a] always matches one of these characters.
1039 nigel 75 .P
1040 nigel 63 The minus (hyphen) character can be used to specify a range of characters in a
1041     character class. For example, [d-m] matches any letter between d and m,
1042     inclusive. If a minus character is required in a class, it must be escaped with
1043     a backslash or appear in a position where it cannot be interpreted as
1044     indicating a range, typically as the first or last character in the class.
1045 nigel 75 .P
1046 nigel 63 It is not possible to have the literal character "]" as the end character of a
1047     range. A pattern such as [W-]46] is interpreted as a class of two characters
1048     ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1049     "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1050 nigel 75 the end of range, so [W-\e]46] is interpreted as a class containing a range
1051     followed by two other characters. The octal or hexadecimal representation of
1052     "]" can also be used to end a range.
1053     .P
1054 nigel 63 Ranges operate in the collating sequence of character values. They can also be
1055 nigel 75 used for characters specified numerically, for example [\e000-\e037]. In UTF-8
1056 nigel 63 mode, ranges can include characters whose values are greater than 255, for
1057 nigel 75 example [\ex{100}-\ex{2ff}].
1058     .P
1059 nigel 63 If a range that includes letters is used when caseless matching is set, it
1060     matches the letters in either case. For example, [W-c] is equivalent to
1061 nigel 75 [][\e\e^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
1062 ph10 139 tables for a French locale are in use, [\exc8-\excb] matches accented E
1063 nigel 75 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
1064     characters with values greater than 128 only when it is compiled with Unicode
1065     property support.
1066     .P
1067 ph10 575 The character escape sequences \ed, \eD, \eh, \eH, \ep, \eP, \es, \eS, \ev,
1068     \eV, \ew, and \eW may appear in a character class, and add the characters that
1069     they match to the class. For example, [\edABCDEF] matches any hexadecimal
1070     digit. In UTF-8 mode, the PCRE_UCP option affects the meanings of \ed, \es, \ew
1071     and their upper case partners, just as it does when they appear outside a
1072     character class, as described in the section entitled
1073     .\" HTML <a href="#genericchartypes">
1074     .\" </a>
1075     "Generic character types"
1076     .\"
1077     above. The escape sequence \eb has a different meaning inside a character
1078     class; it matches the backspace character. The sequences \eB, \eN, \eR, and \eX
1079     are not special inside a character class. Like any other unrecognized escape
1080     sequences, they are treated as the literal characters "B", "N", "R", and "X" by
1081     default, but cause an error if the PCRE_EXTRA option is set.
1082     .P
1083     A circumflex can conveniently be used with the upper case character types to
1084 ph10 518 specify a more restricted set of characters than the matching lower case type.
1085 ph10 575 For example, the class [^\eW_] matches any letter or digit, but not underscore,
1086 ph10 579 whereas [\ew] includes underscore. A positive character class should be read as
1087     "something OR something OR ..." and a negative class as "NOT something AND NOT
1088 ph10 575 something AND NOT ...".
1089 nigel 75 .P
1090     The only metacharacters that are recognized in character classes are backslash,
1091     hyphen (only where it can be interpreted as specifying a range), circumflex
1092     (only at the start), opening square bracket (only when it can be interpreted as
1093     introducing a POSIX class name - see the next section), and the terminating
1094     closing square bracket. However, escaping other non-alphanumeric characters
1095     does no harm.
1096     .
1097     .
1099 nigel 63 .rs
1100     .sp
1101 nigel 75 Perl supports the POSIX notation for character classes. This uses names
1102 nigel 63 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1103     this notation. For example,
1104 nigel 75 .sp
1105 nigel 63 [01[:alpha:]%]
1106 nigel 75 .sp
1107 nigel 63 matches "0", "1", any alphabetic character, or "%". The supported class names
1108 ph10 518 are:
1109 nigel 75 .sp
1110 nigel 63 alnum letters and digits
1111     alpha letters
1112     ascii character codes 0 - 127
1113     blank space or tab only
1114     cntrl control characters
1115 nigel 75 digit decimal digits (same as \ed)
1116 nigel 63 graph printing characters, excluding space
1117     lower lower case letters
1118     print printing characters, including space
1119 ph10 518 punct printing characters, excluding letters and digits and space
1120 nigel 75 space white space (not quite the same as \es)
1121 nigel 63 upper upper case letters
1122 nigel 75 word "word" characters (same as \ew)
1123 nigel 63 xdigit hexadecimal digits
1124 nigel 75 .sp
1125 nigel 63 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1126     space (32). Notice that this list includes the VT character (code 11). This
1127 nigel 75 makes "space" different to \es, which does not include VT (for Perl
1128 nigel 63 compatibility).
1129 nigel 75 .P
1130 nigel 63 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1131     5.8. Another Perl extension is negation, which is indicated by a ^ character
1132     after the colon. For example,
1133 nigel 75 .sp
1134 nigel 63 [12[:^digit:]]
1135 nigel 75 .sp
1136 nigel 63 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1137     syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1138     supported, and an error is given if they are encountered.
1139 nigel 75 .P
1140 ph10 518 By default, in UTF-8 mode, characters with values greater than 128 do not match
1141     any of the POSIX character classes. However, if the PCRE_UCP option is passed
1142 ph10 535 to \fBpcre_compile()\fP, some of the classes are changed so that Unicode
1143     character properties are used. This is achieved by replacing the POSIX classes
1144 ph10 518 by other sequences, as follows:
1145     .sp
1146     [:alnum:] becomes \ep{Xan}
1147     [:alpha:] becomes \ep{L}
1148 ph10 535 [:blank:] becomes \eh
1149 ph10 518 [:digit:] becomes \ep{Nd}
1150     [:lower:] becomes \ep{Ll}
1151 ph10 535 [:space:] becomes \ep{Xps}
1152 ph10 518 [:upper:] becomes \ep{Lu}
1153     [:word:] becomes \ep{Xwd}
1154     .sp
1155     Negated versions, such as [:^alpha:] use \eP instead of \ep. The other POSIX
1156     classes are unchanged, and match only characters with code points less than
1157     128.
1158 nigel 75 .
1159     .
1160     .SH "VERTICAL BAR"
1161 nigel 63 .rs
1162     .sp
1163     Vertical bar characters are used to separate alternative patterns. For example,
1164     the pattern
1165 nigel 75 .sp
1166 nigel 63 gilbert|sullivan
1167 nigel 75 .sp
1168 nigel 63 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1169 nigel 91 and an empty alternative is permitted (matching the empty string). The matching
1170     process tries each alternative in turn, from left to right, and the first one
1171     that succeeds is used. If the alternatives are within a subpattern
1172 nigel 75 .\" HTML <a href="#subpattern">
1173     .\" </a>
1174     (defined below),
1175     .\"
1176     "succeeds" means matching the rest of the main pattern as well as the
1177     alternative in the subpattern.
1178     .
1179     .
1181 nigel 63 .rs
1182     .sp
1183     The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1184 ph10 231 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1185     the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1186     The option letters are
1187 nigel 75 .sp
1188 nigel 63 i for PCRE_CASELESS
1189     m for PCRE_MULTILINE
1190     s for PCRE_DOTALL
1191     x for PCRE_EXTENDED
1192 nigel 75 .sp
1193 nigel 63 For example, (?im) sets caseless, multiline matching. It is also possible to
1194     unset these options by preceding the letter with a hyphen, and a combined
1195     setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1196     PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1197     permitted. If a letter appears both before and after the hyphen, the option is
1198     unset.
1199 nigel 75 .P
1200 ph10 231 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1201     changed in the same way as the Perl-compatible options by using the characters
1202     J, U and X respectively.
1203     .P
1204 ph10 412 When one of these option changes occurs at top level (that is, not inside
1205     subpattern parentheses), the change applies to the remainder of the pattern
1206     that follows. If the change is placed right at the start of a pattern, PCRE
1207     extracts it into the global options (and it will therefore show up in data
1208     extracted by the \fBpcre_fullinfo()\fP function).
1209 nigel 75 .P
1210 nigel 93 An option change within a subpattern (see below for a description of
1211 ph10 572 subpatterns) affects only that part of the subpattern that follows it, so
1212 nigel 75 .sp
1213 nigel 63 (a(?i)b)c
1214 nigel 75 .sp
1215 nigel 63 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1216     By this means, options can be made to have different settings in different
1217     parts of the pattern. Any changes made in one alternative do carry on
1218     into subsequent branches within the same subpattern. For example,
1219 nigel 75 .sp
1220 nigel 63 (a(?i)b|c)
1221 nigel 75 .sp
1222 nigel 63 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1223     branch is abandoned before the option setting. This is because the effects of
1224     option settings happen at compile time. There would be some very weird
1225     behaviour otherwise.
1226 ph10 251 .P
1227     \fBNote:\fP There are other PCRE-specific options that can be set by the
1228     application when the compile or match functions are called. In some cases the
1229 ph10 412 pattern can contain special leading sequences such as (*CRLF) to override what
1230     the application has set or what has been defaulted. Details are given in the
1231     section entitled
1232 ph10 251 .\" HTML <a href="#newlineseq">
1233     .\" </a>
1234     "Newline sequences"
1235     .\"
1236 ph10 518 above. There are also the (*UTF8) and (*UCP) leading sequences that can be used
1237     to set UTF-8 and Unicode property modes; they are equivalent to setting the
1238     PCRE_UTF8 and the PCRE_UCP options, respectively.
1239 nigel 75 .
1240     .
1241     .\" HTML <a name="subpattern"></a>
1242 nigel 63 .SH SUBPATTERNS
1243     .rs
1244     .sp
1245     Subpatterns are delimited by parentheses (round brackets), which can be nested.
1246 nigel 75 Turning part of a pattern into a subpattern does two things:
1247     .sp
1248 nigel 63 1. It localizes a set of alternatives. For example, the pattern
1249 nigel 75 .sp
1250 nigel 63 cat(aract|erpillar|)
1251 nigel 75 .sp
1252 ph10 572 matches "cataract", "caterpillar", or "cat". Without the parentheses, it would
1253     match "cataract", "erpillar" or an empty string.
