ViewVC logotype

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

Parent Directory Parent Directory | Revision Log Revision Log

Revision 737 - (hide annotations) (download)
Wed Oct 19 17:37:29 2011 UTC (3 years, 6 months ago) by ph10
File size: 119225 byte(s)
Add more about \C to documentation.

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 .\" HREF
36 ph10 678 \fBpcreunicode\fP
37 nigel 63 .\"
38     page.
39 nigel 75 .P
40 ph10 535 Another special sequence that may appear at the start of a pattern or in
41 ph10 518 combination with (*UTF8) is:
42     .sp
43     (*UCP)
44     .sp
45 ph10 535 This has the same effect as setting the PCRE_UCP option: it causes sequences
46     such as \ed and \ew to use Unicode properties to determine character types,
47     instead of recognizing only characters with codes less than 128 via a lookup
48 ph10 518 table.
49     .P
50 ph10 576 If a pattern starts with (*NO_START_OPT), it has the same effect as setting the
51 ph10 579 PCRE_NO_START_OPTIMIZE option either at compile or matching time. There are
52     also some more of these special sequences that are concerned with the handling
53 ph10 576 of newlines; they are described below.
54     .P
55 nigel 77 The remainder of this document discusses the patterns that are supported by
56     PCRE when its main matching function, \fBpcre_exec()\fP, is used.
57     From release 6.0, PCRE offers a second matching function,
58     \fBpcre_dfa_exec()\fP, which matches using a different algorithm that is not
59 ph10 172 Perl-compatible. Some of the features discussed below are not available when
60 ph10 168 \fBpcre_dfa_exec()\fP is used. The advantages and disadvantages of the
61     alternative function, and how it differs from the normal function, are
62     discussed in the
63 nigel 77 .\" HREF
64     \fBpcrematching\fP
65     .\"
66     page.
67 nigel 93 .
68     .
69 ph10 556 .\" HTML <a name="newlines"></a>
71     .rs
72     .sp
73     PCRE supports five different conventions for indicating line breaks in
74     strings: a single CR (carriage return) character, a single LF (linefeed)
75     character, the two-character sequence CRLF, any of the three preceding, or any
76     Unicode newline sequence. The
77     .\" HREF
78     \fBpcreapi\fP
79     .\"
80     page has
81     .\" HTML <a href="pcreapi.html#newlines">
82     .\" </a>
83     further discussion
84     .\"
85     about newlines, and shows how to set the newline convention in the
86     \fIoptions\fP arguments for the compiling and matching functions.
87     .P
88     It is also possible to specify a newline convention by starting a pattern
89     string with one of the following five sequences:
90     .sp
91     (*CR) carriage return
92     (*LF) linefeed
93     (*CRLF) carriage return, followed by linefeed
94     (*ANYCRLF) any of the three above
95     (*ANY) all Unicode newline sequences
96     .sp
97 ph10 461 These override the default and the options given to \fBpcre_compile()\fP or
98 ph10 456 \fBpcre_compile2()\fP. For example, on a Unix system where LF is the default
99     newline sequence, the pattern
100 ph10 227 .sp
101     (*CR)a.b
102     .sp
103     changes the convention to CR. That pattern matches "a\enb" because LF is no
104     longer a newline. Note that these special settings, which are not
105     Perl-compatible, are recognized only at the very start of a pattern, and that
106 ph10 231 they must be in upper case. If more than one of them is present, the last one
107     is used.
108     .P
109 ph10 514 The newline convention affects the interpretation of the dot metacharacter when
110     PCRE_DOTALL is not set, and also the behaviour of \eN. However, it does not
111     affect what the \eR escape sequence matches. By default, this is any Unicode
112     newline sequence, for Perl compatibility. However, this can be changed; see the
113     description of \eR in the section entitled
114 ph10 231 .\" HTML <a href="#newlineseq">
115     .\" </a>
116     "Newline sequences"
117     .\"
118 ph10 247 below. A change of \eR setting can be combined with a change of newline
119 ph10 246 convention.
120 ph10 227 .
121     .
123     .rs
124     .sp
125 nigel 63 A regular expression is a pattern that is matched against a subject string from
126     left to right. Most characters stand for themselves in a pattern, and match the
127     corresponding characters in the subject. As a trivial example, the pattern
128 nigel 75 .sp
129 nigel 63 The quick brown fox
130 nigel 75 .sp
131 nigel 77 matches a portion of a subject string that is identical to itself. When
132     caseless matching is specified (the PCRE_CASELESS option), letters are matched
133     independently of case. In UTF-8 mode, PCRE always understands the concept of
134     case for characters whose values are less than 128, so caseless matching is
135     always possible. For characters with higher values, the concept of case is
136     supported if PCRE is compiled with Unicode property support, but not otherwise.
137     If you want to use caseless matching for characters 128 and above, you must
138     ensure that PCRE is compiled with Unicode property support as well as with
139     UTF-8 support.
140     .P
141     The power of regular expressions comes from the ability to include alternatives
142     and repetitions in the pattern. These are encoded in the pattern by the use of
143 nigel 75 \fImetacharacters\fP, which do not stand for themselves but instead are
144 nigel 63 interpreted in some special way.
145 nigel 75 .P
146     There are two different sets of metacharacters: those that are recognized
147 nigel 63 anywhere in the pattern except within square brackets, and those that are
148 nigel 93 recognized within square brackets. Outside square brackets, the metacharacters
149     are as follows:
150 nigel 75 .sp
151     \e general escape character with several uses
152 nigel 63 ^ assert start of string (or line, in multiline mode)
153     $ assert end of string (or line, in multiline mode)
154     . match any character except newline (by default)
155     [ start character class definition
156     | start of alternative branch
157     ( start subpattern
158     ) end subpattern
159     ? extends the meaning of (
160     also 0 or 1 quantifier
161     also quantifier minimizer
162     * 0 or more quantifier
163     + 1 or more quantifier
164     also "possessive quantifier"
165     { start min/max quantifier
166 nigel 75 .sp
167 nigel 63 Part of a pattern that is in square brackets is called a "character class". In
168 nigel 75 a character class the only metacharacters are:
169     .sp
170     \e general escape character
171 nigel 63 ^ negate the class, but only if the first character
172     - indicates character range
173 nigel 75 .\" JOIN
174 nigel 63 [ POSIX character class (only if followed by POSIX
175     syntax)
176     ] terminates the character class
177 nigel 75 .sp
178     The following sections describe the use of each of the metacharacters.
179     .
180 nigel 93 .
181 nigel 63 .SH BACKSLASH
182     .rs
183     .sp
184     The backslash character has several uses. Firstly, if it is followed by a
185 ph10 574 character that is not a number or a letter, it takes away any special meaning
186     that character may have. This use of backslash as an escape character applies
187 ph10 579 both inside and outside character classes.
188 nigel 75 .P
189     For example, if you want to match a * character, you write \e* in the pattern.
190 nigel 63 This escaping action applies whether or not the following character would
191 nigel 75 otherwise be interpreted as a metacharacter, so it is always safe to precede a
192     non-alphanumeric with backslash to specify that it stands for itself. In
193     particular, if you want to match a backslash, you write \e\e.
194     .P
195 ph10 574 In UTF-8 mode, only ASCII numbers and letters have any special meaning after a
196 ph10 579 backslash. All other characters (in particular, those whose codepoints are
197 ph10 574 greater than 127) are treated as literals.
198     .P
199 nigel 63 If a pattern is compiled with the PCRE_EXTENDED option, whitespace in the
200     pattern (other than in a character class) and characters between a # outside
201 nigel 91 a character class and the next newline are ignored. An escaping backslash can
202     be used to include a whitespace or # character as part of the pattern.
203 nigel 75 .P
204 nigel 63 If you want to remove the special meaning from a sequence of characters, you
205 nigel 75 can do so by putting them between \eQ and \eE. This is different from Perl in
206     that $ and @ are handled as literals in \eQ...\eE sequences in PCRE, whereas in
207 nigel 63 Perl, $ and @ cause variable interpolation. Note the following examples:
208 nigel 75 .sp
209 nigel 63 Pattern PCRE matches Perl matches
210 nigel 75 .sp
211     .\" JOIN
212     \eQabc$xyz\eE abc$xyz abc followed by the
213 nigel 63 contents of $xyz
214 nigel 75 \eQabc\e$xyz\eE abc\e$xyz abc\e$xyz
215     \eQabc\eE\e$\eQxyz\eE abc$xyz abc$xyz
216     .sp
217     The \eQ...\eE sequence is recognized both inside and outside character classes.
218 ph10 654 An isolated \eE that is not preceded by \eQ is ignored. If \eQ is not followed
219     by \eE later in the pattern, the literal interpretation continues to the end of
220 ph10 607 the pattern (that is, \eE is assumed at the end). If the isolated \eQ is inside
221     a character class, this causes an error, because the character class is not
222     terminated.
223 nigel 75 .
224     .
225     .\" HTML <a name="digitsafterbackslash"></a>
226     .SS "Non-printing characters"
227     .rs
228     .sp
229 nigel 63 A second use of backslash provides a way of encoding non-printing characters
230     in patterns in a visible manner. There is no restriction on the appearance of
231     non-printing characters, apart from the binary zero that terminates a pattern,
232 ph10 456 but when a pattern is being prepared by text editing, it is often easier to use
233     one of the following escape sequences than the binary character it represents:
234 nigel 75 .sp
235     \ea alarm, that is, the BEL character (hex 07)
236 ph10 574 \ecx "control-x", where x is any ASCII character
237 nigel 75 \ee escape (hex 1B)
238     \ef formfeed (hex 0C)
239 ph10 227 \en linefeed (hex 0A)
240 nigel 75 \er carriage return (hex 0D)
241     \et tab (hex 09)
242 ph10 488 \eddd character with octal code ddd, or back reference
243 nigel 75 \exhh character with hex code hh
244 nigel 87 \ex{hhh..} character with hex code hhh..
245 nigel 75 .sp
246     The precise effect of \ecx is as follows: if x is a lower case letter, it
247 nigel 63 is converted to upper case. Then bit 6 of the character (hex 40) is inverted.
248 ph10 574 Thus \ecz becomes hex 1A (z is 7A), but \ec{ becomes hex 3B ({ is 7B), while
249 ph10 579 \ec; becomes hex 7B (; is 3B). If the byte following \ec has a value greater
250     than 127, a compile-time error occurs. This locks out non-ASCII characters in
251     both byte mode and UTF-8 mode. (When PCRE is compiled in EBCDIC mode, all byte
252     values are valid. A lower case letter is converted to upper case, and then the
253 ph10 574 0xc0 bits are flipped.)
254 nigel 75 .P
255     After \ex, from zero to two hexadecimal digits are read (letters can be in
256 nigel 87 upper or lower case). Any number of hexadecimal digits may appear between \ex{
257     and }, but the value of the character code must be less than 256 in non-UTF-8
258 ph10 211 mode, and less than 2**31 in UTF-8 mode. That is, the maximum value in
259     hexadecimal is 7FFFFFFF. Note that this is bigger than the largest Unicode code
260     point, which is 10FFFF.
261 nigel 75 .P
262 ph10 211 If characters other than hexadecimal digits appear between \ex{ and }, or if
263     there is no terminating }, this form of escape is not recognized. Instead, the
264     initial \ex will be interpreted as a basic hexadecimal escape, with no
265     following digits, giving a character whose value is zero.
266     .P
267 nigel 63 Characters whose value is less than 256 can be defined by either of the two
268 nigel 87 syntaxes for \ex. There is no difference in the way they are handled. For
269     example, \exdc is exactly the same as \ex{dc}.
270 nigel 75 .P
271 nigel 91 After \e0 up to two further octal digits are read. If there are fewer than two
272     digits, just those that are present are used. Thus the sequence \e0\ex\e07
273     specifies two binary zeros followed by a BEL character (code value 7). Make
274     sure you supply two digits after the initial zero if the pattern character that
275     follows is itself an octal digit.
276 nigel 75 .P
277 nigel 63 The handling of a backslash followed by a digit other than 0 is complicated.
278     Outside a character class, PCRE reads it and any following digits as a decimal
279     number. If the number is less than 10, or if there have been at least that many
280     previous capturing left parentheses in the expression, the entire sequence is
281 nigel 75 taken as a \fIback reference\fP. A description of how this works is given
282     .\" HTML <a href="#backreferences">
283     .\" </a>
284     later,
285     .\"
286     following the discussion of
287     .\" HTML <a href="#subpattern">
288     .\" </a>
289     parenthesized subpatterns.
290     .\"
291     .P
292 nigel 63 Inside a character class, or if the decimal number is greater than 9 and there
293     have not been that many capturing subpatterns, PCRE re-reads up to three octal
294 nigel 93 digits following the backslash, and uses them to generate a data character. Any
295 nigel 91 subsequent digits stand for themselves. In non-UTF-8 mode, the value of a
296     character specified in octal must be less than \e400. In UTF-8 mode, values up
297     to \e777 are permitted. For example:
298 nigel 75 .sp
299     \e040 is another way of writing a space
300     .\" JOIN
301     \e40 is the same, provided there are fewer than 40
302 nigel 63 previous capturing subpatterns
303 nigel 75 \e7 is always a back reference
304     .\" JOIN
305     \e11 might be a back reference, or another way of
306 nigel 63 writing a tab
307 nigel 75 \e011 is always a tab
308     \e0113 is a tab followed by the character "3"
309     .\" JOIN
310     \e113 might be a back reference, otherwise the
311 nigel 63 character with octal code 113
312 nigel 75 .\" JOIN
313     \e377 might be a back reference, otherwise
314 nigel 63 the byte consisting entirely of 1 bits
315 nigel 75 .\" JOIN
316     \e81 is either a back reference, or a binary zero
317 nigel 63 followed by the two characters "8" and "1"
318 nigel 75 .sp
319 nigel 63 Note that octal values of 100 or greater must not be introduced by a leading
320     zero, because no more than three octal digits are ever read.
321 nigel 75 .P
322 nigel 91 All the sequences that define a single character value can be used both inside
323     and outside character classes. In addition, inside a character class, the
324 ph10 513 sequence \eb is interpreted as the backspace character (hex 08). The sequences
325 ph10 514 \eB, \eN, \eR, and \eX are not special inside a character class. Like any other
326 ph10 513 unrecognized escape sequences, they are treated as the literal characters "B",
327 ph10 514 "N", "R", and "X" by default, but cause an error if the PCRE_EXTRA option is
328     set. Outside a character class, these sequences have different meanings.
329 nigel 75 .
330     .
331 nigel 93 .SS "Absolute and relative back references"
332     .rs
333     .sp
334 ph10 208 The sequence \eg followed by an unsigned or a negative number, optionally
335     enclosed in braces, is an absolute or relative back reference. A named back
336     reference can be coded as \eg{name}. Back references are discussed
337 nigel 93 .\" HTML <a href="#backreferences">
338     .\" </a>
339     later,
340     .\"
341     following the discussion of
342     .\" HTML <a href="#subpattern">
343     .\" </a>
344     parenthesized subpatterns.
345     .\"
346     .
347     .
348 ph10 333 .SS "Absolute and relative subroutine calls"
349     .rs
350     .sp
351 ph10 345 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
352     a number enclosed either in angle brackets or single quotes, is an alternative
353     syntax for referencing a subpattern as a "subroutine". Details are discussed
354 ph10 333 .\" HTML <a href="#onigurumasubroutines">
355     .\" </a>
356     later.
357     .\"
358 ph10 345 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
359 ph10 461 synonymous. The former is a back reference; the latter is a
360 ph10 454 .\" HTML <a href="#subpatternsassubroutines">
361     .\" </a>
362     subroutine
363     .\"
364     call.
365 ph10 333 .
366     .
367 ph10 518 .\" HTML <a name="genericchartypes"></a>
368 nigel 75 .SS "Generic character types"
369     .rs
370     .sp
371 ph10 514 Another use of backslash is for specifying generic character types:
372 nigel 75 .sp
373 ph10 182 \ed any decimal digit
374 nigel 75 \eD any character that is not a decimal digit
375 ph10 178 \eh any horizontal whitespace character
376 ph10 182 \eH any character that is not a horizontal whitespace character
377 nigel 75 \es any whitespace character
378     \eS any character that is not a whitespace character
379 ph10 178 \ev any vertical whitespace character
380 ph10 182 \eV any character that is not a vertical whitespace character
381 nigel 75 \ew any "word" character
382     \eW any "non-word" character
383     .sp
384 ph10 535 There is also the single sequence \eN, which matches a non-newline character.
385     This is the same as
386 ph10 514 .\" HTML <a href="#fullstopdot">
387     .\" </a>
388 ph10 535 the "." metacharacter
389 ph10 514 .\"
390     when PCRE_DOTALL is not set.
391 nigel 75 .P
392 ph10 514 Each pair of lower and upper case escape sequences partitions the complete set
393     of characters into two disjoint sets. Any given character matches one, and only
394 ph10 518 one, of each pair. The sequences can appear both inside and outside character
395 nigel 75 classes. They each match one character of the appropriate type. If the current
396 ph10 518 matching point is at the end of the subject string, all of them fail, because
397 nigel 75 there is no character to match.
398     .P
399     For compatibility with Perl, \es does not match the VT character (code 11).
400     This makes it different from the the POSIX "space" class. The \es characters
401 ph10 178 are HT (9), LF (10), FF (12), CR (13), and space (32). If "use locale;" is
402 nigel 91 included in a Perl script, \es may match the VT character. In PCRE, it never
403 ph10 178 does.
404 nigel 75 .P
405 ph10 518 A "word" character is an underscore or any character that is a letter or digit.
406     By default, the definition of letters and digits is controlled by PCRE's
407     low-valued character tables, and may vary if locale-specific matching is taking
408     place (see
409     .\" HTML <a href="pcreapi.html#localesupport">
410     .\" </a>
411     "Locale support"
412     .\"
413     in the
414     .\" HREF
415     \fBpcreapi\fP
416     .\"
417     page). For example, in a French locale such as "fr_FR" in Unix-like systems,
418     or "french" in Windows, some character codes greater than 128 are used for
419     accented letters, and these are then matched by \ew. The use of locales with
420     Unicode is discouraged.
421 ph10 178 .P
422 ph10 518 By default, in UTF-8 mode, characters with values greater than 128 never match
423     \ed, \es, or \ew, and always match \eD, \eS, and \eW. These sequences retain
424     their original meanings from before UTF-8 support was available, mainly for
425 ph10 535 efficiency reasons. However, if PCRE is compiled with Unicode property support,
426 ph10 518 and the PCRE_UCP option is set, the behaviour is changed so that Unicode
427     properties are used to determine character types, as follows:
428     .sp
429     \ed any character that \ep{Nd} matches (decimal digit)
430     \es any character that \ep{Z} matches, plus HT, LF, FF, CR
431     \ew any character that \ep{L} or \ep{N} matches, plus underscore
432     .sp
433     The upper case escapes match the inverse sets of characters. Note that \ed
434 ph10 535 matches only decimal digits, whereas \ew matches any Unicode digit, as well as
435 ph10 518 any Unicode letter, and underscore. Note also that PCRE_UCP affects \eb, and
436     \eB because they are defined in terms of \ew and \eW. Matching these sequences
437     is noticeably slower when PCRE_UCP is set.
438     .P
439 ph10 579 The sequences \eh, \eH, \ev, and \eV are features that were added to Perl at
440 ph10 572 release 5.10. In contrast to the other sequences, which match only ASCII
441     characters by default, these always match certain high-valued codepoints in
442     UTF-8 mode, whether or not PCRE_UCP is set. The horizontal space characters
443     are:
444 ph10 178 .sp
445     U+0009 Horizontal tab
446     U+0020 Space
447     U+00A0 Non-break space
448     U+1680 Ogham space mark
449     U+180E Mongolian vowel separator
450     U+2000 En quad
451     U+2001 Em quad
452     U+2002 En space
453     U+2003 Em space
454     U+2004 Three-per-em space
455     U+2005 Four-per-em space
456     U+2006 Six-per-em space
457     U+2007 Figure space
458     U+2008 Punctuation space
459     U+2009 Thin space
460     U+200A Hair space
461     U+202F Narrow no-break space
462     U+205F Medium mathematical space
463     U+3000 Ideographic space
464     .sp
465     The vertical space characters are:
466     .sp
467     U+000A Linefeed
468     U+000B Vertical tab
469     U+000C Formfeed
470     U+000D Carriage return
471     U+0085 Next line
472     U+2028 Line separator
473     U+2029 Paragraph separator
474 nigel 75 .
475     .
476 ph10 231 .\" HTML <a name="newlineseq"></a>
477 nigel 93 .SS "Newline sequences"
478     .rs
479     .sp
480 ph10 231 Outside a character class, by default, the escape sequence \eR matches any
481 ph10 572 Unicode newline sequence. In non-UTF-8 mode \eR is equivalent to the following:
482 nigel 93 .sp
483     (?>\er\en|\en|\ex0b|\ef|\er|\ex85)
484     .sp
485     This is an example of an "atomic group", details of which are given
486     .\" HTML <a href="#atomicgroup">
487     .\" </a>
488     below.
489     .\"
490     This particular group matches either the two-character sequence CR followed by
491     LF, or one of the single characters LF (linefeed, U+000A), VT (vertical tab,
492     U+000B), FF (formfeed, U+000C), CR (carriage return, U+000D), or NEL (next
493     line, U+0085). The two-character sequence is treated as a single unit that
494     cannot be split.
495     .P
496     In UTF-8 mode, two additional characters whose codepoints are greater than 255
497     are added: LS (line separator, U+2028) and PS (paragraph separator, U+2029).
498     Unicode character property support is not needed for these characters to be
499     recognized.
500     .P
501 ph10 231 It is possible to restrict \eR to match only CR, LF, or CRLF (instead of the
502     complete set of Unicode line endings) by setting the option PCRE_BSR_ANYCRLF
503 ph10 247 either at compile time or when the pattern is matched. (BSR is an abbrevation
504 ph10 246 for "backslash R".) This can be made the default when PCRE is built; if this is
505     the case, the other behaviour can be requested via the PCRE_BSR_UNICODE option.
506     It is also possible to specify these settings by starting a pattern string with
507     one of the following sequences:
508 ph10 231 .sp
509     (*BSR_ANYCRLF) CR, LF, or CRLF only
510     (*BSR_UNICODE) any Unicode newline sequence
511     .sp
512 ph10 461 These override the default and the options given to \fBpcre_compile()\fP or
513 ph10 456 \fBpcre_compile2()\fP, but they can be overridden by options given to
514     \fBpcre_exec()\fP or \fBpcre_dfa_exec()\fP. Note that these special settings,
515     which are not Perl-compatible, are recognized only at the very start of a
516     pattern, and that they must be in upper case. If more than one of them is
517     present, the last one is used. They can be combined with a change of newline
518 ph10 518 convention; for example, a pattern can start with:
519 ph10 246 .sp
520     (*ANY)(*BSR_ANYCRLF)
521     .sp
522 ph10 518 They can also be combined with the (*UTF8) or (*UCP) special sequences. Inside
523     a character class, \eR is treated as an unrecognized escape sequence, and so
524     matches the letter "R" by default, but causes an error if PCRE_EXTRA is set.
525 nigel 93 .
526     .
527 nigel 75 .\" HTML <a name="uniextseq"></a>
528     .SS Unicode character properties
529     .rs
530     .sp
531     When PCRE is built with Unicode character property support, three additional
532 ph10 184 escape sequences that match characters with specific properties are available.
533     When not in UTF-8 mode, these sequences are of course limited to testing
534     characters whose codepoints are less than 256, but they do work in this mode.
535     The extra escape sequences are:
536 nigel 75 .sp
537 nigel 87 \ep{\fIxx\fP} a character with the \fIxx\fP property
538     \eP{\fIxx\fP} a character without the \fIxx\fP property
539     \eX an extended Unicode sequence
540 nigel 75 .sp
541 nigel 87 The property names represented by \fIxx\fP above are limited to the Unicode
542 ph10 517 script names, the general category properties, "Any", which matches any
543     character (including newline), and some special PCRE properties (described
544 ph10 535 in the
545 ph10 517 .\" HTML <a href="#extraprops">
546     .\" </a>
547 ph10 535 next section).
548 ph10 517 .\"
549     Other Perl properties such as "InMusicalSymbols" are not currently supported by
550     PCRE. Note that \eP{Any} does not match any characters, so always causes a
551     match failure.
552 nigel 75 .P
553 nigel 87 Sets of Unicode characters are defined as belonging to certain scripts. A
554     character from one of these sets can be matched using a script name. For
555     example:
556 nigel 75 .sp
557 nigel 87 \ep{Greek}
558     \eP{Han}
559     .sp
560     Those that are not part of an identified script are lumped together as
561     "Common". The current list of scripts is:
562     .P
563     Arabic,
564     Armenian,
565 ph10 491 Avestan,
566 nigel 93 Balinese,
567 ph10 491 Bamum,
568 nigel 87 Bengali,
569     Bopomofo,
570     Braille,
571     Buginese,
572     Buhid,
573     Canadian_Aboriginal,
574 ph10 491 Carian,
575     Cham,
576 nigel 87 Cherokee,
577     Common,
578     Coptic,
579 nigel 93 Cuneiform,
580 nigel 87 Cypriot,
581     Cyrillic,
582     Deseret,
583     Devanagari,
584 ph10 491 Egyptian_Hieroglyphs,
585 nigel 87 Ethiopic,
586     Georgian,
587     Glagolitic,
588     Gothic,
589     Greek,
590     Gujarati,
591     Gurmukhi,
592     Han,
593     Hangul,
594     Hanunoo,
595     Hebrew,
596     Hiragana,
597 ph10 491 Imperial_Aramaic,
598 nigel 87 Inherited,
599 ph10 491 Inscriptional_Pahlavi,
600     Inscriptional_Parthian,
601     Javanese,
602     Kaithi,
603 nigel 87 Kannada,
604     Katakana,
605 ph10 491 Kayah_Li,
606 nigel 87 Kharoshthi,
607     Khmer,
608     Lao,
609     Latin,
610 ph10 491 Lepcha,
611 nigel 87 Limbu,
612     Linear_B,
613 ph10 491 Lisu,
614     Lycian,
615     Lydian,
616 nigel 87 Malayalam,
617 ph10 491 Meetei_Mayek,
618 nigel 87 Mongolian,
619     Myanmar,
620     New_Tai_Lue,
621 nigel 93 Nko,
622 nigel 87 Ogham,
623     Old_Italic,
624     Old_Persian,
625 ph10 491 Old_South_Arabian,
626     Old_Turkic,
627     Ol_Chiki,
628 nigel 87 Oriya,
629     Osmanya,
630 nigel 93 Phags_Pa,
631     Phoenician,
632 ph10 491 Rejang,
633 nigel 87 Runic,
634 ph10 491 Samaritan,
635     Saurashtra,
636 nigel 87 Shavian,
637     Sinhala,
638 ph10 491 Sundanese,
639 nigel 87 Syloti_Nagri,
640     Syriac,
641     Tagalog,
642     Tagbanwa,
643     Tai_Le,
644 ph10 491 Tai_Tham,
645     Tai_Viet,
646 nigel 87 Tamil,
647     Telugu,
648     Thaana,
649     Thai,
650     Tibetan,
651     Tifinagh,
652     Ugaritic,
653 ph10 491 Vai,
654 nigel 87 Yi.
655     .P
656 ph10 517 Each character has exactly one Unicode general category property, specified by
657     a two-letter abbreviation. For compatibility with Perl, negation can be
658     specified by including a circumflex between the opening brace and the property
659     name. For example, \ep{^Lu} is the same as \eP{Lu}.
660 nigel 87 .P
661     If only one letter is specified with \ep or \eP, it includes all the general
662     category properties that start with that letter. In this case, in the absence
663     of negation, the curly brackets in the escape sequence are optional; these two
664     examples have the same effect:
665     .sp
666 nigel 75 \ep{L}
667     \epL
668     .sp
669 nigel 87 The following general category property codes are supported:
670 nigel 75 .sp
671     C Other
672     Cc Control
673     Cf Format
674     Cn Unassigned
675     Co Private use
676     Cs Surrogate
677     .sp
678     L Letter
679     Ll Lower case letter
680     Lm Modifier letter
681     Lo Other letter
682     Lt Title case letter
683     Lu Upper case letter
684     .sp
685     M Mark
686     Mc Spacing mark
687     Me Enclosing mark
688     Mn Non-spacing mark
689     .sp
690     N Number
691     Nd Decimal number
692     Nl Letter number
693     No Other number
694     .sp
695     P Punctuation
696     Pc Connector punctuation
697     Pd Dash punctuation
698     Pe Close punctuation
699     Pf Final punctuation
700     Pi Initial punctuation
701     Po Other punctuation
702     Ps Open punctuation
703     .sp
704     S Symbol
705     Sc Currency symbol
706     Sk Modifier symbol
707     Sm Mathematical symbol
708     So Other symbol
709     .sp
710     Z Separator
711     Zl Line separator
712     Zp Paragraph separator
713     Zs Space separator
714     .sp
715 nigel 87 The special property L& is also supported: it matches a character that has
716     the Lu, Ll, or Lt property, in other words, a letter that is not classified as
717     a modifier or "other".
718 nigel 75 .P
719 ph10 211 The Cs (Surrogate) property applies only to characters in the range U+D800 to
720     U+DFFF. Such characters are not valid in UTF-8 strings (see RFC 3629) and so
721     cannot be tested by PCRE, unless UTF-8 validity checking has been turned off
722     (see the discussion of PCRE_NO_UTF8_CHECK in the
723     .\" HREF
724     \fBpcreapi\fP
725     .\"
726 ph10 451 page). Perl does not support the Cs property.
727 ph10 211 .P
728 ph10 451 The long synonyms for property names that Perl supports (such as \ep{Letter})
729 nigel 91 are not supported by PCRE, nor is it permitted to prefix any of these
730 nigel 87 properties with "Is".
731     .P
732     No character that is in the Unicode table has the Cn (unassigned) property.
733     Instead, this property is assumed for any code point that is not in the
734     Unicode table.
735     .P
736 nigel 75 Specifying caseless matching does not affect these escape sequences. For
737     example, \ep{Lu} always matches only upper case letters.
738     .P
739     The \eX escape matches any number of Unicode characters that form an extended
740     Unicode sequence. \eX is equivalent to
741     .sp
742     (?>\ePM\epM*)
743     .sp
744     That is, it matches a character without the "mark" property, followed by zero
745     or more characters with the "mark" property, and treats the sequence as an
746     atomic group
747     .\" HTML <a href="#atomicgroup">
748     .\" </a>
749     (see below).
750     .\"
751     Characters with the "mark" property are typically accents that affect the
752 ph10 185 preceding character. None of them have codepoints less than 256, so in
753 ph10 184 non-UTF-8 mode \eX matches any one character.
754 nigel 75 .P
755 ph10 654 Note that recent versions of Perl have changed \eX to match what Unicode calls
756 ph10 628 an "extended grapheme cluster", which has a more complicated definition.
757     .P
758 nigel 75 Matching characters by Unicode property is not fast, because PCRE has to search
759     a structure that contains data for over fifteen thousand characters. That is
760     why the traditional escape sequences such as \ed and \ew do not use Unicode
761 ph10 535 properties in PCRE by default, though you can make them do so by setting the
762 ph10 518 PCRE_UCP option for \fBpcre_compile()\fP or by starting the pattern with
763     (*UCP).
764 nigel 75 .
765     .
766 ph10 517 .\" HTML <a name="extraprops"></a>
767     .SS PCRE's additional properties
768     .rs
769     .sp
770 ph10 535 As well as the standard Unicode properties described in the previous
771     section, PCRE supports four more that make it possible to convert traditional
772 ph10 517 escape sequences such as \ew and \es and POSIX character classes to use Unicode
773 ph10 518 properties. PCRE uses these non-standard, non-Perl properties internally when
774     PCRE_UCP is set. They are:
775 ph10 517 .sp
776     Xan Any alphanumeric character
777     Xps Any POSIX space character
778     Xsp Any Perl space character
779     Xwd Any Perl "word" character
780     .sp
781 ph10 535 Xan matches characters that have either the L (letter) or the N (number)
782     property. Xps matches the characters tab, linefeed, vertical tab, formfeed, or
783 ph10 517 carriage return, and any other character that has the Z (separator) property.
784 ph10 535 Xsp is the same as Xps, except that vertical tab is excluded. Xwd matches the
785 ph10 517 same characters as Xan, plus underscore.
786     .
787     .
788 ph10 168 .\" HTML <a name="resetmatchstart"></a>
789     .SS "Resetting the match start"
790     .rs
791     .sp
792 ph10 572 The escape sequence \eK causes any previously matched characters not to be
793     included in the final matched sequence. For example, the pattern:
794 ph10 168 .sp
795     foo\eKbar
796     .sp
797 ph10 172 matches "foobar", but reports that it has matched "bar". This feature is
798 ph10 168 similar to a lookbehind assertion
799     .\" HTML <a href="#lookbehind">
800     .\" </a>
801     (described below).
802     .\"
803 ph10 172 However, in this case, the part of the subject before the real match does not
804     have to be of fixed length, as lookbehind assertions do. The use of \eK does
805 ph10 168 not interfere with the setting of
806     .\" HTML <a href="#subpattern">
807     .\" </a>
808     captured substrings.
809 ph10 172 .\"
810 ph10 168 For example, when the pattern
811     .sp
812     (foo)\eKbar
813     .sp
814 ph10 172 matches "foobar", the first substring is still set to "foo".
815 ph10 500 .P
816 ph10 507 Perl documents that the use of \eK within assertions is "not well defined". In
817     PCRE, \eK is acted upon when it occurs inside positive assertions, but is
818 ph10 500 ignored in negative assertions.
819 ph10 168 .
820     .
821 nigel 75 .\" HTML <a name="smallassertions"></a>
822     .SS "Simple assertions"
823     .rs
824     .sp
825 nigel 93 The final use of backslash is for certain simple assertions. An assertion
826 nigel 63 specifies a condition that has to be met at a particular point in a match,
827     without consuming any characters from the subject string. The use of
828 nigel 75 subpatterns for more complicated assertions is described
829     .\" HTML <a href="#bigassertions">
830     .\" </a>
831     below.
832     .\"
833 nigel 91 The backslashed assertions are:
834 nigel 75 .sp
835     \eb matches at a word boundary
836     \eB matches when not at a word boundary
837 nigel 93 \eA matches at the start of the subject
838     \eZ matches at the end of the subject
839     also matches before a newline at the end of the subject
840     \ez matches only at the end of the subject
841     \eG matches at the first matching position in the subject
842 nigel 75 .sp
843 ph10 513 Inside a character class, \eb has a different meaning; it matches the backspace
844 ph10 535 character. If any other of these assertions appears in a character class, by
845 ph10 513 default it matches the corresponding literal character (for example, \eB
846     matches the letter B). However, if the PCRE_EXTRA option is set, an "invalid
847     escape sequence" error is generated instead.