1254 nigel 75 .sp
1255     2. It sets up the subpattern as a capturing subpattern. This means that, when
1256     the whole pattern matches, that portion of the subject string that matched the
1257     subpattern is passed back to the caller via the \fIovector\fP argument of
1258     \fBpcre_exec()\fP. Opening parentheses are counted from left to right (starting
1259 ph10 572 from 1) to obtain numbers for the capturing subpatterns. For example, if the
1260     string "the red king" is matched against the pattern
1261 nigel 75 .sp
1262 nigel 63 the ((red|white) (king|queen))
1263 nigel 75 .sp
1264 nigel 63 the captured substrings are "red king", "red", and "king", and are numbered 1,
1265     2, and 3, respectively.
1266 nigel 75 .P
1267 nigel 63 The fact that plain parentheses fulfil two functions is not always helpful.
1268     There are often times when a grouping subpattern is required without a
1269     capturing requirement. If an opening parenthesis is followed by a question mark
1270     and a colon, the subpattern does not do any capturing, and is not counted when
1271     computing the number of any subsequent capturing subpatterns. For example, if
1272     the string "the white queen" is matched against the pattern
1273 nigel 75 .sp
1274 nigel 63 the ((?:red|white) (king|queen))
1275 nigel 75 .sp
1276 nigel 63 the captured substrings are "white queen" and "queen", and are numbered 1 and
1277 nigel 93 2. The maximum number of capturing subpatterns is 65535.
1278 nigel 75 .P
1279 nigel 63 As a convenient shorthand, if any option settings are required at the start of
1280     a non-capturing subpattern, the option letters may appear between the "?" and
1281     the ":". Thus the two patterns
1282 nigel 75 .sp
1283 nigel 63 (?i:saturday|sunday)
1284     (?:(?i)saturday|sunday)
1285 nigel 75 .sp
1286 nigel 63 match exactly the same set of strings. Because alternative branches are tried
1287     from left to right, and options are not reset until the end of the subpattern
1288     is reached, an option setting in one branch does affect subsequent branches, so
1289     the above patterns match "SUNDAY" as well as "Saturday".
1290 nigel 75 .
1291     .
1292 ph10 456 .\" HTML <a name="dupsubpatternnumber"></a>
1294     .rs
1295     .sp
1296 ph10 182 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1297     the same numbers for its capturing parentheses. Such a subpattern starts with
1298     (?| and is itself a non-capturing subpattern. For example, consider this
1299 ph10 175 pattern:
1300     .sp
1301     (?|(Sat)ur|(Sun))day
1302 ph10 182 .sp
1303     Because the two alternatives are inside a (?| group, both sets of capturing
1304     parentheses are numbered one. Thus, when the pattern matches, you can look
1305     at captured substring number one, whichever alternative matched. This construct
1306     is useful when you want to capture part, but not all, of one of a number of
1307     alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1308 ph10 175 number is reset at the start of each branch. The numbers of any capturing
1309 ph10 572 parentheses that follow the subpattern start after the highest number used in
1310     any branch. The following example is taken from the Perl documentation. The
1311     numbers underneath show in which buffer the captured content will be stored.
1312 ph10 175 .sp
1313     # before ---------------branch-reset----------- after
1314     / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1315     # 1 2 2 3 2 3 4
1316 ph10 182 .sp
1317 ph10 488 A back reference to a numbered subpattern uses the most recent value that is
1318     set for that number by any subpattern. The following pattern matches "abcabc"
1319     or "defdef":
1320 ph10 456 .sp
1321 ph10 461 /(?|(abc)|(def))\e1/
1322 ph10 456 .sp
1323     In contrast, a recursive or "subroutine" call to a numbered subpattern always
1324 ph10 461 refers to the first one in the pattern with the given number. The following
1325 ph10 456 pattern matches "abcabc" or "defabc":
1326     .sp
1327     /(?|(abc)|(def))(?1)/
1328     .sp
1329 ph10 459 If a
1330     .\" HTML <a href="#conditions">
1331     .\" </a>
1332     condition test
1333     .\"
1334     for a subpattern's having matched refers to a non-unique number, the test is
1335     true if any of the subpatterns of that number have matched.
1336     .P
1337     An alternative approach to using this "branch reset" feature is to use
1338 ph10 175 duplicate named subpatterns, as described in the next section.
1339     .
1340     .
1341 nigel 75 .SH "NAMED SUBPATTERNS"
1342 nigel 63 .rs
1343     .sp
1344     Identifying capturing parentheses by number is simple, but it can be very hard
1345     to keep track of the numbers in complicated regular expressions. Furthermore,
1346 nigel 75 if an expression is modified, the numbers may change. To help with this
1347 nigel 93 difficulty, PCRE supports the naming of subpatterns. This feature was not
1348     added to Perl until release 5.10. Python had the feature earlier, and PCRE
1349     introduced it at release 4.0, using the Python syntax. PCRE now supports both
1350 ph10 459 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1351     have different names, but PCRE does not.
1352 nigel 93 .P
1353     In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1354     (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1355 nigel 91 parentheses from other parts of the pattern, such as
1356     .\" HTML <a href="#backreferences">
1357     .\" </a>
1358 ph10 488 back references,
1359 nigel 91 .\"
1360     .\" HTML <a href="#recursion">
1361     .\" </a>
1362     recursion,
1363     .\"
1364     and
1365     .\" HTML <a href="#conditions">
1366     .\" </a>
1367     conditions,
1368     .\"
1369     can be made by name as well as by number.
1370 nigel 75 .P
1371 nigel 91 Names consist of up to 32 alphanumeric characters and underscores. Named
1372 nigel 93 capturing parentheses are still allocated numbers as well as names, exactly as
1373     if the names were not present. The PCRE API provides function calls for
1374     extracting the name-to-number translation table from a compiled pattern. There
1375     is also a convenience function for extracting a captured substring by name.
1376 nigel 91 .P
1377     By default, a name must be unique within a pattern, but it is possible to relax
1378 ph10 457 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1379 ph10 461 names are also always permitted for subpatterns with the same number, set up as
1380 ph10 457 described in the previous section.) Duplicate names can be useful for patterns
1381     where only one instance of the named parentheses can match. Suppose you want to
1382     match the name of a weekday, either as a 3-letter abbreviation or as the full
1383     name, and in both cases you want to extract the abbreviation. This pattern
1384     (ignoring the line breaks) does the job:
1385 nigel 91 .sp
1386 nigel 93 (?<DN>Mon|Fri|Sun)(?:day)?|
1387     (?<DN>Tue)(?:sday)?|
1388     (?<DN>Wed)(?:nesday)?|
1389     (?<DN>Thu)(?:rsday)?|
1390     (?<DN>Sat)(?:urday)?
1391 nigel 91 .sp
1392     There are five capturing substrings, but only one is ever set after a match.
1393 ph10 182 (An alternative way of solving this problem is to use a "branch reset"
1394 ph10 175 subpattern, as described in the previous section.)
1395     .P
1396 nigel 91 The convenience function for extracting the data by name returns the substring
1397 nigel 93 for the first (and in this example, the only) subpattern of that name that
1398 ph10 461 matched. This saves searching to find which numbered subpattern it was.
1399 ph10 459 .P
1400 ph10 488 If you make a back reference to a non-unique named subpattern from elsewhere in
1401 ph10 459 the pattern, the one that corresponds to the first occurrence of the name is
1402     used. In the absence of duplicate numbers (see the previous section) this is
1403     the one with the lowest number. If you use a named reference in a condition
1404     test (see the
1405     .\"
1406     .\" HTML <a href="#conditions">
1407     .\" </a>
1408     section about conditions
1409     .\"
1410 ph10 461 below), either to check whether a subpattern has matched, or to check for
1411 ph10 459 recursion, all subpatterns with the same name are tested. If the condition is
1412     true for any one of them, the overall condition is true. This is the same
1413     behaviour as testing by number. For further details of the interfaces for
1414     handling named subpatterns, see the
1415 nigel 63 .\" HREF
1416 nigel 75 \fBpcreapi\fP
1417 nigel 63 .\"
1418     documentation.
1419 ph10 385 .P
1420     \fBWarning:\fP You cannot use different names to distinguish between two
1421 ph10 457 subpatterns with the same number because PCRE uses only the numbers when
1422     matching. For this reason, an error is given at compile time if different names
1423     are given to subpatterns with the same number. However, you can give the same
1424     name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
1425 nigel 75 .
1426     .
1427 nigel 63 .SH REPETITION
1428     .rs
1429     .sp
1430     Repetition is specified by quantifiers, which can follow any of the following
1431     items:
1432 nigel 75 .sp
1433 nigel 63 a literal data character
1434 nigel 93 the dot metacharacter
1435 nigel 75 the \eC escape sequence
1436     the \eX escape sequence (in UTF-8 mode with Unicode properties)
1437 nigel 93 the \eR escape sequence
1438 ph10 572 an escape such as \ed or \epL that matches a single character
1439 nigel 63 a character class
1440     a back reference (see next section)
1441     a parenthesized subpattern (unless it is an assertion)
1442 ph10 461 a recursive or "subroutine" call to a subpattern
1443 nigel 75 .sp
1444 nigel 63 The general repetition quantifier specifies a minimum and maximum number of
1445     permitted matches, by giving the two numbers in curly brackets (braces),
1446     separated by a comma. The numbers must be less than 65536, and the first must
1447     be less than or equal to the second. For example:
1448 nigel 75 .sp
1449 nigel 63 z{2,4}
1450 nigel 75 .sp
1451 nigel 63 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1452     character. If the second number is omitted, but the comma is present, there is
1453     no upper limit; if the second number and the comma are both omitted, the
1454     quantifier specifies an exact number of required matches. Thus
1455 nigel 75 .sp
1456 nigel 63 [aeiou]{3,}
1457 nigel 75 .sp
1458 nigel 63 matches at least 3 successive vowels, but may match many more, while
1459 nigel 75 .sp
1460     \ed{8}
1461     .sp
1462 nigel 63 matches exactly 8 digits. An opening curly bracket that appears in a position
1463     where a quantifier is not allowed, or one that does not match the syntax of a
1464     quantifier, is taken as a literal character. For example, {,6} is not a
1465     quantifier, but a literal string of four characters.
1466 nigel 75 .P
1467 nigel 63 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1468 nigel 75 bytes. Thus, for example, \ex{100}{2} matches two UTF-8 characters, each of
1469     which is represented by a two-byte sequence. Similarly, when Unicode property
1470     support is available, \eX{3} matches three Unicode extended sequences, each of
1471     which may be several bytes long (and they may be of different lengths).
1472     .P
1473 nigel 63 The quantifier {0} is permitted, causing the expression to behave as if the
1474 ph10 345 previous item and the quantifier were not present. This may be useful for
1475     subpatterns that are referenced as
1476 ph10 335 .\" HTML <a href="#subpatternsassubroutines">
1477     .\" </a>
1478     subroutines
1479     .\"
1480 ph10 572 from elsewhere in the pattern (but see also the section entitled
1481     .\" HTML <a href="#subdefine">
1482     .\" </a>
1483     "Defining subpatterns for use by reference only"
1484     .\"
1485     below). Items other than subpatterns that have a {0} quantifier are omitted
1486     from the compiled pattern.
1487 nigel 75 .P
1488 nigel 93 For convenience, the three most common quantifiers have single-character
1489     abbreviations:
1490 nigel 75 .sp
1491 nigel 63 * is equivalent to {0,}
1492     + is equivalent to {1,}
1493     ? is equivalent to {0,1}
1494 nigel 75 .sp
1495 nigel 63 It is possible to construct infinite loops by following a subpattern that can
1496     match no characters with a quantifier that has no upper limit, for example:
1497 nigel 75 .sp
1498 nigel 63 (a?)*
1499 nigel 75 .sp
1500 nigel 63 Earlier versions of Perl and PCRE used to give an error at compile time for
1501     such patterns. However, because there are cases where this can be useful, such
1502     patterns are now accepted, but if any repetition of the subpattern does in fact
1503     match no characters, the loop is forcibly broken.
1504 nigel 75 .P
1505 nigel 63 By default, the quantifiers are "greedy", that is, they match as much as
1506     possible (up to the maximum number of permitted times), without causing the
1507     rest of the pattern to fail. The classic example of where this gives problems
1508 nigel 75 is in trying to match comments in C programs. These appear between /* and */
1509     and within the comment, individual * and / characters may appear. An attempt to
1510     match C comments by applying the pattern
1511     .sp
1512     /\e*.*\e*/
1513     .sp
1514 nigel 63 to the string
1515 nigel 75 .sp
1516     /* first comment */ not comment /* second comment */
1517     .sp
1518 nigel 63 fails, because it matches the entire string owing to the greediness of the .*
1519     item.
1520 nigel 75 .P
1521 nigel 63 However, if a quantifier is followed by a question mark, it ceases to be
1522     greedy, and instead matches the minimum number of times possible, so the
1523     pattern
1524 nigel 75 .sp
1525     /\e*.*?\e*/
1526     .sp
1527 nigel 63 does the right thing with the C comments. The meaning of the various
1528     quantifiers is not otherwise changed, just the preferred number of matches.
1529     Do not confuse this use of question mark with its use as a quantifier in its
1530     own right. Because it has two uses, it can sometimes appear doubled, as in
1531 nigel 75 .sp
1532     \ed??\ed
1533     .sp
1534 nigel 63 which matches one digit by preference, but can match two if that is the only
1535     way the rest of the pattern matches.
1536 nigel 75 .P
1537 nigel 93 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1538 nigel 63 the quantifiers are not greedy by default, but individual ones can be made
1539     greedy by following them with a question mark. In other words, it inverts the
1540     default behaviour.
1541 nigel 75 .P
1542 nigel 63 When a parenthesized subpattern is quantified with a minimum repeat count that
1543 nigel 75 is greater than 1 or with a limited maximum, more memory is required for the
1544 nigel 63 compiled pattern, in proportion to the size of the minimum or maximum.
1545 nigel 75 .P
1546 nigel 63 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1547 nigel 93 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1548 nigel 63 implicitly anchored, because whatever follows will be tried against every
1549     character position in the subject string, so there is no point in retrying the
1550     overall match at any position after the first. PCRE normally treats such a
1551 nigel 75 pattern as though it were preceded by \eA.
1552     .P
1553 nigel 63 In cases where it is known that the subject string contains no newlines, it is
1554     worth setting PCRE_DOTALL in order to obtain this optimization, or
1555     alternatively using ^ to indicate anchoring explicitly.
1556 nigel 75 .P
1557 nigel 63 However, there is one situation where the optimization cannot be used. When .*
1558 ph10 488 is inside capturing parentheses that are the subject of a back reference
1559 nigel 93 elsewhere in the pattern, a match at the start may fail where a later one
1560     succeeds. Consider, for example:
1561 nigel 75 .sp
1562     (.*)abc\e1
1563     .sp
1564 nigel 63 If the subject is "xyz123abc123" the match point is the fourth character. For
1565     this reason, such a pattern is not implicitly anchored.
1566 nigel 75 .P
1567 nigel 63 When a capturing subpattern is repeated, the value captured is the substring
1568     that matched the final iteration. For example, after
1569 nigel 75 .sp
1570     (tweedle[dume]{3}\es*)+
1571     .sp
1572 nigel 63 has matched "tweedledum tweedledee" the value of the captured substring is
1573     "tweedledee". However, if there are nested capturing subpatterns, the
1574     corresponding captured values may have been set in previous iterations. For
1575     example, after
1576 nigel 75 .sp
1577 nigel 63 /(a|(b))+/
1578 nigel 75 .sp
1579 nigel 63 matches "aba" the value of the second captured substring is "b".
1580 nigel 75 .
1581     .
1582     .\" HTML <a name="atomicgroup"></a>
1584 nigel 63 .rs
1585     .sp
1586 nigel 93 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1587     repetition, failure of what follows normally causes the repeated item to be
1588     re-evaluated to see if a different number of repeats allows the rest of the
1589     pattern to match. Sometimes it is useful to prevent this, either to change the
1590     nature of the match, or to cause it fail earlier than it otherwise might, when
1591     the author of the pattern knows there is no point in carrying on.
1592 nigel 75 .P
1593     Consider, for example, the pattern \ed+foo when applied to the subject line
1594     .sp
1595 nigel 63 123456bar
1596 nigel 75 .sp
1597 nigel 63 After matching all 6 digits and then failing to match "foo", the normal
1598 nigel 75 action of the matcher is to try again with only 5 digits matching the \ed+
1599 nigel 63 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1600     (a term taken from Jeffrey Friedl's book) provides the means for specifying
1601     that once a subpattern has matched, it is not to be re-evaluated in this way.
1602 nigel 75 .P
1603 nigel 93 If we use atomic grouping for the previous example, the matcher gives up
1604 nigel 63 immediately on failing to match "foo" the first time. The notation is a kind of
1605     special parenthesis, starting with (?> as in this example:
1606 nigel 75 .sp
1607     (?>\ed+)foo
1608     .sp
1609 nigel 63 This kind of parenthesis "locks up" the part of the pattern it contains once
1610     it has matched, and a failure further into the pattern is prevented from
1611     backtracking into it. Backtracking past it to previous items, however, works as
1612     normal.
1613 nigel 75 .P
1614 nigel 63 An alternative description is that a subpattern of this type matches the string
1615     of characters that an identical standalone pattern would match, if anchored at
1616     the current point in the subject string.
1617 nigel 75 .P
1618 nigel 63 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1619     the above example can be thought of as a maximizing repeat that must swallow
1620 nigel 75 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1621 nigel 63 number of digits they match in order to make the rest of the pattern match,
1622 nigel 75 (?>\ed+) can only match an entire sequence of digits.
1623     .P
1624 nigel 63 Atomic groups in general can of course contain arbitrarily complicated
1625     subpatterns, and can be nested. However, when the subpattern for an atomic
1626     group is just a single repeated item, as in the example above, a simpler
1627     notation, called a "possessive quantifier" can be used. This consists of an
1628     additional + character following a quantifier. Using this notation, the
1629     previous example can be rewritten as
1630 nigel 75 .sp
1631     \ed++foo
1632     .sp
1633 ph10 208 Note that a possessive quantifier can be used with an entire group, for
1634     example:
1635     .sp
1636     (abc|xyz){2,3}+
1637     .sp
1638 nigel 63 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1639     option is ignored. They are a convenient notation for the simpler forms of
1640 nigel 93 atomic group. However, there is no difference in the meaning of a possessive
1641     quantifier and the equivalent atomic group, though there may be a performance
1642     difference; possessive quantifiers should be slightly faster.
1643 nigel 75 .P
1644 nigel 93 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1645     Jeffrey Friedl originated the idea (and the name) in the first edition of his
1646     book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1647     package, and PCRE copied it from there. It ultimately found its way into Perl
1648     at release 5.10.
1649 nigel 75 .P
1650 nigel 93 PCRE has an optimization that automatically "possessifies" certain simple
1651     pattern constructs. For example, the sequence A+B is treated as A++B because
1652     there is no point in backtracking into a sequence of A's when B must follow.
1653     .P
1654 nigel 63 When a pattern contains an unlimited repeat inside a subpattern that can itself
1655     be repeated an unlimited number of times, the use of an atomic group is the
1656     only way to avoid some failing matches taking a very long time indeed. The
1657     pattern
1658 nigel 75 .sp
1659     (\eD+|<\ed+>)*[!?]
1660     .sp
1661 nigel 63 matches an unlimited number of substrings that either consist of non-digits, or
1662     digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1663     quickly. However, if it is applied to
1664 nigel 75 .sp
1665 nigel 63 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1666 nigel 75 .sp
1667 nigel 63 it takes a long time before reporting failure. This is because the string can
1668 nigel 75 be divided between the internal \eD+ repeat and the external * repeat in a
1669     large number of ways, and all have to be tried. (The example uses [!?] rather
1670     than a single character at the end, because both PCRE and Perl have an
1671     optimization that allows for fast failure when a single character is used. They
1672     remember the last single character that is required for a match, and fail early
1673     if it is not present in the string.) If the pattern is changed so that it uses
1674     an atomic group, like this:
1675     .sp
1676     ((?>\eD+)|<\ed+>)*[!?]
1677     .sp
1678 nigel 63 sequences of non-digits cannot be broken, and failure happens quickly.
1679 nigel 75 .
1680     .
1681     .\" HTML <a name="backreferences"></a>
1683 nigel 63 .rs
1684     .sp
1685     Outside a character class, a backslash followed by a digit greater than 0 (and
1686     possibly further digits) is a back reference to a capturing subpattern earlier
1687     (that is, to its left) in the pattern, provided there have been that many
1688     previous capturing left parentheses.