848 nigel 75 .P
849 nigel 63 A word boundary is a position in the subject string where the current character
850 nigel 75 and the previous character do not both match \ew or \eW (i.e. one matches
851     \ew and the other matches \eW), or the start or end of the string if the
852 ph10 518 first or last character matches \ew, respectively. In UTF-8 mode, the meanings
853     of \ew and \eW can be changed by setting the PCRE_UCP option. When this is
854     done, it also affects \eb and \eB. Neither PCRE nor Perl has a separate "start
855     of word" or "end of word" metasequence. However, whatever follows \eb normally
856     determines which it is. For example, the fragment \eba matches "a" at the start
857     of a word.
858 nigel 75 .P
859     The \eA, \eZ, and \ez assertions differ from the traditional circumflex and
860     dollar (described in the next section) in that they only ever match at the very
861     start and end of the subject string, whatever options are set. Thus, they are
862     independent of multiline mode. These three assertions are not affected by the
863     PCRE_NOTBOL or PCRE_NOTEOL options, which affect only the behaviour of the
864     circumflex and dollar metacharacters. However, if the \fIstartoffset\fP
865     argument of \fBpcre_exec()\fP is non-zero, indicating that matching is to start
866     at a point other than the beginning of the subject, \eA can never match. The
867 nigel 91 difference between \eZ and \ez is that \eZ matches before a newline at the end
868     of the string as well as at the very end, whereas \ez matches only at the end.
869 nigel 75 .P
870     The \eG assertion is true only when the current matching position is at the
871     start point of the match, as specified by the \fIstartoffset\fP argument of
872     \fBpcre_exec()\fP. It differs from \eA when the value of \fIstartoffset\fP is
873     non-zero. By calling \fBpcre_exec()\fP multiple times with appropriate
874 nigel 63 arguments, you can mimic Perl's /g option, and it is in this kind of
875 nigel 75 implementation where \eG can be useful.
876     .P
877     Note, however, that PCRE's interpretation of \eG, as the start of the current
878 nigel 63 match, is subtly different from Perl's, which defines it as the end of the
879     previous match. In Perl, these can be different when the previously matched
880     string was empty. Because PCRE does just one match at a time, it cannot
881     reproduce this behaviour.
882 nigel 75 .P
883     If all the alternatives of a pattern begin with \eG, the expression is anchored
884 nigel 63 to the starting match position, and the "anchored" flag is set in the compiled
885     regular expression.
886 nigel 75 .
887     .
889 nigel 63 .rs
890     .sp
891     Outside a character class, in the default matching mode, the circumflex
892 nigel 75 character is an assertion that is true only if the current matching point is
893     at the start of the subject string. If the \fIstartoffset\fP argument of
894     \fBpcre_exec()\fP is non-zero, circumflex can never match if the PCRE_MULTILINE
895 nigel 63 option is unset. Inside a character class, circumflex has an entirely different
896 nigel 75 meaning
897     .\" HTML <a href="#characterclass">
898     .\" </a>
899     (see below).
900     .\"
901     .P
902 nigel 63 Circumflex need not be the first character of the pattern if a number of
903     alternatives are involved, but it should be the first thing in each alternative
904     in which it appears if the pattern is ever to match that branch. If all
905     possible alternatives start with a circumflex, that is, if the pattern is
906     constrained to match only at the start of the subject, it is said to be an
907     "anchored" pattern. (There are also other constructs that can cause a pattern
908     to be anchored.)
909 nigel 75 .P
910     A dollar character is an assertion that is true only if the current matching
911 nigel 63 point is at the end of the subject string, or immediately before a newline
912 nigel 91 at the end of the string (by default). Dollar need not be the last character of
913     the pattern if a number of alternatives are involved, but it should be the last
914     item in any branch in which it appears. Dollar has no special meaning in a
915     character class.
916 nigel 75 .P
917 nigel 63 The meaning of dollar can be changed so that it matches only at the very end of
918     the string, by setting the PCRE_DOLLAR_ENDONLY option at compile time. This
919 nigel 75 does not affect the \eZ assertion.
920     .P
921 nigel 63 The meanings of the circumflex and dollar characters are changed if the
922 nigel 91 PCRE_MULTILINE option is set. When this is the case, a circumflex matches
923     immediately after internal newlines as well as at the start of the subject
924     string. It does not match after a newline that ends the string. A dollar
925     matches before any newlines in the string, as well as at the very end, when
926     PCRE_MULTILINE is set. When newline is specified as the two-character
927     sequence CRLF, isolated CR and LF characters do not indicate newlines.
928 nigel 75 .P
929 nigel 91 For example, the pattern /^abc$/ matches the subject string "def\enabc" (where
930     \en represents a newline) in multiline mode, but not otherwise. Consequently,
931     patterns that are anchored in single line mode because all branches start with
932     ^ are not anchored in multiline mode, and a match for circumflex is possible
933     when the \fIstartoffset\fP argument of \fBpcre_exec()\fP is non-zero. The
934     PCRE_DOLLAR_ENDONLY option is ignored if PCRE_MULTILINE is set.
935     .P
936 nigel 75 Note that the sequences \eA, \eZ, and \ez can be used to match the start and
937 nigel 63 end of the subject in both modes, and if all branches of a pattern start with
938 nigel 91 \eA it is always anchored, whether or not PCRE_MULTILINE is set.
939 nigel 75 .
940     .
941 ph10 514 .\" HTML <a name="fullstopdot"></a>
943 nigel 63 .rs
944     .sp
945     Outside a character class, a dot in the pattern matches any one character in
946 nigel 91 the subject string except (by default) a character that signifies the end of a
947 nigel 93 line. In UTF-8 mode, the matched character may be more than one byte long.
948 nigel 91 .P
949 nigel 93 When a line ending is defined as a single character, dot never matches that
950     character; when the two-character sequence CRLF is used, dot does not match CR
951     if it is immediately followed by LF, but otherwise it matches all characters
952     (including isolated CRs and LFs). When any Unicode line endings are being
953     recognized, dot does not match CR or LF or any of the other line ending
954     characters.
955     .P
956 nigel 91 The behaviour of dot with regard to newlines can be changed. If the PCRE_DOTALL
957 nigel 93 option is set, a dot matches any one character, without exception. If the
958     two-character sequence CRLF is present in the subject string, it takes two dots
959     to match it.
960 nigel 91 .P
961     The handling of dot is entirely independent of the handling of circumflex and
962     dollar, the only relationship being that they both involve newlines. Dot has no
963     special meaning in a character class.
964 ph10 514 .P
965 ph10 579 The escape sequence \eN behaves like a dot, except that it is not affected by
966 ph10 572 the PCRE_DOTALL option. In other words, it matches any character except one
967     that signifies the end of a line.
968 nigel 75 .
969     .
971 nigel 63 .rs
972     .sp
973 nigel 75 Outside a character class, the escape sequence \eC matches any one byte, both
974 ph10 737 in and out of UTF-8 mode. Unlike a dot, it always matches line-ending
975 nigel 93 characters. The feature is provided in Perl in order to match individual bytes
976 ph10 737 in UTF-8 mode, but it is unclear how it can usefully be used. Because \eC
977     breaks up characters into individual bytes, matching one byte with \eC in UTF-8
978     mode means that the rest of the string may start with a malformed UTF-8
979     character. This has undefined results, because PCRE assumes that it is dealing
980     with valid UTF-8 strings (and by default it checks this at the start of
981     processing unless the PCRE_NO_UTF8_CHECK option is used).
982 nigel 75 .P
983     PCRE does not allow \eC to appear in lookbehind assertions
984     .\" HTML <a href="#lookbehind">
985     .\" </a>
986     (described below),
987     .\"
988     because in UTF-8 mode this would make it impossible to calculate the length of
989     the lookbehind.
990 ph10 737 .P
991     In general, the \eC escape sequence is best avoided in UTF-8 mode. However, one
992     way of using it that avoids the problem of malformed UTF-8 characters is to
993     use a lookahead to check the length of the next character, as in this pattern
994     (ignore white space and line breaks):
995     .sp
996     (?| (?=[\ex00-\ex7f])(\eC) |
997     (?=[\ex80-\ex{7ff}])(\eC)(\eC) |
998     (?=[\ex{800}-\ex{ffff}])(\eC)(\eC)(\eC) |
999     (?=[\ex{10000}-\ex{1fffff}])(\eC)(\eC)(\eC)(\eC))
1000     .sp
1001     A group that starts with (?| resets the capturing parentheses numbers in each
1002     alternative (see
1003     .\" HTML <a href="#dupsubpatternnumber">
1004     .\" </a>
1005     "Duplicate Subpattern Numbers"
1006     .\"
1007     below). The assertions at the start of each branch check the next UTF-8
1008     character for values whose encoding uses 1, 2, 3, or 4 bytes, respectively. The
1009     character's individual bytes are then captured by the appropriate number of
1010     groups.
1011 nigel 75 .
1012     .
1013     .\" HTML <a name="characterclass"></a>
1015 nigel 63 .rs
1016     .sp
1017     An opening square bracket introduces a character class, terminated by a closing
1018 ph10 461 square bracket. A closing square bracket on its own is not special by default.
1019     However, if the PCRE_JAVASCRIPT_COMPAT option is set, a lone closing square
1020 ph10 456 bracket causes a compile-time error. If a closing square bracket is required as
1021     a member of the class, it should be the first data character in the class
1022     (after an initial circumflex, if present) or escaped with a backslash.
1023 nigel 75 .P
1024 nigel 63 A character class matches a single character in the subject. In UTF-8 mode, the
1025 ph10 456 character may be more than one byte long. A matched character must be in the
1026     set of characters defined by the class, unless the first character in the class
1027 nigel 63 definition is a circumflex, in which case the subject character must not be in
1028     the set defined by the class. If a circumflex is actually required as a member
1029     of the class, ensure it is not the first character, or escape it with a
1030     backslash.
1031 nigel 75 .P
1032 nigel 63 For example, the character class [aeiou] matches any lower case vowel, while
1033     [^aeiou] matches any character that is not a lower case vowel. Note that a
1034 nigel 75 circumflex is just a convenient notation for specifying the characters that
1035     are in the class by enumerating those that are not. A class that starts with a
1036 ph10 456 circumflex is not an assertion; it still consumes a character from the subject
1037 nigel 75 string, and therefore it fails if the current pointer is at the end of the
1038     string.
1039     .P
1040 nigel 63 In UTF-8 mode, characters with values greater than 255 can be included in a
1041 nigel 75 class as a literal string of bytes, or by using the \ex{ escaping mechanism.
1042     .P
1043 nigel 63 When caseless matching is set, any letters in a class represent both their
1044     upper case and lower case versions, so for example, a caseless [aeiou] matches
1045     "A" as well as "a", and a caseless [^aeiou] does not match "A", whereas a
1046 nigel 77 caseful version would. In UTF-8 mode, PCRE always understands the concept of
1047     case for characters whose values are less than 128, so caseless matching is
1048     always possible. For characters with higher values, the concept of case is
1049     supported if PCRE is compiled with Unicode property support, but not otherwise.
1050 ph10 456 If you want to use caseless matching in UTF8-mode for characters 128 and above,
1051     you must ensure that PCRE is compiled with Unicode property support as well as
1052     with UTF-8 support.
1053 nigel 75 .P
1054 nigel 93 Characters that might indicate line breaks are never treated in any special way
1055     when matching character classes, whatever line-ending sequence is in use, and
1056     whatever setting of the PCRE_DOTALL and PCRE_MULTILINE options is used. A class
1057     such as [^a] always matches one of these characters.
1058 nigel 75 .P
1059 nigel 63 The minus (hyphen) character can be used to specify a range of characters in a
1060     character class. For example, [d-m] matches any letter between d and m,
1061     inclusive. If a minus character is required in a class, it must be escaped with
1062     a backslash or appear in a position where it cannot be interpreted as
1063     indicating a range, typically as the first or last character in the class.
1064 nigel 75 .P
1065 nigel 63 It is not possible to have the literal character "]" as the end character of a
1066     range. A pattern such as [W-]46] is interpreted as a class of two characters
1067     ("W" and "-") followed by a literal string "46]", so it would match "W46]" or
1068     "-46]". However, if the "]" is escaped with a backslash it is interpreted as
1069 nigel 75 the end of range, so [W-\e]46] is interpreted as a class containing a range
1070     followed by two other characters. The octal or hexadecimal representation of
1071     "]" can also be used to end a range.
1072     .P
1073 nigel 63 Ranges operate in the collating sequence of character values. They can also be
1074 nigel 75 used for characters specified numerically, for example [\e000-\e037]. In UTF-8
1075 nigel 63 mode, ranges can include characters whose values are greater than 255, for
1076 nigel 75 example [\ex{100}-\ex{2ff}].
1077     .P
1078 nigel 63 If a range that includes letters is used when caseless matching is set, it
1079     matches the letters in either case. For example, [W-c] is equivalent to
1080 nigel 75 [][\e\e^_`wxyzabc], matched caselessly, and in non-UTF-8 mode, if character
1081 ph10 139 tables for a French locale are in use, [\exc8-\excb] matches accented E
1082 nigel 75 characters in both cases. In UTF-8 mode, PCRE supports the concept of case for
1083     characters with values greater than 128 only when it is compiled with Unicode
1084     property support.
1085     .P
1086 ph10 575 The character escape sequences \ed, \eD, \eh, \eH, \ep, \eP, \es, \eS, \ev,
1087     \eV, \ew, and \eW may appear in a character class, and add the characters that
1088     they match to the class. For example, [\edABCDEF] matches any hexadecimal
1089     digit. In UTF-8 mode, the PCRE_UCP option affects the meanings of \ed, \es, \ew
1090     and their upper case partners, just as it does when they appear outside a
1091     character class, as described in the section entitled
1092     .\" HTML <a href="#genericchartypes">
1093     .\" </a>
1094     "Generic character types"
1095     .\"
1096     above. The escape sequence \eb has a different meaning inside a character
1097     class; it matches the backspace character. The sequences \eB, \eN, \eR, and \eX
1098     are not special inside a character class. Like any other unrecognized escape
1099     sequences, they are treated as the literal characters "B", "N", "R", and "X" by
1100     default, but cause an error if the PCRE_EXTRA option is set.
1101     .P
1102     A circumflex can conveniently be used with the upper case character types to
1103 ph10 518 specify a more restricted set of characters than the matching lower case type.
1104 ph10 575 For example, the class [^\eW_] matches any letter or digit, but not underscore,
1105 ph10 579 whereas [\ew] includes underscore. A positive character class should be read as
1106     "something OR something OR ..." and a negative class as "NOT something AND NOT
1107 ph10 575 something AND NOT ...".
1108 nigel 75 .P
1109     The only metacharacters that are recognized in character classes are backslash,
1110     hyphen (only where it can be interpreted as specifying a range), circumflex
1111     (only at the start), opening square bracket (only when it can be interpreted as
1112     introducing a POSIX class name - see the next section), and the terminating
1113     closing square bracket. However, escaping other non-alphanumeric characters
1114     does no harm.
1115     .
1116     .
1118 nigel 63 .rs
1119     .sp
1120 nigel 75 Perl supports the POSIX notation for character classes. This uses names
1121 nigel 63 enclosed by [: and :] within the enclosing square brackets. PCRE also supports
1122     this notation. For example,
1123 nigel 75 .sp
1124 nigel 63 [01[:alpha:]%]
1125 nigel 75 .sp
1126 nigel 63 matches "0", "1", any alphabetic character, or "%". The supported class names
1127 ph10 518 are:
1128 nigel 75 .sp
1129 nigel 63 alnum letters and digits
1130     alpha letters
1131     ascii character codes 0 - 127
1132     blank space or tab only
1133     cntrl control characters
1134 nigel 75 digit decimal digits (same as \ed)
1135 nigel 63 graph printing characters, excluding space
1136     lower lower case letters
1137     print printing characters, including space
1138 ph10 518 punct printing characters, excluding letters and digits and space
1139 nigel 75 space white space (not quite the same as \es)
1140 nigel 63 upper upper case letters
1141 nigel 75 word "word" characters (same as \ew)
1142 nigel 63 xdigit hexadecimal digits
1143 nigel 75 .sp
1144 nigel 63 The "space" characters are HT (9), LF (10), VT (11), FF (12), CR (13), and
1145     space (32). Notice that this list includes the VT character (code 11). This
1146 nigel 75 makes "space" different to \es, which does not include VT (for Perl
1147 nigel 63 compatibility).
1148 nigel 75 .P
1149 nigel 63 The name "word" is a Perl extension, and "blank" is a GNU extension from Perl
1150     5.8. Another Perl extension is negation, which is indicated by a ^ character
1151     after the colon. For example,
1152 nigel 75 .sp
1153 nigel 63 [12[:^digit:]]
1154 nigel 75 .sp
1155 nigel 63 matches "1", "2", or any non-digit. PCRE (and Perl) also recognize the POSIX
1156     syntax [.ch.] and [=ch=] where "ch" is a "collating element", but these are not
1157     supported, and an error is given if they are encountered.