1689 nigel 75 .P
1690 nigel 63 However, if the decimal number following the backslash is less than 10, it is
1691     always taken as a back reference, and causes an error only if there are not
1692     that many capturing left parentheses in the entire pattern. In other words, the
1693     parentheses that are referenced need not be to the left of the reference for
1694 nigel 91 numbers less than 10. A "forward back reference" of this type can make sense
1695     when a repetition is involved and the subpattern to the right has participated
1696     in an earlier iteration.
1697     .P
1698 nigel 93 It is not possible to have a numerical "forward back reference" to a subpattern
1699     whose number is 10 or more using this syntax because a sequence such as \e50 is
1700     interpreted as a character defined in octal. See the subsection entitled
1701 nigel 91 "Non-printing characters"
1702 nigel 75 .\" HTML <a href="#digitsafterbackslash">
1703     .\" </a>
1704     above
1705     .\"
1706 nigel 93 for further details of the handling of digits following a backslash. There is
1707     no such problem when named parentheses are used. A back reference to any
1708     subpattern is possible using named parentheses (see below).
1709 nigel 75 .P
1710 nigel 93 Another way of avoiding the ambiguity inherent in the use of digits following a
1711 ph10 572 backslash is to use the \eg escape sequence. This escape must be followed by an
1712     unsigned number or a negative number, optionally enclosed in braces. These
1713     examples are all identical:
1714 nigel 93 .sp
1715     (ring), \e1
1716     (ring), \eg1
1717     (ring), \eg{1}
1718     .sp
1719 ph10 208 An unsigned number specifies an absolute reference without the ambiguity that
1720     is present in the older syntax. It is also useful when literal digits follow
1721     the reference. A negative number is a relative reference. Consider this
1722     example:
1723 nigel 93 .sp
1724     (abc(def)ghi)\eg{-1}
1725     .sp
1726     The sequence \eg{-1} is a reference to the most recently started capturing
1727 ph10 572 subpattern before \eg, that is, is it equivalent to \e2 in this example.
1728     Similarly, \eg{-2} would be equivalent to \e1. The use of relative references
1729     can be helpful in long patterns, and also in patterns that are created by
1730     joining together fragments that contain references within themselves.
1731 nigel 93 .P
1732 nigel 63 A back reference matches whatever actually matched the capturing subpattern in
1733     the current subject string, rather than anything matching the subpattern
1734     itself (see
1735     .\" HTML <a href="#subpatternsassubroutines">
1736     .\" </a>
1737     "Subpatterns as subroutines"
1738     .\"
1739     below for a way of doing that). So the pattern
1740 nigel 75 .sp
1741     (sens|respons)e and \e1ibility
1742     .sp
1743 nigel 63 matches "sense and sensibility" and "response and responsibility", but not
1744     "sense and responsibility". If caseful matching is in force at the time of the
1745     back reference, the case of letters is relevant. For example,
1746 nigel 75 .sp
1747     ((?i)rah)\es+\e1
1748     .sp
1749 nigel 63 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1750     capturing subpattern is matched caselessly.
1751 nigel 75 .P
1752 ph10 171 There are several different ways of writing back references to named
1753     subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1754     \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1755     back reference syntax, in which \eg can be used for both numeric and named
1756     references, is also supported. We could rewrite the above example in any of
1757 nigel 93 the following ways:
1758 nigel 75 .sp
1759 nigel 93 (?<p1>(?i)rah)\es+\ek<p1>
1760 ph10 171 (?'p1'(?i)rah)\es+\ek{p1}
1761 nigel 91 (?P<p1>(?i)rah)\es+(?P=p1)
1762 ph10 171 (?<p1>(?i)rah)\es+\eg{p1}
1763 nigel 75 .sp
1764 nigel 91 A subpattern that is referenced by name may appear in the pattern before or
1765     after the reference.
1766     .P
1767 nigel 63 There may be more than one back reference to the same subpattern. If a
1768     subpattern has not actually been used in a particular match, any back
1769 ph10 456 references to it always fail by default. For example, the pattern
1770 nigel 75 .sp
1771     (a|(bc))\e2
1772     .sp
1773 ph10 461 always fails if it starts to match "a" rather than "bc". However, if the
1774     PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1775 ph10 456 unset value matches an empty string.
1776     .P
1777     Because there may be many capturing parentheses in a pattern, all digits
1778     following a backslash are taken as part of a potential back reference number.
1779     If the pattern continues with a digit character, some delimiter must be used to
1780     terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1781     whitespace. Otherwise, the \eg{ syntax or an empty comment (see
1782 nigel 75 .\" HTML <a href="#comments">
1783     .\" </a>
1784     "Comments"
1785     .\"
1786     below) can be used.
1787 ph10 488 .
1788     .SS "Recursive back references"
1789     .rs
1790     .sp
1791 nigel 63 A back reference that occurs inside the parentheses to which it refers fails
1792 nigel 75 when the subpattern is first used, so, for example, (a\e1) never matches.
1793 nigel 63 However, such references can be useful inside repeated subpatterns. For
1794     example, the pattern
1795 nigel 75 .sp
1796     (a|b\e1)+
1797     .sp
1798 nigel 63 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1799     the subpattern, the back reference matches the character string corresponding
1800     to the previous iteration. In order for this to work, the pattern must be such
1801     that the first iteration does not need to match the back reference. This can be
1802     done using alternation, as in the example above, or by a quantifier with a
1803     minimum of zero.
1804 ph10 488 .P
1805     Back references of this type cause the group that they reference to be treated
1806     as an
1807     .\" HTML <a href="#atomicgroup">
1808     .\" </a>
1809     atomic group.
1810     .\"
1811     Once the whole group has been matched, a subsequent matching failure cannot
1812     cause backtracking into the middle of the group.
1813 nigel 75 .
1814     .
1815     .\" HTML <a name="bigassertions"></a>
1816 nigel 63 .SH ASSERTIONS
1817     .rs
1818     .sp
1819     An assertion is a test on the characters following or preceding the current
1820     matching point that does not actually consume any characters. The simple
1821 nigel 75 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1822     .\" HTML <a href="#smallassertions">
1823     .\" </a>
1824     above.
1825     .\"
1826     .P
1827 nigel 63 More complicated assertions are coded as subpatterns. There are two kinds:
1828     those that look ahead of the current position in the subject string, and those
1829 nigel 75 that look behind it. An assertion subpattern is matched in the normal way,
1830     except that it does not cause the current matching position to be changed.
1831     .P
1832     Assertion subpatterns are not capturing subpatterns, and may not be repeated,
1833     because it makes no sense to assert the same thing several times. If any kind
1834     of assertion contains capturing subpatterns within it, these are counted for
1835     the purposes of numbering the capturing subpatterns in the whole pattern.
1836     However, substring capturing is carried out only for positive assertions,
1837     because it does not make sense for negative assertions.
1838     .
1839     .
1840     .SS "Lookahead assertions"
1841     .rs
1842     .sp
1843 nigel 91 Lookahead assertions start with (?= for positive assertions and (?! for
1844     negative assertions. For example,
1845 nigel 75 .sp
1846     \ew+(?=;)
1847     .sp
1848 nigel 63 matches a word followed by a semicolon, but does not include the semicolon in
1849     the match, and
1850 nigel 75 .sp
1851 nigel 63 foo(?!bar)
1852 nigel 75 .sp
1853 nigel 63 matches any occurrence of "foo" that is not followed by "bar". Note that the
1854     apparently similar pattern
1855 nigel 75 .sp
1856 nigel 63 (?!foo)bar
1857 nigel 75 .sp
1858 nigel 63 does not find an occurrence of "bar" that is preceded by something other than
1859     "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1860     (?!foo) is always true when the next three characters are "bar". A
1861 nigel 75 lookbehind assertion is needed to achieve the other effect.
1862     .P
1863 nigel 63 If you want to force a matching failure at some point in a pattern, the most
1864     convenient way to do it is with (?!) because an empty string always matches, so
1865     an assertion that requires there not to be an empty string must always fail.
1866 ph10 572 The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).
1867 nigel 75 .
1868     .
1869     .\" HTML <a name="lookbehind"></a>
1870     .SS "Lookbehind assertions"
1871     .rs
1872     .sp
1873 nigel 63 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1874     negative assertions. For example,
1875 nigel 75 .sp
1876 nigel 63 (?<!foo)bar
1877 nigel 75 .sp
1878 nigel 63 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1879     a lookbehind assertion are restricted such that all the strings it matches must
1880 nigel 91 have a fixed length. However, if there are several top-level alternatives, they
1881     do not all have to have the same fixed length. Thus
1882 nigel 75 .sp
1883 nigel 63 (?<=bullock|donkey)
1884 nigel 75 .sp
1885 nigel 63 is permitted, but
1886 nigel 75 .sp
1887 nigel 63 (?<!dogs?|cats?)
1888 nigel 75 .sp
1889 nigel 63 causes an error at compile time. Branches that match different length strings
1890     are permitted only at the top level of a lookbehind assertion. This is an
1891 ph10 572 extension compared with Perl, which requires all branches to match the same
1892     length of string. An assertion such as
1893 nigel 75 .sp
1894 nigel 63 (?<=ab(c|de))
1895 nigel 75 .sp
1896 nigel 63 is not permitted, because its single top-level branch can match two different
1897 ph10 454 lengths, but it is acceptable to PCRE if rewritten to use two top-level
1898     branches:
1899 nigel 75 .sp
1900 nigel 63 (?<=abc|abde)
1901 nigel 75 .sp
1902 ph10 572 In some cases, the escape sequence \eK
1903 ph10 168 .\" HTML <a href="#resetmatchstart">
1904     .\" </a>
1905     (see above)
1906     .\"
1907 ph10 461 can be used instead of a lookbehind assertion to get round the fixed-length
1908 ph10 454 restriction.
1909 ph10 168 .P
1910 nigel 63 The implementation of lookbehind assertions is, for each alternative, to
1911 nigel 93 temporarily move the current position back by the fixed length and then try to
1912 nigel 63 match. If there are insufficient characters before the current position, the
1913 nigel 93 assertion fails.
1914 nigel 75 .P
1915     PCRE does not allow the \eC escape (which matches a single byte in UTF-8 mode)
1916 nigel 63 to appear in lookbehind assertions, because it makes it impossible to calculate
1917 nigel 93 the length of the lookbehind. The \eX and \eR escapes, which can match
1918     different numbers of bytes, are also not permitted.