1158 nigel 75 .P
1159 ph10 518 By default, in UTF-8 mode, characters with values greater than 128 do not match
1160     any of the POSIX character classes. However, if the PCRE_UCP option is passed
1161 ph10 535 to \fBpcre_compile()\fP, some of the classes are changed so that Unicode
1162     character properties are used. This is achieved by replacing the POSIX classes
1163 ph10 518 by other sequences, as follows:
1164     .sp
1165     [:alnum:] becomes \ep{Xan}
1166     [:alpha:] becomes \ep{L}
1167 ph10 535 [:blank:] becomes \eh
1168 ph10 518 [:digit:] becomes \ep{Nd}
1169     [:lower:] becomes \ep{Ll}
1170 ph10 535 [:space:] becomes \ep{Xps}
1171 ph10 518 [:upper:] becomes \ep{Lu}
1172     [:word:] becomes \ep{Xwd}
1173     .sp
1174     Negated versions, such as [:^alpha:] use \eP instead of \ep. The other POSIX
1175     classes are unchanged, and match only characters with code points less than
1176     128.
1177 nigel 75 .
1178     .
1179     .SH "VERTICAL BAR"
1180 nigel 63 .rs
1181     .sp
1182     Vertical bar characters are used to separate alternative patterns. For example,
1183     the pattern
1184 nigel 75 .sp
1185 nigel 63 gilbert|sullivan
1186 nigel 75 .sp
1187 nigel 63 matches either "gilbert" or "sullivan". Any number of alternatives may appear,
1188 nigel 91 and an empty alternative is permitted (matching the empty string). The matching
1189     process tries each alternative in turn, from left to right, and the first one
1190     that succeeds is used. If the alternatives are within a subpattern
1191 nigel 75 .\" HTML <a href="#subpattern">
1192     .\" </a>
1193     (defined below),
1194     .\"
1195     "succeeds" means matching the rest of the main pattern as well as the
1196     alternative in the subpattern.
1197     .
1198     .
1200 nigel 63 .rs
1201     .sp
1202     The settings of the PCRE_CASELESS, PCRE_MULTILINE, PCRE_DOTALL, and
1203 ph10 231 PCRE_EXTENDED options (which are Perl-compatible) can be changed from within
1204     the pattern by a sequence of Perl option letters enclosed between "(?" and ")".
1205     The option letters are
1206 nigel 75 .sp
1207 nigel 63 i for PCRE_CASELESS
1208     m for PCRE_MULTILINE
1209     s for PCRE_DOTALL
1210     x for PCRE_EXTENDED
1211 nigel 75 .sp
1212 nigel 63 For example, (?im) sets caseless, multiline matching. It is also possible to
1213     unset these options by preceding the letter with a hyphen, and a combined
1214     setting and unsetting such as (?im-sx), which sets PCRE_CASELESS and
1215     PCRE_MULTILINE while unsetting PCRE_DOTALL and PCRE_EXTENDED, is also
1216     permitted. If a letter appears both before and after the hyphen, the option is
1217     unset.
1218 nigel 75 .P
1219 ph10 231 The PCRE-specific options PCRE_DUPNAMES, PCRE_UNGREEDY, and PCRE_EXTRA can be
1220     changed in the same way as the Perl-compatible options by using the characters
1221     J, U and X respectively.
1222     .P
1223 ph10 412 When one of these option changes occurs at top level (that is, not inside
1224     subpattern parentheses), the change applies to the remainder of the pattern
1225     that follows. If the change is placed right at the start of a pattern, PCRE
1226     extracts it into the global options (and it will therefore show up in data
1227     extracted by the \fBpcre_fullinfo()\fP function).
1228 nigel 75 .P
1229 nigel 93 An option change within a subpattern (see below for a description of
1230 ph10 572 subpatterns) affects only that part of the subpattern that follows it, so
1231 nigel 75 .sp
1232 nigel 63 (a(?i)b)c
1233 nigel 75 .sp
1234 nigel 63 matches abc and aBc and no other strings (assuming PCRE_CASELESS is not used).
1235     By this means, options can be made to have different settings in different
1236     parts of the pattern. Any changes made in one alternative do carry on
1237     into subsequent branches within the same subpattern. For example,
1238 nigel 75 .sp
1239 nigel 63 (a(?i)b|c)
1240 nigel 75 .sp
1241 nigel 63 matches "ab", "aB", "c", and "C", even though when matching "C" the first
1242     branch is abandoned before the option setting. This is because the effects of
1243     option settings happen at compile time. There would be some very weird
1244     behaviour otherwise.
1245 ph10 251 .P
1246     \fBNote:\fP There are other PCRE-specific options that can be set by the
1247     application when the compile or match functions are called. In some cases the
1248 ph10 412 pattern can contain special leading sequences such as (*CRLF) to override what
1249     the application has set or what has been defaulted. Details are given in the
1250     section entitled
1251 ph10 251 .\" HTML <a href="#newlineseq">
1252     .\" </a>
1253     "Newline sequences"
1254     .\"
1255 ph10 518 above. There are also the (*UTF8) and (*UCP) leading sequences that can be used
1256     to set UTF-8 and Unicode property modes; they are equivalent to setting the
1257     PCRE_UTF8 and the PCRE_UCP options, respectively.
1258 nigel 75 .
1259     .
1260     .\" HTML <a name="subpattern"></a>
1261 nigel 63 .SH SUBPATTERNS
1262     .rs
1263     .sp
1264     Subpatterns are delimited by parentheses (round brackets), which can be nested.
1265 nigel 75 Turning part of a pattern into a subpattern does two things:
1266     .sp
1267 nigel 63 1. It localizes a set of alternatives. For example, the pattern
1268 nigel 75 .sp
1269 nigel 63 cat(aract|erpillar|)
1270 nigel 75 .sp
1271 ph10 572 matches "cataract", "caterpillar", or "cat". Without the parentheses, it would
1272     match "cataract", "erpillar" or an empty string.
1273 nigel 75 .sp
1274     2. It sets up the subpattern as a capturing subpattern. This means that, when
1275     the whole pattern matches, that portion of the subject string that matched the
1276     subpattern is passed back to the caller via the \fIovector\fP argument of
1277     \fBpcre_exec()\fP. Opening parentheses are counted from left to right (starting
1278 ph10 572 from 1) to obtain numbers for the capturing subpatterns. For example, if the
1279     string "the red king" is matched against the pattern
1280 nigel 75 .sp
1281 nigel 63 the ((red|white) (king|queen))
1282 nigel 75 .sp
1283 nigel 63 the captured substrings are "red king", "red", and "king", and are numbered 1,
1284     2, and 3, respectively.
1285 nigel 75 .P
1286 nigel 63 The fact that plain parentheses fulfil two functions is not always helpful.
1287     There are often times when a grouping subpattern is required without a
1288     capturing requirement. If an opening parenthesis is followed by a question mark
1289     and a colon, the subpattern does not do any capturing, and is not counted when
1290     computing the number of any subsequent capturing subpatterns. For example, if
1291     the string "the white queen" is matched against the pattern
1292 nigel 75 .sp
1293 nigel 63 the ((?:red|white) (king|queen))
1294 nigel 75 .sp
1295 nigel 63 the captured substrings are "white queen" and "queen", and are numbered 1 and
1296 nigel 93 2. The maximum number of capturing subpatterns is 65535.
1297 nigel 75 .P
1298 nigel 63 As a convenient shorthand, if any option settings are required at the start of
1299     a non-capturing subpattern, the option letters may appear between the "?" and
1300     the ":". Thus the two patterns
1301 nigel 75 .sp
1302 nigel 63 (?i:saturday|sunday)
1303     (?:(?i)saturday|sunday)
1304 nigel 75 .sp
1305 nigel 63 match exactly the same set of strings. Because alternative branches are tried
1306     from left to right, and options are not reset until the end of the subpattern
1307     is reached, an option setting in one branch does affect subsequent branches, so
1308     the above patterns match "SUNDAY" as well as "Saturday".
1309 nigel 75 .
1310     .
1311 ph10 456 .\" HTML <a name="dupsubpatternnumber"></a>
1313     .rs
1314     .sp
1315 ph10 182 Perl 5.10 introduced a feature whereby each alternative in a subpattern uses
1316     the same numbers for its capturing parentheses. Such a subpattern starts with
1317     (?| and is itself a non-capturing subpattern. For example, consider this
1318 ph10 175 pattern:
1319     .sp
1320     (?|(Sat)ur|(Sun))day
1321 ph10 182 .sp
1322     Because the two alternatives are inside a (?| group, both sets of capturing
1323     parentheses are numbered one. Thus, when the pattern matches, you can look
1324     at captured substring number one, whichever alternative matched. This construct
1325     is useful when you want to capture part, but not all, of one of a number of
1326     alternatives. Inside a (?| group, parentheses are numbered as usual, but the
1327 ph10 175 number is reset at the start of each branch. The numbers of any capturing
1328 ph10 572 parentheses that follow the subpattern start after the highest number used in
1329     any branch. The following example is taken from the Perl documentation. The
1330     numbers underneath show in which buffer the captured content will be stored.
1331 ph10 175 .sp
1332     # before ---------------branch-reset----------- after
1333     / ( a ) (?| x ( y ) z | (p (q) r) | (t) u (v) ) ( z ) /x
1334     # 1 2 2 3 2 3 4
1335 ph10 182 .sp
1336 ph10 488 A back reference to a numbered subpattern uses the most recent value that is
1337     set for that number by any subpattern. The following pattern matches "abcabc"
1338     or "defdef":
1339 ph10 456 .sp
1340 ph10 461 /(?|(abc)|(def))\e1/
1341 ph10 456 .sp
1342 ph10 716 In contrast, a subroutine call to a numbered subpattern always refers to the
1343     first one in the pattern with the given number. The following pattern matches
1344     "abcabc" or "defabc":
1345 ph10 456 .sp
1346     /(?|(abc)|(def))(?1)/
1347     .sp
1348 ph10 459 If a
1349     .\" HTML <a href="#conditions">
1350     .\" </a>
1351     condition test
1352     .\"
1353     for a subpattern's having matched refers to a non-unique number, the test is
1354     true if any of the subpatterns of that number have matched.
1355     .P
1356     An alternative approach to using this "branch reset" feature is to use
1357 ph10 175 duplicate named subpatterns, as described in the next section.
1358     .
1359     .
1360 nigel 75 .SH "NAMED SUBPATTERNS"
1361 nigel 63 .rs
1362     .sp
1363     Identifying capturing parentheses by number is simple, but it can be very hard
1364     to keep track of the numbers in complicated regular expressions. Furthermore,
1365 nigel 75 if an expression is modified, the numbers may change. To help with this
1366 nigel 93 difficulty, PCRE supports the naming of subpatterns. This feature was not
1367     added to Perl until release 5.10. Python had the feature earlier, and PCRE
1368     introduced it at release 4.0, using the Python syntax. PCRE now supports both
1369 ph10 459 the Perl and the Python syntax. Perl allows identically numbered subpatterns to
1370     have different names, but PCRE does not.
1371 nigel 93 .P
1372     In PCRE, a subpattern can be named in one of three ways: (?<name>...) or
1373     (?'name'...) as in Perl, or (?P<name>...) as in Python. References to capturing
1374 nigel 91 parentheses from other parts of the pattern, such as
1375     .\" HTML <a href="#backreferences">
1376     .\" </a>
1377 ph10 488 back references,
1378 nigel 91 .\"
1379     .\" HTML <a href="#recursion">
1380     .\" </a>
1381     recursion,
1382     .\"
1383     and
1384     .\" HTML <a href="#conditions">
1385     .\" </a>
1386     conditions,
1387     .\"
1388     can be made by name as well as by number.
1389 nigel 75 .P
1390 nigel 91 Names consist of up to 32 alphanumeric characters and underscores. Named
1391 nigel 93 capturing parentheses are still allocated numbers as well as names, exactly as
1392     if the names were not present. The PCRE API provides function calls for
1393     extracting the name-to-number translation table from a compiled pattern. There
1394     is also a convenience function for extracting a captured substring by name.
1395 nigel 91 .P
1396     By default, a name must be unique within a pattern, but it is possible to relax
1397 ph10 457 this constraint by setting the PCRE_DUPNAMES option at compile time. (Duplicate
1398 ph10 461 names are also always permitted for subpatterns with the same number, set up as
1399 ph10 457 described in the previous section.) Duplicate names can be useful for patterns
1400     where only one instance of the named parentheses can match. Suppose you want to
1401     match the name of a weekday, either as a 3-letter abbreviation or as the full
1402     name, and in both cases you want to extract the abbreviation. This pattern
1403     (ignoring the line breaks) does the job:
1404 nigel 91 .sp
1405 nigel 93 (?<DN>Mon|Fri|Sun)(?:day)?|
1406     (?<DN>Tue)(?:sday)?|
1407     (?<DN>Wed)(?:nesday)?|
1408     (?<DN>Thu)(?:rsday)?|
1409     (?<DN>Sat)(?:urday)?
1410 nigel 91 .sp
1411     There are five capturing substrings, but only one is ever set after a match.
1412 ph10 182 (An alternative way of solving this problem is to use a "branch reset"
1413 ph10 175 subpattern, as described in the previous section.)
1414     .P
1415 nigel 91 The convenience function for extracting the data by name returns the substring
1416 nigel 93 for the first (and in this example, the only) subpattern of that name that
1417 ph10 461 matched. This saves searching to find which numbered subpattern it was.
1418 ph10 459 .P
1419 ph10 488 If you make a back reference to a non-unique named subpattern from elsewhere in
1420 ph10 459 the pattern, the one that corresponds to the first occurrence of the name is
1421     used. In the absence of duplicate numbers (see the previous section) this is
1422     the one with the lowest number. If you use a named reference in a condition
1423     test (see the
1424     .\"
1425     .\" HTML <a href="#conditions">
1426     .\" </a>
1427     section about conditions
1428     .\"
1429 ph10 461 below), either to check whether a subpattern has matched, or to check for
1430 ph10 459 recursion, all subpatterns with the same name are tested. If the condition is
1431     true for any one of them, the overall condition is true. This is the same
1432     behaviour as testing by number. For further details of the interfaces for
1433     handling named subpatterns, see the
1434 nigel 63 .\" HREF
1435 nigel 75 \fBpcreapi\fP
1436 nigel 63 .\"
1437     documentation.
1438 ph10 385 .P
1439     \fBWarning:\fP You cannot use different names to distinguish between two
1440 ph10 457 subpatterns with the same number because PCRE uses only the numbers when
1441     matching. For this reason, an error is given at compile time if different names
1442     are given to subpatterns with the same number. However, you can give the same
1443     name to subpatterns with the same number, even when PCRE_DUPNAMES is not set.
1444 nigel 75 .
1445     .
1446 nigel 63 .SH REPETITION
1447     .rs
1448     .sp
1449     Repetition is specified by quantifiers, which can follow any of the following
1450     items:
1451 nigel 75 .sp
1452 nigel 63 a literal data character
1453 nigel 93 the dot metacharacter
1454 nigel 75 the \eC escape sequence
1455     the \eX escape sequence (in UTF-8 mode with Unicode properties)
1456 nigel 93 the \eR escape sequence
1457 ph10 572 an escape such as \ed or \epL that matches a single character
1458 nigel 63 a character class
1459     a back reference (see next section)
1460 ph10 637 a parenthesized subpattern (including assertions)
1461 ph10 716 a subroutine call to a subpattern (recursive or otherwise)
1462 nigel 75 .sp
1463 nigel 63 The general repetition quantifier specifies a minimum and maximum number of
1464     permitted matches, by giving the two numbers in curly brackets (braces),
1465     separated by a comma. The numbers must be less than 65536, and the first must
1466     be less than or equal to the second. For example:
1467 nigel 75 .sp
1468 nigel 63 z{2,4}
1469 nigel 75 .sp
1470 nigel 63 matches "zz", "zzz", or "zzzz". A closing brace on its own is not a special
1471     character. If the second number is omitted, but the comma is present, there is
1472     no upper limit; if the second number and the comma are both omitted, the
1473     quantifier specifies an exact number of required matches. Thus
1474 nigel 75 .sp
1475 nigel 63 [aeiou]{3,}
1476 nigel 75 .sp
1477 nigel 63 matches at least 3 successive vowels, but may match many more, while
1478 nigel 75 .sp
1479     \ed{8}
1480     .sp
1481 nigel 63 matches exactly 8 digits. An opening curly bracket that appears in a position
1482     where a quantifier is not allowed, or one that does not match the syntax of a
1483     quantifier, is taken as a literal character. For example, {,6} is not a
1484     quantifier, but a literal string of four characters.
1485 nigel 75 .P
1486 nigel 63 In UTF-8 mode, quantifiers apply to UTF-8 characters rather than to individual
1487 nigel 75 bytes. Thus, for example, \ex{100}{2} matches two UTF-8 characters, each of
1488     which is represented by a two-byte sequence. Similarly, when Unicode property
1489     support is available, \eX{3} matches three Unicode extended sequences, each of
1490     which may be several bytes long (and they may be of different lengths).
1491     .P
1492 nigel 63 The quantifier {0} is permitted, causing the expression to behave as if the
1493 ph10 345 previous item and the quantifier were not present. This may be useful for
1494     subpatterns that are referenced as
1495 ph10 335 .\" HTML <a href="#subpatternsassubroutines">
1496     .\" </a>
1497     subroutines
1498     .\"
1499 ph10 572 from elsewhere in the pattern (but see also the section entitled
1500     .\" HTML <a href="#subdefine">
1501     .\" </a>
1502     "Defining subpatterns for use by reference only"
1503     .\"
1504     below). Items other than subpatterns that have a {0} quantifier are omitted
1505     from the compiled pattern.
1506 nigel 75 .P
1507 nigel 93 For convenience, the three most common quantifiers have single-character
1508     abbreviations:
1509 nigel 75 .sp
1510 nigel 63 * is equivalent to {0,}
1511     + is equivalent to {1,}
1512     ? is equivalent to {0,1}
1513 nigel 75 .sp
1514 nigel 63 It is possible to construct infinite loops by following a subpattern that can
1515     match no characters with a quantifier that has no upper limit, for example:
1516 nigel 75 .sp
1517 nigel 63 (a?)*
1518 nigel 75 .sp
1519 nigel 63 Earlier versions of Perl and PCRE used to give an error at compile time for
1520     such patterns. However, because there are cases where this can be useful, such
1521     patterns are now accepted, but if any repetition of the subpattern does in fact
1522     match no characters, the loop is forcibly broken.
1523 nigel 75 .P
1524 nigel 63 By default, the quantifiers are "greedy", that is, they match as much as
1525     possible (up to the maximum number of permitted times), without causing the
1526     rest of the pattern to fail. The classic example of where this gives problems
1527 nigel 75 is in trying to match comments in C programs. These appear between /* and */
1528     and within the comment, individual * and / characters may appear. An attempt to
1529     match C comments by applying the pattern
1530     .sp
1531     /\e*.*\e*/
1532     .sp
1533 nigel 63 to the string
1534 nigel 75 .sp
1535     /* first comment */ not comment /* second comment */
1536     .sp
1537 nigel 63 fails, because it matches the entire string owing to the greediness of the .*
1538     item.
1539 nigel 75 .P
1540 nigel 63 However, if a quantifier is followed by a question mark, it ceases to be
1541     greedy, and instead matches the minimum number of times possible, so the
1542     pattern
1543 nigel 75 .sp
1544     /\e*.*?\e*/
1545     .sp
1546 nigel 63 does the right thing with the C comments. The meaning of the various
1547     quantifiers is not otherwise changed, just the preferred number of matches.
1548     Do not confuse this use of question mark with its use as a quantifier in its
1549     own right. Because it has two uses, it can sometimes appear doubled, as in
1550 nigel 75 .sp
1551     \ed??\ed
1552     .sp
1553 nigel 63 which matches one digit by preference, but can match two if that is the only
1554     way the rest of the pattern matches.
1555 nigel 75 .P
1556 nigel 93 If the PCRE_UNGREEDY option is set (an option that is not available in Perl),
1557 nigel 63 the quantifiers are not greedy by default, but individual ones can be made
1558     greedy by following them with a question mark. In other words, it inverts the
1559     default behaviour.
1560 nigel 75 .P
1561 nigel 63 When a parenthesized subpattern is quantified with a minimum repeat count that
1562 nigel 75 is greater than 1 or with a limited maximum, more memory is required for the
1563 nigel 63 compiled pattern, in proportion to the size of the minimum or maximum.
1564 nigel 75 .P
1565 nigel 63 If a pattern starts with .* or .{0,} and the PCRE_DOTALL option (equivalent
1566 nigel 93 to Perl's /s) is set, thus allowing the dot to match newlines, the pattern is
1567 nigel 63 implicitly anchored, because whatever follows will be tried against every
1568     character position in the subject string, so there is no point in retrying the
1569     overall match at any position after the first. PCRE normally treats such a
1570 nigel 75 pattern as though it were preceded by \eA.
1571     .P
1572 nigel 63 In cases where it is known that the subject string contains no newlines, it is
1573     worth setting PCRE_DOTALL in order to obtain this optimization, or
1574     alternatively using ^ to indicate anchoring explicitly.
1575 nigel 75 .P
1576 nigel 63 However, there is one situation where the optimization cannot be used. When .*
1577 ph10 488 is inside capturing parentheses that are the subject of a back reference
1578 nigel 93 elsewhere in the pattern, a match at the start may fail where a later one
1579     succeeds. Consider, for example:
1580 nigel 75 .sp
1581     (.*)abc\e1
1582     .sp
1583 nigel 63 If the subject is "xyz123abc123" the match point is the fourth character. For
1584     this reason, such a pattern is not implicitly anchored.
1585 nigel 75 .P
1586 nigel 63 When a capturing subpattern is repeated, the value captured is the substring
1587     that matched the final iteration. For example, after
1588 nigel 75 .sp
1589     (tweedle[dume]{3}\es*)+
1590     .sp
1591 nigel 63 has matched "tweedledum tweedledee" the value of the captured substring is
1592     "tweedledee". However, if there are nested capturing subpatterns, the
1593     corresponding captured values may have been set in previous iterations. For
1594     example, after
1595 nigel 75 .sp
1596 nigel 63 /(a|(b))+/
1597 nigel 75 .sp
1598 nigel 63 matches "aba" the value of the second captured substring is "b".
1599 nigel 75 .
1600     .
1601     .\" HTML <a name="atomicgroup"></a>
1603 nigel 63 .rs
1604     .sp
1605 nigel 93 With both maximizing ("greedy") and minimizing ("ungreedy" or "lazy")
1606     repetition, failure of what follows normally causes the repeated item to be
1607     re-evaluated to see if a different number of repeats allows the rest of the
1608     pattern to match. Sometimes it is useful to prevent this, either to change the
1609     nature of the match, or to cause it fail earlier than it otherwise might, when
1610     the author of the pattern knows there is no point in carrying on.
1611 nigel 75 .P
1612     Consider, for example, the pattern \ed+foo when applied to the subject line
1613     .sp
1614 nigel 63 123456bar
1615 nigel 75 .sp
1616 nigel 63 After matching all 6 digits and then failing to match "foo", the normal
1617 nigel 75 action of the matcher is to try again with only 5 digits matching the \ed+
1618 nigel 63 item, and then with 4, and so on, before ultimately failing. "Atomic grouping"
1619     (a term taken from Jeffrey Friedl's book) provides the means for specifying
1620     that once a subpattern has matched, it is not to be re-evaluated in this way.
1621 nigel 75 .P
1622 nigel 93 If we use atomic grouping for the previous example, the matcher gives up
1623 nigel 63 immediately on failing to match "foo" the first time. The notation is a kind of
1624     special parenthesis, starting with (?> as in this example:
1625 nigel 75 .sp
1626     (?>\ed+)foo
1627     .sp
1628 nigel 63 This kind of parenthesis "locks up" the part of the pattern it contains once
1629     it has matched, and a failure further into the pattern is prevented from
1630     backtracking into it. Backtracking past it to previous items, however, works as
1631     normal.
1632 nigel 75 .P
1633 nigel 63 An alternative description is that a subpattern of this type matches the string
1634     of characters that an identical standalone pattern would match, if anchored at
1635     the current point in the subject string.
1636 nigel 75 .P
1637 nigel 63 Atomic grouping subpatterns are not capturing subpatterns. Simple cases such as
1638     the above example can be thought of as a maximizing repeat that must swallow
1639 nigel 75 everything it can. So, while both \ed+ and \ed+? are prepared to adjust the
1640 nigel 63 number of digits they match in order to make the rest of the pattern match,
1641 nigel 75 (?>\ed+) can only match an entire sequence of digits.
1642     .P
1643 nigel 63 Atomic groups in general can of course contain arbitrarily complicated
1644     subpatterns, and can be nested. However, when the subpattern for an atomic
1645     group is just a single repeated item, as in the example above, a simpler
1646     notation, called a "possessive quantifier" can be used. This consists of an
1647     additional + character following a quantifier. Using this notation, the
1648     previous example can be rewritten as
1649 nigel 75 .sp
1650     \ed++foo
1651     .sp
1652 ph10 208 Note that a possessive quantifier can be used with an entire group, for
1653     example:
1654     .sp
1655     (abc|xyz){2,3}+
1656     .sp
1657 nigel 63 Possessive quantifiers are always greedy; the setting of the PCRE_UNGREEDY
1658     option is ignored. They are a convenient notation for the simpler forms of
1659 nigel 93 atomic group. However, there is no difference in the meaning of a possessive
1660     quantifier and the equivalent atomic group, though there may be a performance
1661     difference; possessive quantifiers should be slightly faster.
1662 nigel 75 .P
1663 nigel 93 The possessive quantifier syntax is an extension to the Perl 5.8 syntax.
1664     Jeffrey Friedl originated the idea (and the name) in the first edition of his
1665     book. Mike McCloskey liked it, so implemented it when he built Sun's Java
1666     package, and PCRE copied it from there. It ultimately found its way into Perl
1667     at release 5.10.
1668 nigel 75 .P
1669 nigel 93 PCRE has an optimization that automatically "possessifies" certain simple
1670     pattern constructs. For example, the sequence A+B is treated as A++B because
1671     there is no point in backtracking into a sequence of A's when B must follow.
1672     .P
1673 nigel 63 When a pattern contains an unlimited repeat inside a subpattern that can itself
1674     be repeated an unlimited number of times, the use of an atomic group is the
1675     only way to avoid some failing matches taking a very long time indeed. The
1676     pattern
1677 nigel 75 .sp
1678     (\eD+|<\ed+>)*[!?]
1679     .sp
1680 nigel 63 matches an unlimited number of substrings that either consist of non-digits, or
1681     digits enclosed in <>, followed by either ! or ?. When it matches, it runs
1682     quickly. However, if it is applied to
1683 nigel 75 .sp
1684 nigel 63 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
1685 nigel 75 .sp
1686 nigel 63 it takes a long time before reporting failure. This is because the string can
1687 nigel 75 be divided between the internal \eD+ repeat and the external * repeat in a
1688     large number of ways, and all have to be tried. (The example uses [!?] rather
1689     than a single character at the end, because both PCRE and Perl have an
1690     optimization that allows for fast failure when a single character is used. They
1691     remember the last single character that is required for a match, and fail early
1692     if it is not present in the string.) If the pattern is changed so that it uses
1693     an atomic group, like this:
1694     .sp
1695     ((?>\eD+)|<\ed+>)*[!?]
1696     .sp
1697 nigel 63 sequences of non-digits cannot be broken, and failure happens quickly.
1698 nigel 75 .
1699     .
1700     .\" HTML <a name="backreferences"></a>
1702 nigel 63 .rs
1703     .sp
1704     Outside a character class, a backslash followed by a digit greater than 0 (and
1705     possibly further digits) is a back reference to a capturing subpattern earlier
1706     (that is, to its left) in the pattern, provided there have been that many
1707     previous capturing left parentheses.
1708 nigel 75 .P
1709 nigel 63 However, if the decimal number following the backslash is less than 10, it is
1710     always taken as a back reference, and causes an error only if there are not
1711     that many capturing left parentheses in the entire pattern. In other words, the
1712     parentheses that are referenced need not be to the left of the reference for
1713 nigel 91 numbers less than 10. A "forward back reference" of this type can make sense
1714     when a repetition is involved and the subpattern to the right has participated
1715     in an earlier iteration.
1716     .P
1717 nigel 93 It is not possible to have a numerical "forward back reference" to a subpattern
1718     whose number is 10 or more using this syntax because a sequence such as \e50 is
1719     interpreted as a character defined in octal. See the subsection entitled
1720 nigel 91 "Non-printing characters"
1721 nigel 75 .\" HTML <a href="#digitsafterbackslash">
1722     .\" </a>
1723     above
1724     .\"
1725 nigel 93 for further details of the handling of digits following a backslash. There is
1726     no such problem when named parentheses are used. A back reference to any
1727     subpattern is possible using named parentheses (see below).
1728 nigel 75 .P
1729 nigel 93 Another way of avoiding the ambiguity inherent in the use of digits following a
1730 ph10 572 backslash is to use the \eg escape sequence. This escape must be followed by an
1731     unsigned number or a negative number, optionally enclosed in braces. These
1732     examples are all identical:
1733 nigel 93 .sp
1734     (ring), \e1
1735     (ring), \eg1
1736     (ring), \eg{1}
1737     .sp
1738 ph10 208 An unsigned number specifies an absolute reference without the ambiguity that
1739     is present in the older syntax. It is also useful when literal digits follow
1740     the reference. A negative number is a relative reference. Consider this
1741     example:
1742 nigel 93 .sp
1743     (abc(def)ghi)\eg{-1}
1744     .sp
1745     The sequence \eg{-1} is a reference to the most recently started capturing
1746 ph10 572 subpattern before \eg, that is, is it equivalent to \e2 in this example.
1747     Similarly, \eg{-2} would be equivalent to \e1. The use of relative references
1748     can be helpful in long patterns, and also in patterns that are created by
1749     joining together fragments that contain references within themselves.
1750 nigel 93 .P
1751 nigel 63 A back reference matches whatever actually matched the capturing subpattern in
1752     the current subject string, rather than anything matching the subpattern
1753     itself (see
1754     .\" HTML <a href="#subpatternsassubroutines">
1755     .\" </a>
1756     "Subpatterns as subroutines"
1757     .\"
1758     below for a way of doing that). So the pattern
1759 nigel 75 .sp
1760     (sens|respons)e and \e1ibility
1761     .sp
1762 nigel 63 matches "sense and sensibility" and "response and responsibility", but not
1763     "sense and responsibility". If caseful matching is in force at the time of the
1764     back reference, the case of letters is relevant. For example,
1765 nigel 75 .sp
1766     ((?i)rah)\es+\e1
1767     .sp
1768 nigel 63 matches "rah rah" and "RAH RAH", but not "RAH rah", even though the original
1769     capturing subpattern is matched caselessly.
1770 nigel 75 .P
1771 ph10 171 There are several different ways of writing back references to named
1772     subpatterns. The .NET syntax \ek{name} and the Perl syntax \ek<name> or
1773     \ek'name' are supported, as is the Python syntax (?P=name). Perl 5.10's unified
1774     back reference syntax, in which \eg can be used for both numeric and named
1775     references, is also supported. We could rewrite the above example in any of
1776 nigel 93 the following ways:
1777 nigel 75 .sp
1778 nigel 93 (?<p1>(?i)rah)\es+\ek<p1>
1779 ph10 171 (?'p1'(?i)rah)\es+\ek{p1}
1780 nigel 91 (?P<p1>(?i)rah)\es+(?P=p1)
1781 ph10 171 (?<p1>(?i)rah)\es+\eg{p1}
1782 nigel 75 .sp
1783 nigel 91 A subpattern that is referenced by name may appear in the pattern before or
1784     after the reference.
1785     .P
1786 nigel 63 There may be more than one back reference to the same subpattern. If a
1787     subpattern has not actually been used in a particular match, any back
1788 ph10 456 references to it always fail by default. For example, the pattern
1789 nigel 75 .sp
1790     (a|(bc))\e2
1791     .sp
1792 ph10 461 always fails if it starts to match "a" rather than "bc". However, if the
1793     PCRE_JAVASCRIPT_COMPAT option is set at compile time, a back reference to an
1794 ph10 456 unset value matches an empty string.
1795     .P
1796     Because there may be many capturing parentheses in a pattern, all digits
1797     following a backslash are taken as part of a potential back reference number.
1798     If the pattern continues with a digit character, some delimiter must be used to
1799     terminate the back reference. If the PCRE_EXTENDED option is set, this can be
1800     whitespace. Otherwise, the \eg{ syntax or an empty comment (see
1801 nigel 75 .\" HTML <a href="#comments">
1802     .\" </a>
1803     "Comments"
1804     .\"
1805     below) can be used.
1806 ph10 488 .
1807     .SS "Recursive back references"
1808     .rs
1809     .sp
1810 nigel 63 A back reference that occurs inside the parentheses to which it refers fails
1811 nigel 75 when the subpattern is first used, so, for example, (a\e1) never matches.
1812 nigel 63 However, such references can be useful inside repeated subpatterns. For
1813     example, the pattern
1814 nigel 75 .sp
1815     (a|b\e1)+
1816     .sp
1817 nigel 63 matches any number of "a"s and also "aba", "ababbaa" etc. At each iteration of
1818     the subpattern, the back reference matches the character string corresponding
1819     to the previous iteration. In order for this to work, the pattern must be such
1820     that the first iteration does not need to match the back reference. This can be
1821     done using alternation, as in the example above, or by a quantifier with a
1822     minimum of zero.
1823 ph10 488 .P
1824     Back references of this type cause the group that they reference to be treated
1825     as an
1826     .\" HTML <a href="#atomicgroup">
1827     .\" </a>
1828     atomic group.
1829     .\"
1830     Once the whole group has been matched, a subsequent matching failure cannot
1831     cause backtracking into the middle of the group.
1832 nigel 75 .
1833     .
1834     .\" HTML <a name="bigassertions"></a>
1835 nigel 63 .SH ASSERTIONS
1836     .rs
1837     .sp
1838     An assertion is a test on the characters following or preceding the current
1839     matching point that does not actually consume any characters. The simple
1840 nigel 75 assertions coded as \eb, \eB, \eA, \eG, \eZ, \ez, ^ and $ are described
1841     .\" HTML <a href="#smallassertions">
1842     .\" </a>
1843     above.
1844     .\"
1845     .P
1846 nigel 63 More complicated assertions are coded as subpatterns. There are two kinds:
1847     those that look ahead of the current position in the subject string, and those
1848 nigel 75 that look behind it. An assertion subpattern is matched in the normal way,
1849     except that it does not cause the current matching position to be changed.
1850     .P
1851 ph10 637 Assertion subpatterns are not capturing subpatterns. If such an assertion
1852     contains capturing subpatterns within it, these are counted for the purposes of
1853     numbering the capturing subpatterns in the whole pattern. However, substring
1854     capturing is carried out only for positive assertions, because it does not make
1855     sense for negative assertions.