1919 nigel 75 .P
1920 ph10 454 .\" HTML <a href="#subpatternsassubroutines">
1921     .\" </a>
1922     "Subroutine"
1923     .\"
1924     calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
1925 ph10 461 as the subpattern matches a fixed-length string.
1926 ph10 454 .\" HTML <a href="#recursion">
1927     .\" </a>
1928     Recursion,
1929     .\"
1930     however, is not supported.
1931     .P
1932 nigel 93 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1933 ph10 456 specify efficient matching of fixed-length strings at the end of subject
1934     strings. Consider a simple pattern such as
1935 nigel 75 .sp
1936 nigel 63 abcd$
1937 nigel 75 .sp
1938 nigel 63 when applied to a long string that does not match. Because matching proceeds
1939     from left to right, PCRE will look for each "a" in the subject and then see if
1940     what follows matches the rest of the pattern. If the pattern is specified as
1941 nigel 75 .sp
1942 nigel 63 ^.*abcd$
1943 nigel 75 .sp
1944 nigel 63 the initial .* matches the entire string at first, but when this fails (because
1945     there is no following "a"), it backtracks to match all but the last character,
1946     then all but the last two characters, and so on. Once again the search for "a"
1947     covers the entire string, from right to left, so we are no better off. However,
1948     if the pattern is written as
1949 nigel 75 .sp
1950 nigel 63 ^.*+(?<=abcd)
1951 nigel 75 .sp
1952 nigel 93 there can be no backtracking for the .*+ item; it can match only the entire
1953 nigel 63 string. The subsequent lookbehind assertion does a single test on the last four
1954     characters. If it fails, the match fails immediately. For long strings, this
1955     approach makes a significant difference to the processing time.
1956 nigel 75 .
1957     .
1958     .SS "Using multiple assertions"
1959     .rs
1960     .sp
1961 nigel 63 Several assertions (of any sort) may occur in succession. For example,
1962 nigel 75 .sp
1963     (?<=\ed{3})(?<!999)foo
1964     .sp
1965 nigel 63 matches "foo" preceded by three digits that are not "999". Notice that each of
1966     the assertions is applied independently at the same point in the subject
1967     string. First there is a check that the previous three characters are all
1968     digits, and then there is a check that the same three characters are not "999".
1969 nigel 75 This pattern does \fInot\fP match "foo" preceded by six characters, the first
1970 nigel 63 of which are digits and the last three of which are not "999". For example, it
1971     doesn't match "123abcfoo". A pattern to do that is
1972 nigel 75 .sp
1973     (?<=\ed{3}...)(?<!999)foo
1974     .sp
1975 nigel 63 This time the first assertion looks at the preceding six characters, checking
1976     that the first three are digits, and then the second assertion checks that the
1977     preceding three characters are not "999".
1978 nigel 75 .P
1979 nigel 63 Assertions can be nested in any combination. For example,
1980 nigel 75 .sp
1981 nigel 63 (?<=(?<!foo)bar)baz
1982 nigel 75 .sp
1983 nigel 63 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
1984     preceded by "foo", while
1985 nigel 75 .sp
1986     (?<=\ed{3}(?!999)...)foo
1987     .sp
1988     is another pattern that matches "foo" preceded by three digits and any three
1989 nigel 63 characters that are not "999".
1990 nigel 75 .
1991     .
1992 nigel 91 .\" HTML <a name="conditions"></a>
1994 nigel 63 .rs
1995     .sp
1996     It is possible to cause the matching process to obey a subpattern
1997     conditionally or to choose between two alternative subpatterns, depending on
1998 ph10 461 the result of an assertion, or whether a specific capturing subpattern has
1999 ph10 456 already been matched. The two possible forms of conditional subpattern are:
2000 nigel 75 .sp
2001 nigel 63 (?(condition)yes-pattern)
2002     (?(condition)yes-pattern|no-pattern)
2003 nigel 75 .sp
2004 nigel 63 If the condition is satisfied, the yes-pattern is used; otherwise the
2005     no-pattern (if present) is used. If there are more than two alternatives in the
2006 ph10 557 subpattern, a compile-time error occurs. Each of the two alternatives may
2007 ph10 579 itself contain nested subpatterns of any form, including conditional
2008 ph10 557 subpatterns; the restriction to two alternatives applies only at the level of
2009 ph10 579 the condition. This pattern fragment is an example where the alternatives are
2010 ph10 557 complex:
2011     .sp
2012     (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
2013     .sp
2014 nigel 75 .P
2015 nigel 93 There are four kinds of condition: references to subpatterns, references to
2016     recursion, a pseudo-condition called DEFINE, and assertions.
2017     .
2018     .SS "Checking for a used subpattern by number"
2019     .rs
2020     .sp
2021     If the text between the parentheses consists of a sequence of digits, the
2022 ph10 456 condition is true if a capturing subpattern of that number has previously
2023 ph10 461 matched. If there is more than one capturing subpattern with the same number
2024     (see the earlier
2025 ph10 456 .\"
2026     .\" HTML <a href="#recursion">
2027     .\" </a>
2028     section about duplicate subpattern numbers),
2029     .\"
2030 ph10 572 the condition is true if any of them have matched. An alternative notation is
2031 ph10 456 to precede the digits with a plus or minus sign. In this case, the subpattern
2032     number is relative rather than absolute. The most recently opened parentheses
2033 ph10 572 can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside
2034 ph10 579 loops it can also make sense to refer to subsequent groups. The next
2035 ph10 572 parentheses to be opened can be referenced as (?(+1), and so on. (The value
2036     zero in any of these forms is not used; it provokes a compile-time error.)
2037 nigel 91 .P
2038     Consider the following pattern, which contains non-significant white space to
2039     make it more readable (assume the PCRE_EXTENDED option) and to divide it into
2040     three parts for ease of discussion:
2041 nigel 75 .sp
2042     ( \e( )? [^()]+ (?(1) \e) )
2043     .sp
2044 nigel 63 The first part matches an optional opening parenthesis, and if that
2045     character is present, sets it as the first captured substring. The second part
2046     matches one or more characters that are not parentheses. The third part is a
2047 ph10 572 conditional subpattern that tests whether or not the first set of parentheses
2048     matched. If they did, that is, if subject started with an opening parenthesis,
2049 nigel 63 the condition is true, and so the yes-pattern is executed and a closing
2050     parenthesis is required. Otherwise, since no-pattern is not present, the
2051     subpattern matches nothing. In other words, this pattern matches a sequence of
2052 nigel 93 non-parentheses, optionally enclosed in parentheses.
2053 ph10 167 .P
2054 ph10 172 If you were embedding this pattern in a larger one, you could use a relative
2055 ph10 167 reference:
2056     .sp
2057     ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
2058     .sp
2059     This makes the fragment independent of the parentheses in the larger pattern.
2060 nigel 93 .
2061     .SS "Checking for a used subpattern by name"
2062     .rs
2063 nigel 91 .sp
2064 nigel 93 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
2065     subpattern by name. For compatibility with earlier versions of PCRE, which had
2066     this facility before Perl, the syntax (?(name)...) is also recognized. However,
2067     there is a possible ambiguity with this syntax, because subpattern names may
2068     consist entirely of digits. PCRE looks first for a named subpattern; if it
2069     cannot find one and the name consists entirely of digits, PCRE looks for a
2070     subpattern of that number, which must be greater than zero. Using subpattern
2071     names that consist entirely of digits is not recommended.
2072     .P
2073     Rewriting the above example to use a named subpattern gives this:
2074 nigel 91 .sp
2075 nigel 93 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
2076     .sp
2077 ph10 461 If the name used in a condition of this kind is a duplicate, the test is
2078     applied to all subpatterns of the same name, and is true if any one of them has
2079 ph10 459 matched.
2080 nigel 93 .
2081     .SS "Checking for pattern recursion"
2082     .rs
2083     .sp
2084 nigel 91 If the condition is the string (R), and there is no subpattern with the name R,
2085 nigel 93 the condition is true if a recursive call to the whole pattern or any
2086     subpattern has been made. If digits or a name preceded by ampersand follow the
2087     letter R, for example:
2088     .sp
2089     (?(R3)...) or (?(R&name)...)
2090     .sp
2091 ph10 456 the condition is true if the most recent recursion is into a subpattern whose
2092 nigel 93 number or name is given. This condition does not check the entire recursion
2093 ph10 461 stack. If the name used in a condition of this kind is a duplicate, the test is
2094     applied to all subpatterns of the same name, and is true if any one of them is
2095     the most recent recursion.
2096 nigel 75 .P
2097 ph10 461 At "top level", all these recursion test conditions are false.
2098 ph10 454 .\" HTML <a href="#recursion">
2099     .\" </a>
2100 ph10 459 The syntax for recursive patterns
2101 ph10 454 .\"
2102 ph10 459 is described below.
2103 nigel 93 .
2104 ph10 572 .\" HTML <a name="subdefine"></a>
2105 nigel 93 .SS "Defining subpatterns for use by reference only"
2106     .rs
2107     .sp
2108     If the condition is the string (DEFINE), and there is no subpattern with the
2109     name DEFINE, the condition is always false. In this case, there may be only one
2110     alternative in the subpattern. It is always skipped if control reaches this
2111     point in the pattern; the idea of DEFINE is that it can be used to define
2112 ph10 461 "subroutines" that can be referenced from elsewhere. (The use of
2113 ph10 454 .\" HTML <a href="#subpatternsassubroutines">
2114     .\" </a>
2115     "subroutines"
2116     .\"
2117 ph10 572 is described below.) For example, a pattern to match an IPv4 address such as
2118     "" could be written like this (ignore whitespace and line
2119     breaks):
2120 nigel 93 .sp
2121     (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
2122     \eb (?&byte) (\e.(?&byte)){3} \eb
2123     .sp
2124     The first part of the pattern is a DEFINE group inside which a another group
2125     named "byte" is defined. This matches an individual component of an IPv4
2126     address (a number less than 256). When matching takes place, this part of the
2127 ph10 456 pattern is skipped because DEFINE acts like a false condition. The rest of the
2128     pattern uses references to the named group to match the four dot-separated
2129     components of an IPv4 address, insisting on a word boundary at each end.
2130 nigel 93 .