1856     .P
1857 ph10 643 For compatibility with Perl, assertion subpatterns may be repeated; though
1858 ph10 654 it makes no sense to assert the same thing several times, the side effect of
1859 ph10 643 capturing parentheses may occasionally be useful. In practice, there only three
1860     cases:
1861 ph10 637 .sp
1862 ph10 654 (1) If the quantifier is {0}, the assertion is never obeyed during matching.
1863     However, it may contain internal capturing parenthesized groups that are called
1864 ph10 637 from elsewhere via the
1865     .\" HTML <a href="#subpatternsassubroutines">
1866     .\" </a>
1867     subroutine mechanism.
1868     .\"
1869     .sp
1870 ph10 654 (2) If quantifier is {0,n} where n is greater than zero, it is treated as if it
1871     were {0,1}. At run time, the rest of the pattern match is tried with and
1872 ph10 637 without the assertion, the order depending on the greediness of the quantifier.
1873     .sp
1874 ph10 654 (3) If the minimum repetition is greater than zero, the quantifier is ignored.
1875 ph10 637 The assertion is obeyed just once when encountered during matching.
1876 nigel 75 .
1877     .
1878     .SS "Lookahead assertions"
1879     .rs
1880     .sp
1881 nigel 91 Lookahead assertions start with (?= for positive assertions and (?! for
1882     negative assertions. For example,
1883 nigel 75 .sp
1884     \ew+(?=;)
1885     .sp
1886 nigel 63 matches a word followed by a semicolon, but does not include the semicolon in
1887     the match, and
1888 nigel 75 .sp
1889 nigel 63 foo(?!bar)
1890 nigel 75 .sp
1891 nigel 63 matches any occurrence of "foo" that is not followed by "bar". Note that the
1892     apparently similar pattern
1893 nigel 75 .sp
1894 nigel 63 (?!foo)bar
1895 nigel 75 .sp
1896 nigel 63 does not find an occurrence of "bar" that is preceded by something other than
1897     "foo"; it finds any occurrence of "bar" whatsoever, because the assertion
1898     (?!foo) is always true when the next three characters are "bar". A
1899 nigel 75 lookbehind assertion is needed to achieve the other effect.
1900     .P
1901 nigel 63 If you want to force a matching failure at some point in a pattern, the most
1902     convenient way to do it is with (?!) because an empty string always matches, so
1903     an assertion that requires there not to be an empty string must always fail.
1904 ph10 572 The backtracking control verb (*FAIL) or (*F) is a synonym for (?!).
1905 nigel 75 .
1906     .
1907     .\" HTML <a name="lookbehind"></a>
1908     .SS "Lookbehind assertions"
1909     .rs
1910     .sp
1911 nigel 63 Lookbehind assertions start with (?<= for positive assertions and (?<! for
1912     negative assertions. For example,
1913 nigel 75 .sp
1914 nigel 63 (?<!foo)bar
1915 nigel 75 .sp
1916 nigel 63 does find an occurrence of "bar" that is not preceded by "foo". The contents of
1917     a lookbehind assertion are restricted such that all the strings it matches must
1918 nigel 91 have a fixed length. However, if there are several top-level alternatives, they
1919     do not all have to have the same fixed length. Thus
1920 nigel 75 .sp
1921 nigel 63 (?<=bullock|donkey)
1922 nigel 75 .sp
1923 nigel 63 is permitted, but
1924 nigel 75 .sp
1925 nigel 63 (?<!dogs?|cats?)
1926 nigel 75 .sp
1927 nigel 63 causes an error at compile time. Branches that match different length strings
1928     are permitted only at the top level of a lookbehind assertion. This is an
1929 ph10 572 extension compared with Perl, which requires all branches to match the same
1930     length of string. An assertion such as
1931 nigel 75 .sp
1932 nigel 63 (?<=ab(c|de))
1933 nigel 75 .sp
1934 nigel 63 is not permitted, because its single top-level branch can match two different
1935 ph10 454 lengths, but it is acceptable to PCRE if rewritten to use two top-level
1936     branches:
1937 nigel 75 .sp
1938 nigel 63 (?<=abc|abde)
1939 nigel 75 .sp
1940 ph10 572 In some cases, the escape sequence \eK
1941 ph10 168 .\" HTML <a href="#resetmatchstart">
1942     .\" </a>
1943     (see above)
1944     .\"
1945 ph10 461 can be used instead of a lookbehind assertion to get round the fixed-length
1946 ph10 454 restriction.
1947 ph10 168 .P
1948 nigel 63 The implementation of lookbehind assertions is, for each alternative, to
1949 nigel 93 temporarily move the current position back by the fixed length and then try to
1950 nigel 63 match. If there are insufficient characters before the current position, the
1951 nigel 93 assertion fails.
1952 nigel 75 .P
1953     PCRE does not allow the \eC escape (which matches a single byte in UTF-8 mode)
1954 nigel 63 to appear in lookbehind assertions, because it makes it impossible to calculate
1955 nigel 93 the length of the lookbehind. The \eX and \eR escapes, which can match
1956     different numbers of bytes, are also not permitted.
1957 nigel 75 .P
1958 ph10 454 .\" HTML <a href="#subpatternsassubroutines">
1959     .\" </a>
1960     "Subroutine"
1961     .\"
1962     calls (see below) such as (?2) or (?&X) are permitted in lookbehinds, as long
1963 ph10 461 as the subpattern matches a fixed-length string.
1964 ph10 454 .\" HTML <a href="#recursion">
1965     .\" </a>
1966     Recursion,
1967     .\"
1968     however, is not supported.
1969     .P
1970 nigel 93 Possessive quantifiers can be used in conjunction with lookbehind assertions to
1971 ph10 456 specify efficient matching of fixed-length strings at the end of subject
1972     strings. Consider a simple pattern such as
1973 nigel 75 .sp
1974 nigel 63 abcd$
1975 nigel 75 .sp
1976 nigel 63 when applied to a long string that does not match. Because matching proceeds
1977     from left to right, PCRE will look for each "a" in the subject and then see if
1978     what follows matches the rest of the pattern. If the pattern is specified as
1979 nigel 75 .sp
1980 nigel 63 ^.*abcd$
1981 nigel 75 .sp
1982 nigel 63 the initial .* matches the entire string at first, but when this fails (because
1983     there is no following "a"), it backtracks to match all but the last character,
1984     then all but the last two characters, and so on. Once again the search for "a"
1985     covers the entire string, from right to left, so we are no better off. However,
1986     if the pattern is written as
1987 nigel 75 .sp
1988 nigel 63 ^.*+(?<=abcd)
1989 nigel 75 .sp
1990 nigel 93 there can be no backtracking for the .*+ item; it can match only the entire
1991 nigel 63 string. The subsequent lookbehind assertion does a single test on the last four
1992     characters. If it fails, the match fails immediately. For long strings, this
1993     approach makes a significant difference to the processing time.
1994 nigel 75 .
1995     .
1996     .SS "Using multiple assertions"
1997     .rs
1998     .sp
1999 nigel 63 Several assertions (of any sort) may occur in succession. For example,
2000 nigel 75 .sp
2001     (?<=\ed{3})(?<!999)foo
2002     .sp
2003 nigel 63 matches "foo" preceded by three digits that are not "999". Notice that each of
2004     the assertions is applied independently at the same point in the subject
2005     string. First there is a check that the previous three characters are all
2006     digits, and then there is a check that the same three characters are not "999".
2007 nigel 75 This pattern does \fInot\fP match "foo" preceded by six characters, the first
2008 nigel 63 of which are digits and the last three of which are not "999". For example, it
2009     doesn't match "123abcfoo". A pattern to do that is
2010 nigel 75 .sp
2011     (?<=\ed{3}...)(?<!999)foo
2012     .sp
2013 nigel 63 This time the first assertion looks at the preceding six characters, checking
2014     that the first three are digits, and then the second assertion checks that the
2015     preceding three characters are not "999".
2016 nigel 75 .P
2017 nigel 63 Assertions can be nested in any combination. For example,
2018 nigel 75 .sp
2019 nigel 63 (?<=(?<!foo)bar)baz
2020 nigel 75 .sp
2021 nigel 63 matches an occurrence of "baz" that is preceded by "bar" which in turn is not
2022     preceded by "foo", while
2023 nigel 75 .sp
2024     (?<=\ed{3}(?!999)...)foo
2025     .sp
2026     is another pattern that matches "foo" preceded by three digits and any three
2027 nigel 63 characters that are not "999".
2028 nigel 75 .
2029     .
2030 nigel 91 .\" HTML <a name="conditions"></a>
2032 nigel 63 .rs
2033     .sp
2034     It is possible to cause the matching process to obey a subpattern
2035     conditionally or to choose between two alternative subpatterns, depending on
2036 ph10 461 the result of an assertion, or whether a specific capturing subpattern has
2037 ph10 456 already been matched. The two possible forms of conditional subpattern are:
2038 nigel 75 .sp
2039 nigel 63 (?(condition)yes-pattern)
2040     (?(condition)yes-pattern|no-pattern)
2041 nigel 75 .sp
2042 nigel 63 If the condition is satisfied, the yes-pattern is used; otherwise the
2043     no-pattern (if present) is used. If there are more than two alternatives in the
2044 ph10 557 subpattern, a compile-time error occurs. Each of the two alternatives may
2045 ph10 579 itself contain nested subpatterns of any form, including conditional
2046 ph10 557 subpatterns; the restriction to two alternatives applies only at the level of
2047 ph10 579 the condition. This pattern fragment is an example where the alternatives are
2048 ph10 557 complex:
2049     .sp
2050     (?(1) (A|B|C) | (D | (?(2)E|F) | E) )
2051     .sp
2052 nigel 75 .P
2053 nigel 93 There are four kinds of condition: references to subpatterns, references to
2054     recursion, a pseudo-condition called DEFINE, and assertions.
2055     .
2056     .SS "Checking for a used subpattern by number"
2057     .rs
2058     .sp
2059     If the text between the parentheses consists of a sequence of digits, the
2060 ph10 456 condition is true if a capturing subpattern of that number has previously
2061 ph10 461 matched. If there is more than one capturing subpattern with the same number
2062     (see the earlier
2063 ph10 456 .\"
2064     .\" HTML <a href="#recursion">
2065     .\" </a>
2066     section about duplicate subpattern numbers),
2067     .\"
2068 ph10 572 the condition is true if any of them have matched. An alternative notation is
2069 ph10 456 to precede the digits with a plus or minus sign. In this case, the subpattern
2070     number is relative rather than absolute. The most recently opened parentheses
2071 ph10 572 can be referenced by (?(-1), the next most recent by (?(-2), and so on. Inside
2072 ph10 579 loops it can also make sense to refer to subsequent groups. The next
2073 ph10 572 parentheses to be opened can be referenced as (?(+1), and so on. (The value
2074     zero in any of these forms is not used; it provokes a compile-time error.)
2075 nigel 91 .P
2076     Consider the following pattern, which contains non-significant white space to
2077     make it more readable (assume the PCRE_EXTENDED option) and to divide it into
2078     three parts for ease of discussion:
2079 nigel 75 .sp
2080     ( \e( )? [^()]+ (?(1) \e) )
2081     .sp
2082 nigel 63 The first part matches an optional opening parenthesis, and if that
2083     character is present, sets it as the first captured substring. The second part
2084     matches one or more characters that are not parentheses. The third part is a
2085 ph10 572 conditional subpattern that tests whether or not the first set of parentheses
2086     matched. If they did, that is, if subject started with an opening parenthesis,
2087 nigel 63 the condition is true, and so the yes-pattern is executed and a closing
2088     parenthesis is required. Otherwise, since no-pattern is not present, the
2089     subpattern matches nothing. In other words, this pattern matches a sequence of
2090 nigel 93 non-parentheses, optionally enclosed in parentheses.
2091 ph10 167 .P
2092 ph10 172 If you were embedding this pattern in a larger one, you could use a relative
2093 ph10 167 reference:
2094     .sp
2095     ...other stuff... ( \e( )? [^()]+ (?(-1) \e) ) ...
2096     .sp
2097     This makes the fragment independent of the parentheses in the larger pattern.
2098 nigel 93 .
2099     .SS "Checking for a used subpattern by name"
2100     .rs
2101 nigel 91 .sp
2102 nigel 93 Perl uses the syntax (?(<name>)...) or (?('name')...) to test for a used
2103     subpattern by name. For compatibility with earlier versions of PCRE, which had
2104     this facility before Perl, the syntax (?(name)...) is also recognized. However,
2105     there is a possible ambiguity with this syntax, because subpattern names may
2106     consist entirely of digits. PCRE looks first for a named subpattern; if it
2107     cannot find one and the name consists entirely of digits, PCRE looks for a
2108     subpattern of that number, which must be greater than zero. Using subpattern
2109     names that consist entirely of digits is not recommended.
2110     .P
2111     Rewriting the above example to use a named subpattern gives this:
2112 nigel 91 .sp
2113 nigel 93 (?<OPEN> \e( )? [^()]+ (?(<OPEN>) \e) )
2114     .sp
2115 ph10 461 If the name used in a condition of this kind is a duplicate, the test is
2116     applied to all subpatterns of the same name, and is true if any one of them has
2117 ph10 459 matched.
2118 nigel 93 .
2119     .SS "Checking for pattern recursion"
2120     .rs
2121     .sp
2122 nigel 91 If the condition is the string (R), and there is no subpattern with the name R,
2123 nigel 93 the condition is true if a recursive call to the whole pattern or any
2124     subpattern has been made. If digits or a name preceded by ampersand follow the
2125     letter R, for example:
2126     .sp
2127     (?(R3)...) or (?(R&name)...)
2128     .sp
2129 ph10 456 the condition is true if the most recent recursion is into a subpattern whose
2130 nigel 93 number or name is given. This condition does not check the entire recursion
2131 ph10 461 stack. If the name used in a condition of this kind is a duplicate, the test is
2132     applied to all subpatterns of the same name, and is true if any one of them is
2133     the most recent recursion.
2134 nigel 75 .P
2135 ph10 461 At "top level", all these recursion test conditions are false.
2136 ph10 454 .\" HTML <a href="#recursion">
2137     .\" </a>
2138 ph10 459 The syntax for recursive patterns
2139 ph10 454 .\"
2140 ph10 459 is described below.
2141 nigel 93 .
2142 ph10 572 .\" HTML <a name="subdefine"></a>
2143 nigel 93 .SS "Defining subpatterns for use by reference only"
2144     .rs
2145     .sp
2146     If the condition is the string (DEFINE), and there is no subpattern with the
2147     name DEFINE, the condition is always false. In this case, there may be only one
2148     alternative in the subpattern. It is always skipped if control reaches this
2149     point in the pattern; the idea of DEFINE is that it can be used to define
2150 ph10 716 subroutines that can be referenced from elsewhere. (The use of
2151 ph10 454 .\" HTML <a href="#subpatternsassubroutines">
2152     .\" </a>
2153 ph10 716 subroutines
2154 ph10 454 .\"
2155 ph10 572 is described below.) For example, a pattern to match an IPv4 address such as
2156     "" could be written like this (ignore whitespace and line
2157     breaks):
2158 nigel 93 .sp
2159     (?(DEFINE) (?<byte> 2[0-4]\ed | 25[0-5] | 1\ed\ed | [1-9]?\ed) )
2160     \eb (?&byte) (\e.(?&byte)){3} \eb
2161     .sp
2162     The first part of the pattern is a DEFINE group inside which a another group
2163     named "byte" is defined. This matches an individual component of an IPv4
2164     address (a number less than 256). When matching takes place, this part of the
2165 ph10 456 pattern is skipped because DEFINE acts like a false condition. The rest of the
2166     pattern uses references to the named group to match the four dot-separated
2167     components of an IPv4 address, insisting on a word boundary at each end.
2168 nigel 93 .
2169     .SS "Assertion conditions"
2170     .rs
2171     .sp
2172     If the condition is not in any of the above formats, it must be an assertion.
2173 nigel 63 This may be a positive or negative lookahead or lookbehind assertion. Consider
2174     this pattern, again containing non-significant white space, and with the two
2175     alternatives on the second line:
2176 nigel 75 .sp
2177 nigel 63 (?(?=[^a-z]*[a-z])
2178 nigel 75 \ed{2}-[a-z]{3}-\ed{2} | \ed{2}-\ed{2}-\ed{2} )
2179     .sp
2180 nigel 63 The condition is a positive lookahead assertion that matches an optional
2181     sequence of non-letters followed by a letter. In other words, it tests for the
2182     presence of at least one letter in the subject. If a letter is found, the
2183     subject is matched against the first alternative; otherwise it is matched
2184     against the second. This pattern matches strings in one of the two forms
2185     dd-aaa-dd or dd-dd-dd, where aaa are letters and dd are digits.
2186 nigel 75 .
2187     .
2188     .\" HTML <a name="comments"></a>
2189 nigel 63 .SH COMMENTS
2190     .rs
2191     .sp
2192 ph10 579 There are two ways of including comments in patterns that are processed by
2193 ph10 562 PCRE. In both cases, the start of the comment must not be in a character class,
2194     nor in the middle of any other sequence of related characters such as (?: or a
2195     subpattern name or number. The characters that make up a comment play no part
2196     in the pattern matching.
2197     .P
2198 nigel 75 The sequence (?# marks the start of a comment that continues up to the next
2199 ph10 562 closing parenthesis. Nested parentheses are not permitted. If the PCRE_EXTENDED
2200     option is set, an unescaped # character also introduces a comment, which in
2201     this case continues to immediately after the next newline character or
2202     character sequence in the pattern. Which characters are interpreted as newlines
2203     is controlled by the options passed to \fBpcre_compile()\fP or by a special
2204     sequence at the start of the pattern, as described in the section entitled
2205 ph10 572 .\" HTML <a href="#newlines">
2206 ph10 556 .\" </a>
2207     "Newline conventions"
2208     .\"
2209 ph10 572 above. Note that the end of this type of comment is a literal newline sequence
2210     in the pattern; escape sequences that happen to represent a newline do not
2211     count. For example, consider this pattern when PCRE_EXTENDED is set, and the
2212     default newline convention is in force:
2213 ph10 556 .sp
2214     abc #comment \en still comment
2215     .sp
2216 ph10 579 On encountering the # character, \fBpcre_compile()\fP skips along, looking for
2217 ph10 556 a newline in the pattern. The sequence \en is still literal at this stage, so
2218     it does not terminate the comment. Only an actual character with the code value
2219 ph10 562 0x0a (the default newline) does so.
2220 nigel 75 .
2221     .
2222 nigel 91 .\" HTML <a name="recursion"></a>
2223 nigel 75 .SH "RECURSIVE PATTERNS"
2224 nigel 63 .rs
2225     .sp
2226     Consider the problem of matching a string in parentheses, allowing for
2227     unlimited nested parentheses. Without the use of recursion, the best that can
2228     be done is to use a pattern that matches up to some fixed depth of nesting. It
2229 nigel 93 is not possible to handle an arbitrary nesting depth.
2230     .P
2231     For some time, Perl has provided a facility that allows regular expressions to
2232     recurse (amongst other things). It does this by interpolating Perl code in the
2233     expression at run time, and the code can refer to the expression itself. A Perl
2234     pattern using code interpolation to solve the parentheses problem can be
2235     created like this:
2236 nigel 75 .sp
2237     $re = qr{\e( (?: (?>[^()]+) | (?p{$re}) )* \e)}x;
2238     .sp
2239 nigel 63 The (?p{...}) item interpolates Perl code at run time, and in this case refers
2240 nigel 93 recursively to the pattern in which it appears.
2241 nigel 75 .P
2242 nigel 93 Obviously, PCRE cannot support the interpolation of Perl code. Instead, it
2243     supports special syntax for recursion of the entire pattern, and also for
2244     individual subpattern recursion. After its introduction in PCRE and Python,
2245 ph10 453 this kind of recursion was subsequently introduced into Perl at release 5.10.
2246 nigel 75 .P
2247 nigel 93 A special item that consists of (? followed by a number greater than zero and a
2248 ph10 716 closing parenthesis is a recursive subroutine call of the subpattern of the
2249     given number, provided that it occurs inside that subpattern. (If not, it is a
2250 ph10 454 .\" HTML <a href="#subpatternsassubroutines">
2251     .\" </a>
2252 ph10 716 non-recursive subroutine
2253 ph10 454 .\"
2254 nigel 93 call, which is described in the next section.) The special item (?R) or (?0) is
2255     a recursive call of the entire regular expression.
2256 nigel 87 .P
2257     This PCRE pattern solves the nested parentheses problem (assume the
2258     PCRE_EXTENDED option is set so that white space is ignored):
2259 nigel 75 .sp
2260 ph10 456 \e( ( [^()]++ | (?R) )* \e)
2261 nigel 75 .sp
2262 nigel 63 First it matches an opening parenthesis. Then it matches any number of
2263     substrings which can either be a sequence of non-parentheses, or a recursive
2264 nigel 87 match of the pattern itself (that is, a correctly parenthesized substring).
2265 ph10 461 Finally there is a closing parenthesis. Note the use of a possessive quantifier
2266 ph10 456 to avoid backtracking into sequences of non-parentheses.
2267 nigel 75 .P
2268 nigel 63 If this were part of a larger pattern, you would not want to recurse the entire
2269     pattern, so instead you could use this:
2270 nigel 75 .sp
2271 ph10 456 ( \e( ( [^()]++ | (?1) )* \e) )
2272 nigel 75 .sp
2273 nigel 63 We have put the pattern into parentheses, and caused the recursion to refer to
2274 ph10 172 them instead of the whole pattern.
2275 ph10 166 .P
2276     In a larger pattern, keeping track of parenthesis numbers can be tricky. This
2277 ph10 572 is made easier by the use of relative references. Instead of (?1) in the
2278     pattern above you can write (?-2) to refer to the second most recently opened
2279     parentheses preceding the recursion. In other words, a negative number counts
2280     capturing parentheses leftwards from the point at which it is encountered.
2281 ph10 166 .P
2282     It is also possible to refer to subsequently opened parentheses, by writing
2283     references such as (?+2). However, these cannot be recursive because the
2284     reference is not inside the parentheses that are referenced. They are always
2285 ph10 454 .\" HTML <a href="#subpatternsassubroutines">
2286     .\" </a>
2287 ph10 716 non-recursive subroutine
2288 ph10 454 .\"
2289     calls, as described in the next section.
2290 ph10 166 .P
2291     An alternative approach is to use named parentheses instead. The Perl syntax
2292     for this is (?&name); PCRE's earlier syntax (?P>name) is also supported. We
2293     could rewrite the above example as follows:
2294 nigel 75 .sp
2295 ph10 456 (?<pn> \e( ( [^()]++ | (?&pn) )* \e) )
2296 nigel 75 .sp
2297 nigel 93 If there is more than one subpattern with the same name, the earliest one is
2298 ph10 172 used.
2299 ph10 166 .P
2300     This particular example pattern that we have been looking at contains nested
2301 ph10 456 unlimited repeats, and so the use of a possessive quantifier for matching
2302     strings of non-parentheses is important when applying the pattern to strings
2303     that do not match. For example, when this pattern is applied to
2304 nigel 75 .sp
2305 nigel 63 (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()
2306 nigel 75 .sp
2307 ph10 456 it yields "no match" quickly. However, if a possessive quantifier is not used,
2308 nigel 63 the match runs for a very long time indeed because there are so many different
2309     ways the + and * repeats can carve up the subject, and all have to be tested
2310     before failure can be reported.
2311 nigel 75 .P
2312 ph10 464 At the end of a match, the values of capturing parentheses are those from
2313     the outermost level. If you want to obtain intermediate values, a callout
2314     function can be used (see below and the
2315 nigel 63 .\" HREF
2316 nigel 75 \fBpcrecallout\fP
2317 nigel 63 .\"
2318     documentation). If the pattern above is matched against
2319 nigel 75 .sp
2320 nigel 63 (ab(cd)ef)
2321 nigel 75 .sp
2322 ph10 464 the value for the inner capturing parentheses (numbered 2) is "ef", which is
2323     the last value taken on at the top level. If a capturing subpattern is not
2324 ph10 724 matched at the top level, its final captured value is unset, even if it was
2325     (temporarily) set at a deeper level during the matching process.
2326 nigel 75 .P
2327 ph10 464 If there are more than 15 capturing parentheses in a pattern, PCRE has to
2328     obtain extra memory to store data during a recursion, which it does by using
2329     \fBpcre_malloc\fP, freeing it via \fBpcre_free\fP afterwards. If no memory can
2330     be obtained, the match fails with the PCRE_ERROR_NOMEMORY error.
2331     .P
2332 nigel 63 Do not confuse the (?R) item with the condition (R), which tests for recursion.
2333     Consider this pattern, which matches text in angle brackets, allowing for
2334     arbitrary nesting. Only digits are allowed in nested brackets (that is, when
2335     recursing), whereas any characters are permitted at the outer level.
2336 nigel 75 .sp
2337     < (?: (?(R) \ed++ | [^<>]*+) | (?R)) * >
2338     .sp
2339 nigel 63 In this pattern, (?(R) is the start of a conditional subpattern, with two
2340     different alternatives for the recursive and non-recursive cases. The (?R) item
2341     is the actual recursive call.
2342 nigel 75 .
2343     .
2344 ph10 453 .\" HTML <a name="recursiondifference"></a>
2345 ph10 724 .SS "Differences in recursion processing between PCRE and Perl"
2346 ph10 453 .rs
2347     .sp
2348 ph10 724 Recursion processing in PCRE differs from Perl in two important ways. In PCRE
2349     (like Python, but unlike Perl), a recursive subpattern call is always treated
2350     as an atomic group. That is, once it has matched some of the subject string, it
2351     is never re-entered, even if it contains untried alternatives and there is a
2352     subsequent matching failure. This can be illustrated by the following pattern,
2353     which purports to match a palindromic string that contains an odd number of
2354     characters (for example, "a", "aba", "abcba", "abcdcba"):
2355 ph10 453 .sp
2356     ^(.|(.)(?1)\e2)$
2357     .sp
2358 ph10 461 The idea is that it either matches a single character, or two identical
2359     characters surrounding a sub-palindrome. In Perl, this pattern works; in PCRE
2360 ph10 453 it does not if the pattern is longer than three characters. Consider the
2361     subject string "abcba":
2362     .P
2363 ph10 461 At the top level, the first character is matched, but as it is not at the end
2364 ph10 453 of the string, the first alternative fails; the second alternative is taken
2365     and the recursion kicks in. The recursive call to subpattern 1 successfully
2366     matches the next character ("b"). (Note that the beginning and end of line
2367     tests are not part of the recursion).
2368     .P
2369     Back at the top level, the next character ("c") is compared with what
2370 ph10 461 subpattern 2 matched, which was "a". This fails. Because the recursion is
2371 ph10 453 treated as an atomic group, there are now no backtracking points, and so the
2372     entire match fails. (Perl is able, at this point, to re-enter the recursion and
2373     try the second alternative.) However, if the pattern is written with the
2374     alternatives in the other order, things are different:
2375     .sp
2376     ^((.)(?1)\e2|.)$
2377     .sp
2378 ph10 461 This time, the recursing alternative is tried first, and continues to recurse
2379     until it runs out of characters, at which point the recursion fails. But this
2380     time we do have another alternative to try at the higher level. That is the big
2381 ph10 453 difference: in the previous case the remaining alternative is at a deeper
2382     recursion level, which PCRE cannot use.
2383     .P
2384 ph10 572 To change the pattern so that it matches all palindromic strings, not just
2385     those with an odd number of characters, it is tempting to change the pattern to
2386     this:
2387 ph10 453 .sp
2388     ^((.)(?1)\e2|.?)$
2389     .sp
2390 ph10 461 Again, this works in Perl, but not in PCRE, and for the same reason. When a
2391     deeper recursion has matched a single character, it cannot be entered again in
2392     order to match an empty string. The solution is to separate the two cases, and
2393 ph10 453 write out the odd and even cases as alternatives at the higher level:
2394     .sp
2395     ^(?:((.)(?1)\e2|)|((.)(?3)\e4|.))
2396 ph10 461 .sp
2397     If you want to match typical palindromic phrases, the pattern has to ignore all
2398 ph10 453 non-word characters, which can be done like this:
2399     .sp
2400 ph10 461 ^\eW*+(?:((.)\eW*+(?1)\eW*+\e2|)|((.)\eW*+(?3)\eW*+\e4|\eW*+.\eW*+))\eW*+$
2401 ph10 453 .sp
2402 ph10 461 If run with the PCRE_CASELESS option, this pattern matches phrases such as "A
2403     man, a plan, a canal: Panama!" and it works well in both PCRE and Perl. Note
2404     the use of the possessive quantifier *+ to avoid backtracking into sequences of
2405 ph10 453 non-word characters. Without this, PCRE takes a great deal longer (ten times or
2406     more) to match typical phrases, and Perl takes so long that you think it has
2407     gone into a loop.
2408 ph10 456 .P
2409     \fBWARNING\fP: The palindrome-matching patterns above work only if the subject
2410     string does not start with a palindrome that is shorter than the entire string.
2411     For example, although "abcba" is correctly matched, if the subject is "ababa",
2412     PCRE finds the palindrome "aba" at the start, then fails at top level because
2413     the end of the string does not follow. Once again, it cannot jump back into the
2414     recursion to try other alternatives, so the entire match fails.
2415 ph10 724 .P
2416 ph10 733 The second way in which PCRE and Perl differ in their recursion processing is
2417     in the handling of captured values. In Perl, when a subpattern is called
2418     recursively or as a subpattern (see the next section), it has no access to any
2419     values that were captured outside the recursion, whereas in PCRE these values
2420 ph10 724 can be referenced. Consider this pattern:
2421     .sp
2422     ^(.)(\e1|a(?2))
2423     .sp
2424 ph10 733 In PCRE, this pattern matches "bab". The first capturing parentheses match "b",
2425 ph10 724 then in the second group, when the back reference \e1 fails to match "b", the
2426     second alternative matches "a" and then recurses. In the recursion, \e1 does
2427     now match "b" and so the whole match succeeds. In Perl, the pattern fails to
2428     match because inside the recursive call \e1 cannot access the externally set
2429     value.
2430 ph10 453 .
2431     .
2432 nigel 63 .\" HTML <a name="subpatternsassubroutines"></a>
2434 nigel 63 .rs
2435     .sp
2436 ph10 716 If the syntax for a recursive subpattern call (either by number or by
2437 nigel 63 name) is used outside the parentheses to which it refers, it operates like a
2438 ph10 716 subroutine in a programming language. The called subpattern may be defined
2439 ph10 166 before or after the reference. A numbered reference can be absolute or
2440     relative, as in these examples:
2441 nigel 75 .sp
2442 ph10 166 (...(absolute)...)...(?2)...
2443     (...(relative)...)...(?-1)...
2444 ph10 172 (...(?+1)...(relative)...
2445 ph10 166 .sp
2446     An earlier example pointed out that the pattern
2447     .sp
2448 nigel 75 (sens|respons)e and \e1ibility
2449     .sp
2450 nigel 63 matches "sense and sensibility" and "response and responsibility", but not
2451     "sense and responsibility". If instead the pattern
2452 nigel 75 .sp
2453 nigel 63 (sens|respons)e and (?1)ibility
2454 nigel 75 .sp
2455 nigel 63 is used, it does match "sense and responsibility" as well as the other two
2456 nigel 93 strings. Another example is given in the discussion of DEFINE above.
2457 nigel 87 .P
2458 ph10 716 All subroutine calls, whether recursive or not, are always treated as atomic
2459     groups. That is, once a subroutine has matched some of the subject string, it
2460     is never re-entered, even if it contains untried alternatives and there is a
2461     subsequent matching failure. Any capturing parentheses that are set during the
2462     subroutine call revert to their previous values afterwards.
2463 nigel 93 .P
2464 ph10 716 Processing options such as case-independence are fixed when a subpattern is
2465     defined, so if it is used as a subroutine, such options cannot be changed for
2466     different calls. For example, consider this pattern:
2467 nigel 93 .sp
2468 ph10 166 (abc)(?i:(?-1))
2469 nigel 93 .sp
2470     It matches "abcabc". It does not match "abcABC" because the change of
2471     processing option does not affect the called subpattern.
2472 nigel 75 .
2473     .
2474 ph10 333 .\" HTML <a name="onigurumasubroutines"></a>
2476     .rs
2477     .sp
2478 ph10 345 For compatibility with Oniguruma, the non-Perl syntax \eg followed by a name or
2479     a number enclosed either in angle brackets or single quotes, is an alternative
2480     syntax for referencing a subpattern as a subroutine, possibly recursively. Here
2481 ph10 333 are two of the examples used above, rewritten using this syntax:
2482     .sp
2483     (?<pn> \e( ( (?>[^()]+) | \eg<pn> )* \e) )
2484     (sens|respons)e and \eg'1'ibility
2485     .sp
2486 ph10 345 PCRE supports an extension to Oniguruma: if a number is preceded by a
2487 ph10 333 plus or a minus sign it is taken as a relative reference. For example:
2488     .sp
2489     (abc)(?i:\eg<-1>)
2490     .sp
2491 ph10 345 Note that \eg{...} (Perl syntax) and \eg<...> (Oniguruma syntax) are \fInot\fP
2492 ph10 333 synonymous. The former is a back reference; the latter is a subroutine call.
2493     .
2494     .
2495 nigel 63 .SH CALLOUTS
2496     .rs
2497     .sp
2498     Perl has a feature whereby using the sequence (?{...}) causes arbitrary Perl
2499     code to be obeyed in the middle of matching a regular expression. This makes it
2500     possible, amongst other things, to extract different substrings that match the
2501     same pair of parentheses when there is a repetition.
2502 nigel 75 .P
2503 nigel 63 PCRE provides a similar feature, but of course it cannot obey arbitrary Perl
2504     code. The feature is called "callout". The caller of PCRE provides an external
2505 nigel 75 function by putting its entry point in the global variable \fIpcre_callout\fP.
2506 nigel 63 By default, this variable contains NULL, which disables all calling out.
2507 nigel 75 .P
2508 nigel 63 Within a regular expression, (?C) indicates the points at which the external
2509     function is to be called. If you want to identify different callout points, you
2510     can put a number less than 256 after the letter C. The default value is zero.
2511     For example, this pattern has two callout points:
2512 nigel 75 .sp
2513 ph10 155 (?C1)abc(?C2)def
2514 nigel 75 .sp
2515     If the PCRE_AUTO_CALLOUT flag is passed to \fBpcre_compile()\fP, callouts are
2516     automatically installed before each item in the pattern. They are all numbered
2517     255.
2518     .P
2519     During matching, when PCRE reaches a callout point (and \fIpcre_callout\fP is
2520 nigel 63 set), the external function is called. It is provided with the number of the
2521 nigel 75 callout, the position in the pattern, and, optionally, one item of data
2522     originally supplied by the caller of \fBpcre_exec()\fP. The callout function
2523     may cause matching to proceed, to backtrack, or to fail altogether. A complete
2524     description of the interface to the callout function is given in the
2525 nigel 63 .\" HREF
2526 nigel 75 \fBpcrecallout\fP
2527 nigel 63 .\"
2528     documentation.