2131     .SS "Assertion conditions"
2132     .rs
2133     .sp
2134     If the condition is not in any of the above formats, it must be an assertion.
2135 nigel 63 This may be a positive or negative lookahead or lookbehind assertion. Consider
2136     this pattern, again containing non-significant white space, and with the two
2137     alternatives on the second line:
2138 nigel 75 .sp
2139 nigel 63 (?(?=[^a-z]*[a-z])
2140 nigel 75 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
2141     .sp
2142 nigel 63 The condition is a positive lookahead assertion that matches an optional
2143     sequence of non-letters followed by a letter. In other words, it tests for the
2144     presence of at least one letter in the subject. If a letter is found, the
2145     subject is matched against the first alternative; otherwise it is matched
2146     against the second. This pattern matches strings in one of the two forms
2147     dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2148 nigel 75 .
2149     .
2150     .\" HTML <a name="comments"></a>
2151 nigel 63 .SH COMMENTS
2152     .rs
2153     .sp
2154 ph10 579 There are two ways of including comments in patterns that are processed by
2155 ph10 562 PCRE. In both cases, the start of the comment must not be in a character class,
2156     nor in the middle of any other sequence of related characters such as (?: or a
2157     subpattern name or number. The characters that make up a comment play no part
2158     in the pattern matching.
2159     .P
2160 nigel 75 The sequence (?# marks the start of a comment that continues up to the next
2161 ph10 562 closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED
2162     option is set, an unescaped # character also introduces a comment, which in
2163     this case continues to immediately after the next newline character or
2164     character sequence in the pattern. Which characters are interpreted as newlines
2165     is controlled by the options passed to \fBpcre_compile()\fP or by a special
2166     sequence at the start of the pattern, as described in the section entitled
2167 ph10 572 .\" HTML <a href="#newlines">
2168 ph10 556 .\" </a>
2169     "Newline conventions"
2170     .\"
2171 ph10 572 above. Note that the end of this type of comment is a literal newline sequence
2172     in the pattern; escape sequences that happen to represent a newline do not
2173     count. For example, consider this pattern when PCRE_EXTENDED is set, and the
2174     default newline convention is in force:
2175 ph10 556 .sp
2176     abc #comment \en still comment
2177     .sp
2178 ph10 579 On encountering the # character, \fBpcre_compile()\fP skips along, looking for
2179 ph10 556 a newline in the pattern. The sequence \en is still literal at this stage, so
2180     it does not terminate the comment. Only an actual character with the code value
2181 ph10 562 0x0a (the default newline) does so.
2182 nigel 75 .
2183     .
2184 nigel 91 .\" HTML <a name="recursion"></a>
2185 nigel 75 .SH "RECURSIVE PATTERNS"
2186 nigel 63 .rs
2187     .sp
2188     Consider the problem of matching a string in parentheses, allowing for
2189     unlimited nested parentheses. Without the use of recursion, the best that can
2190     be done is to use a pattern that matches up to some fixed depth of nesting. It
2191 nigel 93 is not possible to handle an arbitrary nesting depth.
2192     .P
2193     For some time, Perl has provided a facility that allows regular expressions to
2194     recurse (amongst other things). It does this by interpolating Perl code in the
2195     expression at run time, and the code can refer to the expression itself. A Perl
2196     pattern using code interpolation to solve the parentheses problem can be
2197     created like this:
2198 nigel 75 .sp
2199     $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
2200     .sp
2201 nigel 63 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2202 nigel 93 recursively to the pattern in which it appears.
2203 nigel 75 .P
2204 nigel 93 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2205     supports special syntax for recursion of the entire pattern, and also for
2206     individual subpattern recursion. After its introduction in PCRE and Python,
2207 ph10 453 this kind of recursion was subsequently introduced into Perl at release 5.10.
2208 nigel 75 .P
2209 nigel 93 A special item that consists of (? followed by a number greater than zero and a
2210     closing parenthesis is a recursive call of the subpattern of the given number,
2211 ph10 461 provided that it occurs inside that subpattern. (If not, it is a
2212 ph10 454 .\" HTML <a href="#subpatternsassubroutines">
2213     .\" </a>
2214     "subroutine"
2215     .\"
2216 nigel 93 call, which is described in the next section.) The special item (?R) or (?0) is
2217     a recursive call of the entire regular expression.
2218 nigel 87 .P
2219     This PCRE pattern solves the nested parentheses problem (assume the
2220     PCRE_EXTENDED option is set so that white space is ignored):
2221 nigel 75 .sp
2222 ph10 456 \e( ( [^()]++ | (?R) )* \e)
2223 nigel 75 .sp
2224 nigel 63 First it matches an opening parenthesis. Then it matches any number of
2225     substrings which can either be a sequence of non-parentheses, or a recursive
2226 nigel 87 match of the pattern itself (that is, a correctly parenthesized substring).
2227 ph10 461 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2228 ph10 456 to avoid backtracking into sequences of non-parentheses.
2229 nigel 75 .P
2230 nigel 63 If this were part of a larger pattern, you would not want to recurse the entire
2231     pattern, so instead you could use this:
2232 nigel 75 .sp
2233 ph10 456 ( \e( ( [^()]++ | (?1) )* \e) )
2234 nigel 75 .sp
2235 nigel 63 We have put the pattern into parentheses, and caused the recursion to refer to
2236 ph10 172 them instead of the whole pattern.
2237 ph10 166 .P
2238     In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2239 ph10 572 is made easier by the use of relative references. Instead of (?1) in the
2240     pattern above you can write (?-2) to refer to the second most recently opened
2241     parentheses preceding the recursion. In other words, a negative number counts
2242     capturing parentheses leftwards from the point at which it is encountered.
2243 ph10 166 .P
2244     It is also possible to refer to subsequently opened parentheses, by writing
2245     references such as (?+2). However, these cannot be recursive because the
2246     reference is not inside the parentheses that are referenced. They are always
2247 ph10 454 .\" HTML <a href="#subpatternsassubroutines">
2248     .\" </a>
2249     "subroutine"
2250     .\"
2251     calls, as described in the next section.
2252 ph10 166 .P
2253     An alternative approach is to use named parentheses instead. The Perl syntax
2254     for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
2255     could rewrite the above example as follows:
2256 nigel 75 .sp
2257 ph10 456 (?<pn> \e( ( [^()]++ | (?&pn) )* \e) )
2258 nigel 75 .sp
2259 nigel 93 If there is more than one subpattern with the same name, the earliest one is
2260 ph10 172 used.
2261 ph10 166 .P
2262     This particular example pattern that we have been looking at contains nested
2263 ph10 456 unlimited repeats, and so the use of a possessive quantifier for matching
2264     strings of non-parentheses is important when applying the pattern to strings
2265     that do not match. For example, when this pattern is applied to
2266 nigel 75 .sp
2267 nigel 63 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2268 nigel 75 .sp
2269 ph10 456 it yields "no match" quickly. However, if a possessive quantifier is not used,
2270 nigel 63 the match runs for a very long time indeed because there are so many different
2271     ways the + and * repeats can carve up the subject, and all have to be tested
2272     before failure can be reported.
2273 nigel 75 .P
2274 ph10 464 At the end of a match, the values of capturing parentheses are those from
2275     the outermost level. If you want to obtain intermediate values, a callout
2276     function can be used (see below and the
2277 nigel 63 .\" HREF
2278 nigel 75 \fBpcrecallout\fP
2279 nigel 63 .\"
2280     documentation). If the pattern above is matched against
2281 nigel 75 .sp
2282 nigel 63 (ab(cd)ef)
2283 nigel 75 .sp
2284 ph10 464 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2285     the last value taken on at the top level. If a capturing subpattern is not
2286     matched at the top level, its final value is unset, even if it is (temporarily)
2287     set at a deeper level.
2288 nigel 75 .P
2289 ph10 464 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2290     obtain extra memory to store data during a recursion, which it does by using
2291     \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no memory can
2292     be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2293     .P
2294 nigel 63 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2295     Consider this pattern, which matches text in angle brackets, allowing for
2296     arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2297     recursing), whereas any characters are permitted at the outer level.
2298 nigel 75 .sp
2299     < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
2300     .sp
2301 nigel 63 In this pattern, (?(R) is the start of a conditional subpattern, with two
2302     different alternatives for the recursive and non-recursive cases. The (?R) item
2303     is the actual recursive call.
2304 nigel 75 .
2305     .
2306 ph10 453 .\" HTML <a name="recursiondifference"></a>
2307     .SS "Recursion difference from Perl"
2308     .rs
2309     .sp
2310     In PCRE (like Python, but unlike Perl), a recursive subpattern call is always
2311     treated as an atomic group. That is, once it has matched some of the subject
2312     string, it is never re-entered, even if it contains untried alternatives and
2313 ph10 461 there is a subsequent matching failure. This can be illustrated by the
2314     following pattern, which purports to match a palindromic string that contains
2315 ph10 453 an odd number of characters (for example, "a", "aba", "abcba", "abcdcba"):
2316     .sp
2317     ^(.|(.)(?1)\e2)$
2318     .sp
2319 ph10 461 The idea is that it either matches a single character, or two identical
2320     characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2321 ph10 453 it does not if the pattern is longer than three characters. Consider the
2322     subject string "abcba":
2323     .P
2324 ph10 461 At the top level, the first character is matched, but as it is not at the end
2325 ph10 453 of the string, the first alternative fails; the second alternative is taken
2326     and the recursion kicks in. The recursive call to subpattern 1 successfully
2327     matches the next character ("b"). (Note that the beginning and end of line
2328     tests are not part of the recursion).
2329     .P
2330     Back at the top level, the next character ("c") is compared with what
2331 ph10 461 subpattern 2 matched, which was "a". This fails. Because the recursion is
2332 ph10 453 treated as an atomic group, there are now no backtracking points, and so the
2333     entire match fails. (Perl is able, at this point, to re-enter the recursion and
2334     try the second alternative.) However, if the pattern is written with the
2335     alternatives in the other order, things are different:
2336     .sp
2337     ^((.)(?1)\e2|.)$
2338     .sp
2339 ph10 461 This time, the recursing alternative is tried first, and continues to recurse
2340     until it runs out of characters, at which point the recursion fails. But this
2341     time we do have another alternative to try at the higher level. That is the big
2342 ph10 453 difference: in the previous case the remaining alternative is at a deeper
2343     recursion level, which PCRE cannot use.