2529 nigel 93 .
2530     .
2531 ph10 510 .\" HTML <a name="backtrackcontrol"></a>
2533 ph10 210 .rs
2534     .sp
2535 ph10 211 Perl 5.10 introduced a number of "Special Backtracking Control Verbs", which
2536 ph10 210 are described in the Perl documentation as "experimental and subject to change
2537 ph10 211 or removal in a future version of Perl". It goes on to say: "Their usage in
2538     production code should be noted to avoid problems during upgrades." The same
2539 ph10 210 remarks apply to the PCRE features described in this section.
2540     .P
2541 ph10 341 Since these verbs are specifically related to backtracking, most of them can be
2542     used only when the pattern is to be matched using \fBpcre_exec()\fP, which uses
2543 ph10 345 a backtracking algorithm. With the exception of (*FAIL), which behaves like a
2544 ph10 341 failing negative assertion, they cause an error if encountered by
2545 ph10 210 \fBpcre_dfa_exec()\fP.
2546     .P
2547 ph10 716 If any of these verbs are used in an assertion or in a subpattern that is
2548     called as a subroutine (whether or not recursively), their effect is confined
2549     to that subpattern; it does not extend to the surrounding pattern, with one
2550     exception: a *MARK that is encountered in a positive assertion \fIis\fP passed
2551     back (compare capturing parentheses in assertions). Note that such subpatterns
2552     are processed as anchored at the point where they are tested. Note also that
2553     Perl's treatment of subroutines is different in some cases.
2554 ph10 445 .P
2555 ph10 211 The new verbs make use of what was previously invalid syntax: an opening
2556 ph10 510 parenthesis followed by an asterisk. They are generally of the form
2557 ph10 512 (*VERB) or (*VERB:NAME). Some may take either form, with differing behaviour,
2558 ph10 716 depending on whether or not an argument is present. A name is any sequence of
2559     characters that does not include a closing parenthesis. If the name is empty,
2560     that is, if the closing parenthesis immediately follows the colon, the effect
2561     is as if the colon were not there. Any number of these verbs may occur in a
2562     pattern.
2563 ph10 510 .P
2564 ph10 512 PCRE contains some optimizations that are used to speed up matching by running
2565     some checks at the start of each match attempt. For example, it may know the
2566     minimum length of matching subject, or that a particular character must be
2567     present. When one of these optimizations suppresses the running of a match, any
2568     included backtracking verbs will not, of course, be processed. You can suppress
2569     the start-of-match optimizations by setting the PCRE_NO_START_OPTIMIZE option
2570 ph10 577 when calling \fBpcre_compile()\fP or \fBpcre_exec()\fP, or by starting the
2571     pattern with (*NO_START_OPT).
2572 ph10 210 .
2573 ph10 510 .
2574 ph10 210 .SS "Verbs that act immediately"
2575     .rs
2576     .sp
2577 ph10 512 The following verbs act as soon as they are encountered. They may not be
2578 ph10 510 followed by a name.
2579 ph10 210 .sp
2580     (*ACCEPT)
2581     .sp
2582     This verb causes the match to end successfully, skipping the remainder of the
2583 ph10 716 pattern. However, when it is inside a subpattern that is called as a
2584     subroutine, only that subpattern is ended successfully. Matching then continues
2585     at the outer level. If (*ACCEPT) is inside capturing parentheses, the data so
2586     far is captured. For example:
2587 ph10 210 .sp
2588 ph10 447 A((?:A|B(*ACCEPT)|C)D)
2589 ph10 210 .sp
2590 ph10 461 This matches "AB", "AAD", or "ACD"; when it matches "AB", "B" is captured by
2591 ph10 447 the outer parentheses.
2592 ph10 210 .sp
2593     (*FAIL) or (*F)
2594     .sp
2595 ph10 716 This verb causes a matching failure, forcing backtracking to occur. It is
2596 ph10 210 equivalent to (?!) but easier to read. The Perl documentation notes that it is
2597     probably useful only when combined with (?{}) or (??{}). Those are, of course,
2598     Perl features that are not present in PCRE. The nearest equivalent is the
2599     callout feature, as for example in this pattern:
2600     .sp
2601     a+(?C)(*FAIL)
2602     .sp
2603 ph10 211 A match with the string "aaaa" always fails, but the callout is taken before
2604     each backtrack happens (in this example, 10 times).
2605 ph10 210 .
2606 ph10 510 .
2607     .SS "Recording which path was taken"
2608     .rs
2609     .sp
2610 ph10 512 There is one verb whose main purpose is to track how a match was arrived at,
2611     though it also has a secondary use in conjunction with advancing the match
2612 ph10 510 starting point (see (*SKIP) below).
2613     .sp
2614     (*MARK:NAME) or (*:NAME)
2615     .sp
2616     A name is always required with this verb. There may be as many instances of
2617     (*MARK) as you like in a pattern, and their names do not have to be unique.
2618     .P
2619     When a match succeeds, the name of the last-encountered (*MARK) is passed back
2620     to the caller via the \fIpcre_extra\fP data structure, as described in the
2621     .\" HTML <a href="pcreapi.html#extradata">
2622     .\" </a>
2623     section on \fIpcre_extra\fP
2624     .\"
2625 ph10 512 in the
2626 ph10 510 .\" HREF
2627     \fBpcreapi\fP
2628     .\"
2629     documentation. No data is returned for a partial match. Here is an example of
2630     \fBpcretest\fP output, where the /K modifier requests the retrieval and
2631     outputting of (*MARK) data:
2632     .sp
2633     /X(*MARK:A)Y|X(*MARK:B)Z/K
2634     XY
2635     0: XY
2636     MK: A
2637     XZ
2638     0: XZ
2639     MK: B
2640     .sp
2641 ph10 512 The (*MARK) name is tagged with "MK:" in this output, and in this example it
2642     indicates which of the two alternatives matched. This is a more efficient way
2643 ph10 510 of obtaining this information than putting each alternative in its own
2644     capturing parentheses.
2645     .P
2646 ph10 654 If (*MARK) is encountered in a positive assertion, its name is recorded and
2647     passed back if it is the last-encountered. This does not happen for negative
2648 ph10 716 assertions.
2649 ph10 630 .P
2650 ph10 510 A name may also be returned after a failed match if the final path through the
2651     pattern involves (*MARK). However, unless (*MARK) used in conjunction with
2652     (*COMMIT), this is unlikely to happen for an unanchored pattern because, as the
2653     starting point for matching is advanced, the final check is often with an empty
2654     string, causing a failure before (*MARK) is reached. For example:
2655     .sp
2656     /X(*MARK:A)Y|X(*MARK:B)Z/K
2657     XP
2658     No match
2659     .sp
2660     There are three potential starting points for this match (starting with X,
2661 ph10 512 starting with P, and with an empty string). If the pattern is anchored, the
2662 ph10 510 result is different:
2663     .sp
2664     /^X(*MARK:A)Y|^X(*MARK:B)Z/K
2665     XP
2666     No match, mark = B
2667     .sp
2668 ph10 512 PCRE's start-of-match optimizations can also interfere with this. For example,
2669     if, as a result of a call to \fBpcre_study()\fP, it knows the minimum
2670 ph10 510 subject length for a match, a shorter subject will not be scanned at all.
2671     .P
2672 ph10 512 Note that similar anomalies (though different in detail) exist in Perl, no
2673     doubt for the same reasons. The use of (*MARK) data after a failed match of an
2674 ph10 510 unanchored pattern is not recommended, unless (*COMMIT) is involved.
2675     .
2676     .
2677 ph10 210 .SS "Verbs that act after backtracking"
2678     .rs
2679     .sp
2680 ph10 211 The following verbs do nothing when they are encountered. Matching continues
2681 ph10 510 with what follows, but if there is no subsequent match, causing a backtrack to
2682     the verb, a failure is forced. That is, backtracking cannot pass to the left of
2683     the verb. However, when one of these verbs appears inside an atomic group, its
2684     effect is confined to that group, because once the group has been matched,
2685     there is never any backtracking into it. In this situation, backtracking can
2686 ph10 512 "jump back" to the left of the entire atomic group. (Remember also, as stated
2687 ph10 510 above, that this localization also applies in subroutine calls and assertions.)
2688     .P
2689     These verbs differ in exactly what kind of failure occurs when backtracking
2690     reaches them.
2691 ph10 210 .sp
2692     (*COMMIT)
2693     .sp
2694 ph10 510 This verb, which may not be followed by a name, causes the whole match to fail
2695     outright if the rest of the pattern does not match. Even if the pattern is
2696     unanchored, no further attempts to find a match by advancing the starting point
2697     take place. Once (*COMMIT) has been passed, \fBpcre_exec()\fP is committed to
2698     finding a match at the current starting point, or not at all. For example:
2699 ph10 210 .sp
2700     a+(*COMMIT)b
2701     .sp
2702 ph10 211 This matches "xxaab" but not "aacaab". It can be thought of as a kind of
2703 ph10 512 dynamic anchor, or "I've started, so I must finish." The name of the most
2704     recently passed (*MARK) in the path is passed back when (*COMMIT) forces a
2705 ph10 510 match failure.
2706     .P
2707 ph10 512 Note that (*COMMIT) at the start of a pattern is not the same as an anchor,
2708     unless PCRE's start-of-match optimizations are turned off, as shown in this
2709 ph10 510 \fBpcretest\fP example:
2710 ph10 210 .sp
2711 ph10 510 /(*COMMIT)abc/
2712     xyzabc
2713     0: abc
2714     xyzabc\eY
2715     No match
2716 ph10 210 .sp
2717 ph10 512 PCRE knows that any match must start with "a", so the optimization skips along
2718 ph10 510 the subject to "a" before running the first match attempt, which succeeds. When
2719     the optimization is disabled by the \eY escape in the second subject, the match
2720 ph10 512 starts at "x" and so the (*COMMIT) causes it to fail without trying any other
2721 ph10 510 starting points.
2722 ph10 210 .sp
2723 ph10 510 (*PRUNE) or (*PRUNE:NAME)
2724     .sp
2725 ph10 512 This verb causes the match to fail at the current starting position in the
2726 ph10 510 subject if the rest of the pattern does not match. If the pattern is
2727     unanchored, the normal "bumpalong" advance to the next starting character then
2728     happens. Backtracking can occur as usual to the left of (*PRUNE), before it is
2729     reached, or when matching to the right of (*PRUNE), but if there is no match to
2730     the right, backtracking cannot cross (*PRUNE). In simple cases, the use of
2731     (*PRUNE) is just an alternative to an atomic group or possessive quantifier,
2732     but there are some uses of (*PRUNE) that cannot be expressed in any other way.
2733     The behaviour of (*PRUNE:NAME) is the same as (*MARK:NAME)(*PRUNE) when the
2734     match fails completely; the name is passed back if this is the final attempt.
2735     (*PRUNE:NAME) does not pass back a name if the match succeeds. In an anchored
2736     pattern (*PRUNE) has the same effect as (*COMMIT).
2737     .sp
2738 ph10 210 (*SKIP)
2739     .sp
2740 ph10 510 This verb, when given without a name, is like (*PRUNE), except that if the
2741     pattern is unanchored, the "bumpalong" advance is not to the next character,
2742     but to the position in the subject where (*SKIP) was encountered. (*SKIP)
2743     signifies that whatever text was matched leading up to it cannot be part of a
2744     successful match. Consider:
2745 ph10 210 .sp
2746     a+(*SKIP)b
2747     .sp
2748 ph10 211 If the subject is "aaaac...", after the first match attempt fails (starting at
2749 ph10 210 the first character in the string), the starting point skips on to start the
2750 ph10 211 next attempt at "c". Note that a possessive quantifer does not have the same
2751 ph10 456 effect as this example; although it would suppress backtracking during the
2752 ph10 210 first match attempt, the second attempt would start at the second character
2753     instead of skipping on to "c".
2754     .sp
2755 ph10 510 (*SKIP:NAME)
2756 ph10 211 .sp
2757 ph10 512 When (*SKIP) has an associated name, its behaviour is modified. If the
2758     following pattern fails to match, the previous path through the pattern is
2759     searched for the most recent (*MARK) that has the same name. If one is found,
2760     the "bumpalong" advance is to the subject position that corresponds to that
2761     (*MARK) instead of to where (*SKIP) was encountered. If no (*MARK) with a
2762 ph10 716 matching name is found, normal "bumpalong" of one character happens (that is,
2763     the (*SKIP) is ignored).
2764 ph10 510 .sp
2765     (*THEN) or (*THEN:NAME)
2766     .sp
2767 ph10 716 This verb causes a skip to the next innermost alternative if the rest of the
2768     pattern does not match. That is, it cancels pending backtracking, but only
2769     within the current alternative. Its name comes from the observation that it can
2770     be used for a pattern-based if-then-else block:
2771 ph10 210 .sp
2772     ( COND1 (*THEN) FOO | COND2 (*THEN) BAR | COND3 (*THEN) BAZ ) ...
2773     .sp
2774 ph10 211 If the COND1 pattern matches, FOO is tried (and possibly further items after
2775 ph10 716 the end of the group if FOO succeeds); on failure, the matcher skips to the
2776 ph10 510 second alternative and tries COND2, without backtracking into COND1. The
2777     behaviour of (*THEN:NAME) is exactly the same as (*MARK:NAME)(*THEN) if the
2778 ph10 716 overall match fails. If (*THEN) is not inside an alternation, it acts like
2779     (*PRUNE).
2780 ph10 551 .P
2781 ph10 716 Note that a subpattern that does not contain a | character is just a part of
2782     the enclosing alternative; it is not a nested alternation with only one
2783     alternative. The effect of (*THEN) extends beyond such a subpattern to the
2784     enclosing alternative. Consider this pattern, where A, B, etc. are complex
2785     pattern fragments that do not contain any | characters at this level:
2786     .sp
2787     A (B(*THEN)C) | D
2788     .sp
2789 ph10 733 If A and B are matched, but there is a failure in C, matching does not
2790 ph10 716 backtrack into A; instead it moves to the next alternative, that is, D.
2791     However, if the subpattern containing (*THEN) is given an alternative, it
2792     behaves differently:
2793     .sp
2794     A (B(*THEN)C | (*FAIL)) | D
2795     .sp
2796     The effect of (*THEN) is now confined to the inner subpattern. After a failure
2797 ph10 733 in C, matching moves to (*FAIL), which causes the whole subpattern to fail
2798     because there are no more alternatives to try. In this case, matching does now
2799 ph10 716 backtrack into A.
2800 ph10 551 .P
2801 ph10 733 Note also that a conditional subpattern is not considered as having two
2802     alternatives, because only one is ever used. In other words, the | character in
2803 ph10 716 a conditional subpattern has a different meaning. Ignoring white space,
2804     consider:
2805 ph10 551 .sp
2806 ph10 716 ^.*? (?(?=a) a | b(*THEN)c )
2807     .sp
2808 ph10 733 If the subject is "ba", this pattern does not match. Because .*? is ungreedy,
2809     it initially matches zero characters. The condition (?=a) then fails, the
2810 ph10 716 character "b" is matched, but "c" is not. At this point, matching does not
2811     backtrack to .*? as might perhaps be expected from the presence of the |
2812     character. The conditional subpattern is part of the single alternative that
2813 ph10 733 comprises the whole pattern, and so the match fails. (If there was a backtrack
2814 ph10 716 into .*?, allowing it to match "b", the match would succeed.)
2815     .P
2816     The verbs just described provide four different "strengths" of control when
2817     subsequent matching fails. (*THEN) is the weakest, carrying on the match at the
2818     next alternative. (*PRUNE) comes next, failing the match at the current
2819     starting position, but allowing an advance to the next character (for an
2820     unanchored pattern). (*SKIP) is similar, except that the advance may be more
2821     than one character. (*COMMIT) is the strongest, causing the entire match to
2822     fail.
2823     .P
2824     If more than one such verb is present in a pattern, the "strongest" one wins.
2825     For example, consider this pattern, where A, B, etc. are complex pattern
2826     fragments:
2827     .sp
2828 ph10 551 (A(*COMMIT)B(*THEN)C|D)
2829     .sp
2830     Once A has matched, PCRE is committed to this match, at the current starting
2831     position. If subsequently B matches, but C does not, the normal (*THEN) action
2832 ph10 716 of trying the next alternative (that is, D) does not happen because (*COMMIT)
2833 ph10 551 overrides.
2834 ph10 210 .
2835 ph10 551 .
2836 nigel 93 .SH "SEE ALSO"
2837     .rs
2838     .sp
2839 ph10 461 \fBpcreapi\fP(3), \fBpcrecallout\fP(3), \fBpcrematching\fP(3),
2840 ph10 456 \fBpcresyntax\fP(3), \fBpcre\fP(3).
2841 ph10 99 .
2842     .
2843     .SH AUTHOR
2844     .rs
2845     .sp
2846     .nf
2847     Philip Hazel
2848     University Computing Service
2849     Cambridge CB2 3QH, England.
2850     .fi
2851     .
2852     .
2853     .SH REVISION
2854     .rs
2855     .sp
2856     .nf
2857 ph10 737 Last updated: 19 October 2011
2858 ph10 607 Copyright (c) 1997-2011 University of Cambridge.
2859 ph10 99 .fi


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

ViewVC Help
Powered by ViewVC 1.1.12