2344     .P
2345 ph10 572 To change the pattern so that it matches all palindromic strings, not just
2346     those with an odd number of characters, it is tempting to change the pattern to
2347     this:
2348 ph10 453 .sp
2349     ^((.)(?1)\e2|.?)$
2350     .sp
2351 ph10 461 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2352     deeper recursion has matched a single character, it cannot be entered again in
2353     order to match an empty string. The solution is to separate the two cases, and
2354 ph10 453 write out the odd and even cases as alternatives at the higher level:
2355     .sp
2356     ^(?:((.)(?1)\e2|)|((.)(?3)\e4|.))
2357 ph10 461 .sp
2358     If you want to match typical palindromic phrases, the pattern has to ignore all
2359 ph10 453 non-word characters, which can be done like this:
2360     .sp
2361 ph10 461 ^\eW*+(?:((.)\eW*+(?1)\eW*+\e2|)|((.)\eW*+(?3)\eW*+\e4|\eW*+.\eW*+))\eW*+$
2362 ph10 453 .sp
2363 ph10 461 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2364     man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2365     the use of the possessive quantifier *+ to avoid backtracking into sequences of
2366 ph10 453 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2367     more) to match typical phrases, and Perl takes so long that you think it has
2368     gone into a loop.
2369 ph10 456 .P
2370     \fBWARNING\fP: The palindrome-matching patterns above work only if the subject
2371     string does not start with a palindrome that is shorter than the entire string.
2372     For example, although "abcba" is correctly matched, if the subject is "ababa",
2373     PCRE finds the palindrome "aba" at the start, then fails at top level because
2374     the end of the string does not follow. Once again, it cannot jump back into the
2375     recursion to try other alternatives, so the entire match fails.
2376 ph10 453 .
2377     .
2378 nigel 63 .\" HTML <a name="subpatternsassubroutines"></a>
2380 nigel 63 .rs
2381     .sp
2382     If the syntax for a recursive subpattern reference (either by number or by
2383     name) is used outside the parentheses to which it refers, it operates like a
2384 nigel 93 subroutine in a programming language. The "called" subpattern may be defined
2385 ph10 166 before or after the reference. A numbered reference can be absolute or
2386     relative, as in these examples:
2387 nigel 75 .sp
2388 ph10 166 (...(absolute)...)...(?2)...
2389     (...(relative)...)...(?-1)...
2390 ph10 172 (...(?+1)...(relative)...
2391 ph10 166 .sp
2392     An earlier example pointed out that the pattern
2393     .sp
2394 nigel 75 (sens|respons)e and \e1ibility
2395     .sp
2396 nigel 63 matches "sense and sensibility" and "response and responsibility", but not
2397     "sense and responsibility". If instead the pattern
2398 nigel 75 .sp
2399 nigel 63 (sens|respons)e and (?1)ibility
2400 nigel 75 .sp
2401 nigel 63 is used, it does match "sense and responsibility" as well as the other two
2402 nigel 93 strings. Another example is given in the discussion of DEFINE above.
2403 nigel 87 .P
2404 ph10 464 Like recursive subpatterns, a subroutine call is always treated as an atomic
2405 nigel 87 group. That is, once it has matched some of the subject string, it is never
2406     re-entered, even if it contains untried alternatives and there is a subsequent
2407 ph10 469 matching failure. Any capturing parentheses that are set during the subroutine
2408 ph10 464 call revert to their previous values afterwards.
2409 nigel 93 .P
2410     When a subpattern is used as a subroutine, processing options such as
2411     case-independence are fixed when the subpattern is defined. They cannot be
2412     changed for different calls. For example, consider this pattern:
2413     .sp
2414 ph10 166 (abc)(?i:(?-1))
2415 nigel 93 .sp
2416     It matches "abcabc". It does not match "abcABC" because the change of
2417     processing option does not affect the called subpattern.
2418 nigel 75 .
2419     .
2420 ph10 333 .\" HTML <a name="onigurumasubroutines"></a>
2422     .rs
2423     .sp
2424 ph10 345 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
2425     a number enclosed either in angle brackets or single quotes, is an alternative
2426     syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2427 ph10 333 are two of the examples used above, rewritten using this syntax:
2428     .sp
2429     (?<pn> \e( ( (?>[^()]+) | \eg<pn> )* \e) )
2430     (sens|respons)e and \eg'1'ibility
2431     .sp
2432 ph10 345 PCRE supports an extension to Oniguruma: if a number is preceded by a
2433 ph10 333 plus or a minus sign it is taken as a relative reference. For example:
2434     .sp
2435     (abc)(?i:\eg<-1>)
2436     .sp
2437 ph10 345 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
2438 ph10 333 synonymous. The former is a back reference; the latter is a subroutine call.
2439     .
2440     .
2441 nigel 63 .SH CALLOUTS
2442     .rs
2443     .sp
2444     Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2445     code to be obeyed in the middle of matching a regular expression. This makes it
2446     possible, amongst other things, to extract different substrings that match the
2447     same pair of parentheses when there is a repetition.
2448 nigel 75 .P
2449 nigel 63 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2450     code. The feature is called "callout". The caller of PCRE provides an external
2451 nigel 75 function by putting its entry point in the global variable \fIpcre_callout\fP.
2452 nigel 63 By default, this variable contains NULL, which disables all calling out.
2453 nigel 75 .P
2454 nigel 63 Within a regular expression, (?C) indicates the points at which the external
2455     function is to be called. If you want to identify different callout points, you
2456     can put a number less than 256 after the letter C. The default value is zero.
2457     For example, this pattern has two callout points:
2458 nigel 75 .sp
2459 ph10 155 (?C1)abc(?C2)def
2460 nigel 75 .sp
2461     If the PCRE_AUTO_CALLOUT flag is passed to \fBpcre_compile()\fP, callouts are
2462     automatically installed before each item in the pattern. They are all numbered
2463     255.
2464     .P
2465     During matching, when PCRE reaches a callout point (and \fIpcre_callout\fP is
2466 nigel 63 set), the external function is called. It is provided with the number of the
2467 nigel 75 callout, the position in the pattern, and, optionally, one item of data
2468     originally supplied by the caller of \fBpcre_exec()\fP. The callout function
2469     may cause matching to proceed, to backtrack, or to fail altogether. A complete
2470     description of the interface to the callout function is given in the
2471 nigel 63 .\" HREF
2472 nigel 75 \fBpcrecallout\fP
2473 nigel 63 .\"
2474     documentation.
2475 nigel 93 .
2476     .
2477 ph10 510 .\" HTML <a name="backtrackcontrol"></a>
2479 ph10 210 .rs
2480     .sp
2481 ph10 211 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2482 ph10 210 are described in the Perl documentation as "experimental and subject to change
2483 ph10 211 or removal in a future version of Perl". It goes on to say: "Their usage in
2484     production code should be noted to avoid problems during upgrades." The same
2485 ph10 210 remarks apply to the PCRE features described in this section.
2486     .P
2487 ph10 341 Since these verbs are specifically related to backtracking, most of them can be
2488     used only when the pattern is to be matched using \fBpcre_exec()\fP, which uses
2489 ph10 345 a backtracking algorithm. With the exception of (*FAIL), which behaves like a
2490 ph10 341 failing negative assertion, they cause an error if encountered by
2491 ph10 210 \fBpcre_dfa_exec()\fP.
2492     .P
2493 ph10 469 If any of these verbs are used in an assertion or subroutine subpattern
2494 ph10 464 (including recursive subpatterns), their effect is confined to that subpattern;
2495 ph10 630 it does not extend to the surrounding pattern, with one exception: a *MARK that
2496     is encountered in a positive assertion \fIis\fP passed back (compare capturing
2497     parentheses in assertions). Note that such subpatterns are processed as
2498     anchored at the point where they are tested.
2499 ph10 445 .P
2500 ph10 211 The new verbs make use of what was previously invalid syntax: an opening
2501 ph10 510 parenthesis followed by an asterisk. They are generally of the form
2502 ph10 512 (*VERB) or (*VERB:NAME). Some may take either form, with differing behaviour,
2503 ph10 510 depending on whether or not an argument is present. An name is a sequence of
2504     letters, digits, and underscores. If the name is empty, that is, if the closing
2505     parenthesis immediately follows the colon, the effect is as if the colon were
2506     not there. Any number of these verbs may occur in a pattern.
2507     .P
2508 ph10 512 PCRE contains some optimizations that are used to speed up matching by running
2509     some checks at the start of each match attempt. For example, it may know the
2510     minimum length of matching subject, or that a particular character must be
2511     present. When one of these optimizations suppresses the running of a match, any
2512     included backtracking verbs will not, of course, be processed. You can suppress
2513     the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2514 ph10 577 when calling \fBpcre_compile()\fP or \fBpcre_exec()\fP, or by starting the
2515     pattern with (*NO_START_OPT).
2516 ph10 210 .
2517 ph10 510 .
2518 ph10 210 .SS "Verbs that act immediately"
2519     .rs
2520     .sp
2521 ph10 512 The following verbs act as soon as they are encountered. They may not be
2522 ph10 510 followed by a name.
2523 ph10 210 .sp
2524     (*ACCEPT)
2525     .sp
2526     This verb causes the match to end successfully, skipping the remainder of the
2527 ph10 211 pattern. When inside a recursion, only the innermost pattern is ended
2528 ph10 456 immediately. If (*ACCEPT) is inside capturing parentheses, the data so far is
2529     captured. (This feature was added to PCRE at release 8.00.) For example:
2530 ph10 210 .sp
2531 ph10 447 A((?:A|B(*ACCEPT)|C)D)
2532 ph10 210 .sp
2533 ph10 461 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2534 ph10 447 the outer parentheses.
2535 ph10 210 .sp
2536     (*FAIL) or (*F)
2537     .sp
2538 ph10 211 This verb causes the match to fail, forcing backtracking to occur. It is
2539 ph10 210 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2540     probably useful only when combined with (?{}) or (??{}). Those are, of course,
2541     Perl features that are not present in PCRE. The nearest equivalent is the
2542     callout feature, as for example in this pattern:
2543     .sp
2544     a+(?C)(*FAIL)
2545     .sp
2546 ph10 211 A match with the string "aaaa" always fails, but the callout is taken before
2547     each backtrack happens (in this example, 10 times).
2548 ph10 210 .
2549 ph10 510 .
2550     .SS "Recording which path was taken"
2551     .rs
2552     .sp
2553 ph10 512 There is one verb whose main purpose is to track how a match was arrived at,
2554     though it also has a secondary use in conjunction with advancing the match
2555 ph10 510 starting point (see (*SKIP) below).
2556     .sp
2557     (*MARK:NAME) or (*:NAME)
2558     .sp
2559     A name is always required with this verb. There may be as many instances of
2560     (*MARK) as you like in a pattern, and their names do not have to be unique.
2561     .P
2562     When a match succeeds, the name of the last-encountered (*MARK) is passed back
2563     to the caller via the \fIpcre_extra\fP data structure, as described in the
2564     .\" HTML <a href="pcreapi.html#extradata">
2565     .\" </a>
2566     section on \fIpcre_extra\fP
2567     .\"
2568 ph10 512 in the
2569 ph10 510 .\" HREF
2570     \fBpcreapi\fP
2571     .\"
2572     documentation. No data is returned for a partial match. Here is an example of
2573     \fBpcretest\fP output, where the /K modifier requests the retrieval and
2574     outputting of (*MARK) data:
2575     .sp
2576     /X(*MARK:A)Y|X(*MARK:B)Z/K
2577     XY
2578     0: XY
2579     MK: A
2580     XZ
2581     0: XZ
2582     MK: B
2583     .sp
2584 ph10 512 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2585     indicates which of the two alternatives matched. This is a more efficient way
2586 ph10 510 of obtaining this information than putting each alternative in its own
2587     capturing parentheses.
2588     .P
2589 ph10 630 If (*MARK) is encountered in a positive assertion, its name is recorded and
2590     passed back if it is the last-encountered. This does not happen for negative
2591     assetions.
2592     .P
2593 ph10 510 A name may also be returned after a failed match if the final path through the
2594     pattern involves (*MARK). However, unless (*MARK) used in conjunction with
2595     (*COMMIT), this is unlikely to happen for an unanchored pattern because, as the
2596     starting point for matching is advanced, the final check is often with an empty
2597     string, causing a failure before (*MARK) is reached. For example:
2598     .sp
2599     /X(*MARK:A)Y|X(*MARK:B)Z/K
2600     XP
2601     No match
2602     .sp
2603     There are three potential starting points for this match (starting with X,
2604 ph10 512 starting with P, and with an empty string). If the pattern is anchored, the
2605 ph10 510 result is different:
2606     .sp
2607     /^X(*MARK:A)Y|^X(*MARK:B)Z/K
2608     XP
2609     No match, mark = B
2610     .sp
2611 ph10 512 PCRE's start-of-match optimizations can also interfere with this. For example,
2612     if, as a result of a call to \fBpcre_study()\fP, it knows the minimum
2613 ph10 510 subject length for a match, a shorter subject will not be scanned at all.
2614     .P
2615 ph10 512 Note that similar anomalies (though different in detail) exist in Perl, no
2616     doubt for the same reasons. The use of (*MARK) data after a failed match of an
2617 ph10 510 unanchored pattern is not recommended, unless (*COMMIT) is involved.
2618     .
2619     .
2620 ph10 210 .SS "Verbs that act after backtracking"
2621     .rs
2622     .sp
2623 ph10 211 The following verbs do nothing when they are encountered. Matching continues
2624 ph10 510 with what follows, but if there is no subsequent match, causing a backtrack to
2625     the verb, a failure is forced. That is, backtracking cannot pass to the left of
2626     the verb. However, when one of these verbs appears inside an atomic group, its
2627     effect is confined to that group, because once the group has been matched,
2628     there is never any backtracking into it. In this situation, backtracking can
2629 ph10 512 "jump back" to the left of the entire atomic group. (Remember also, as stated
2630 ph10 510 above, that this localization also applies in subroutine calls and assertions.)
2631     .P
2632     These verbs differ in exactly what kind of failure occurs when backtracking
2633     reaches them.
2634 ph10 210 .sp
2635     (*COMMIT)
2636     .sp
2637 ph10 510 This verb, which may not be followed by a name, causes the whole match to fail
2638     outright if the rest of the pattern does not match. Even if the pattern is
2639     unanchored, no further attempts to find a match by advancing the starting point
2640     take place. Once (*COMMIT) has been passed, \fBpcre_exec()\fP is committed to
2641     finding a match at the current starting point, or not at all. For example:
2642 ph10 210 .sp
2643     a+(*COMMIT)b
2644     .sp
2645 ph10 211 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2646 ph10 512 dynamic anchor, or "I've started, so I must finish." The name of the most
2647     recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2648 ph10 510 match failure.
2649     .P
2650 ph10 512 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2651     unless PCRE's start-of-match optimizations are turned off, as shown in this
2652 ph10 510 \fBpcretest\fP example:
2653 ph10 210 .sp
2654 ph10 510 /(*COMMIT)abc/
2655     xyzabc
2656     0: abc
2657     xyzabc\eY
2658     No match
2659 ph10 210 .sp
2660 ph10 512 PCRE knows that any match must start with "a", so the optimization skips along
2661 ph10 510 the subject to "a" before running the first match attempt, which succeeds. When
2662     the optimization is disabled by the \eY escape in the second subject, the match
2663 ph10 512 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2664 ph10 510 starting points.
2665 ph10 210 .sp
2666 ph10 510 (*PRUNE) or (*PRUNE:NAME)
2667     .sp
2668 ph10 512 This verb causes the match to fail at the current starting position in the
2669 ph10 510 subject if the rest of the pattern does not match. If the pattern is
2670     unanchored, the normal "bumpalong" advance to the next starting character then
2671     happens. Backtracking can occur as usual to the left of (*PRUNE), before it is
2672     reached, or when matching to the right of (*PRUNE), but if there is no match to
2673     the right, backtracking cannot cross (*PRUNE). In simple cases, the use of
2674     (*PRUNE) is just an alternative to an atomic group or possessive quantifier,
2675     but there are some uses of (*PRUNE) that cannot be expressed in any other way.
2676     The behaviour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE) when the
2677     match fails completely; the name is passed back if this is the final attempt.
2678     (*PRUNE:NAME) does not pass back a name if the match succeeds. In an anchored
2679     pattern (*PRUNE) has the same effect as (*COMMIT).
2680     .sp
2681 ph10 210 (*SKIP)
2682     .sp
2683 ph10 510 This verb, when given without a name, is like (*PRUNE), except that if the
2684     pattern is unanchored, the "bumpalong" advance is not to the next character,
2685     but to the position in the subject where (*SKIP) was encountered. (*SKIP)
2686     signifies that whatever text was matched leading up to it cannot be part of a
2687     successful match. Consider:
2688 ph10 210 .sp
2689     a+(*SKIP)b
2690     .sp
2691 ph10 211 If the subject is "aaaac...", after the first match attempt fails (starting at
2692 ph10 210 the first character in the string), the starting point skips on to start the
2693 ph10 211 next attempt at "c". Note that a possessive quantifer does not have the same
2694 ph10 456 effect as this example; although it would suppress backtracking during the
2695 ph10 210 first match attempt, the second attempt would start at the second character
2696     instead of skipping on to "c".
2697     .sp
2698 ph10 510 (*SKIP:NAME)
2699 ph10 211 .sp
2700 ph10 512 When (*SKIP) has an associated name, its behaviour is modified. If the
2701     following pattern fails to match, the previous path through the pattern is
2702     searched for the most recent (*MARK) that has the same name. If one is found,
2703     the "bumpalong" advance is to the subject position that corresponds to that
2704     (*MARK) instead of to where (*SKIP) was encountered. If no (*MARK) with a
2705     matching name is found, normal "bumpalong" of one character happens (the
2706 ph10 510 (*SKIP) is ignored).
2707     .sp
2708     (*THEN) or (*THEN:NAME)
2709     .sp
2710 ph10 579 This verb causes a skip to the next alternation in the innermost enclosing
2711 ph10 550 group if the rest of the pattern does not match. That is, it cancels pending
2712     backtracking, but only within the current alternation. Its name comes from the
2713     observation that it can be used for a pattern-based if-then-else block:
2714 ph10 210 .sp
2715     ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2716     .sp
2717 ph10 211 If the COND1 pattern matches, FOO is tried (and possibly further items after
2718 ph10 210 the end of the group if FOO succeeds); on failure the matcher skips to the
2719 ph10 510 second alternative and tries COND2, without backtracking into COND1. The
2720     behaviour of (*THEN:NAME) is exactly the same as (*MARK:NAME)(*THEN) if the
2721     overall match fails. If (*THEN) is not directly inside an alternation, it acts
2722     like (*PRUNE).
2723 ph10 210 .
2724 ph10 551 .P
2725 ph10 579 The above verbs provide four different "strengths" of control when subsequent
2726 ph10 551 matching fails. (*THEN) is the weakest, carrying on the match at the next
2727     alternation. (*PRUNE) comes next, failing the match at the current starting
2728     position, but allowing an advance to the next character (for an unanchored
2729     pattern). (*SKIP) is similar, except that the advance may be more than one
2730     character. (*COMMIT) is the strongest, causing the entire match to fail.
2731     .P
2732     If more than one is present in a pattern, the "stongest" one wins. For example,
2733     consider this pattern, where A, B, etc. are complex pattern fragments:
2734     .sp
2735     (A(*COMMIT)B(*THEN)C|D)
2736     .sp
2737     Once A has matched, PCRE is committed to this match, at the current starting
2738     position. If subsequently B matches, but C does not, the normal (*THEN) action
2739     of trying the next alternation (that is, D) does not happen because (*COMMIT)
2740     overrides.
2741 ph10 210 .
2742 ph10 551 .
2743 nigel 93 .SH "SEE ALSO"
2744     .rs
2745     .sp
2746 ph10 461 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3),
2747 ph10 456 \fBpcresyntax\fP(3), \fBpcre\fP(3).
2748 ph10 99 .
2749     .
2750     .SH AUTHOR
2751     .rs
2752     .sp
2753     .nf
2754     Philip Hazel
2755     University Computing Service
2756     Cambridge CB2 3QH, England.
2757     .fi
2758     .
2759     .
2760     .SH REVISION
2761     .rs
2762     .sp
2763     .nf
2764 ph10 630 Last updated: 22 July 2011
2765 ph10 607 Copyright (c) 1997-2011 University of Cambridge.
2766 ph10 99 .fi


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

ViewVC Help
Powered by ViewVC 1.1.12